the quarterly journal of economics · the quarterly journal of economics vol.cxix february 2004...
TRANSCRIPT
THE
QUARTERLY JOURNALOF ECONOMICS
Vol CXIX February 2004 Issue 1
THE MODERN HISTORY OF EXCHANGE RATEARRANGEMENTS A REINTERPRETATION
CARMEN M REINHART AND KENNETH S ROGOFF
We develop a novel system of reclassifying historical exchange rate regimesOne key difference between our study and previous classications is that weemploy monthly data on market-determined parallel exchange rates going back to1946 for 153 countries Our approach differs from the IMF ofcial classication(which we show to be only a little better than random) it also differs radicallyfrom all previous attempts at historical reclassication Our classication pointsto a rethinking of economic performance under alternative exchange rate regimesIndeed the breakup of Bretton Woods had less impact on exchange rate regimesthan is popularly believed
I INTRODUCTION
This paper rewrites the history of post-World War II ex-change rate arrangements based on an extensive new monthlydata set spanning across 153 countries for 1946 ndash2001 Our ap-proach differs not only from countriesrsquo ofcially declared classi-cations (which we show to be only a little better than random) italso differs radically from the small number of previous attemptsat historical reclassication1
The authors wish to thank Alberto Alesina Arminio Fraga Amartya LahiriVincent Reinhart Andrew Rose Miguel Savastano participants at Harvard Uni-versityrsquos Canada-US Economic and Monetary Integration Conference Interna-tional Monetary Fund-World Bank Joint Seminar National Bureau of EconomicResearch Summer Institute New York University Princeton University andthree anonymous referees for useful comments and suggestions and KenichiroKashiwase Daouda Sembene and Ioannis Tokatlidis for excellent research as-sistance Data and background material to this paper are available athttpwwwpuafumdedufacultypapersreinhartreinharthtm
1 The ofcial classication is given in the IMFrsquos Annual Report on ExchangeRate Arrangements and Exchange Restrictions which until recently asked mem-ber states to self-declare their arrangement as belonging to one of four categories
copy 2004 by the President and Fellows of Harvard College and the Massachusetts Institute ofTechnologyThe Quarterly Journal of Economics February 2004
1
As a rst innovation we incorporate data on parallel anddual exchange rate markets which have been enormously impor-tant not only in developing countries but in virtually all theEuropean countries up until the late 1950s and sometimes wellbeyond We argue that any classication algorithm that fails todistinguish between unied rate systems (with one ofcial ex-change rate and no signicant ldquoblackrdquo or parallel market) and allothers is fundamentally awed Indeed in the vast majority ofmultiple exchange rate or dual systems the oating dual orparallel rate is not only a far better barometer of monetary policythan is the ofcial exchange rate it is often the most economicallymeaningful rate2 Very frequentlymdashroughly half the time forofcial pegsmdashwe nd that dualparallel rates have been used as aform of ldquoback doorrdquo oating albeit one usually accompanied byexchange controls The second novelty in our approach is that wedevelop extensive chronologies of the history of exchange ar-rangements and related factors such as exchange controls andcurrency reforms Together with a battery of descriptive statis-tics this allows us to draw a nuanced distinction between whatcountries declare as their ofcial de jure regime and their actualde facto exchange rate practices To capture the wide range ofarrangements our approach allows for fourteen categories ofexchange rate regimes ranging from no separate legal tender ora strict peg to a dysfunctional ldquofreely fallingrdquo or ldquohyperoatrdquo
Some highlights from our reclassication of exchange ratearrangements are as follows
First dual or multiple rates and parallel markets haveprevailed far more frequently than is commonly acknowledged In1950 45 percent of the countries in our sample had dual ormultiple rates many more had thriving parallel markets Amongthe industrialized economies dual or multiple rates were the
Previous studies have either extended the four-way ofcial classication into amore informative taxonomy (see Ghosh et al [1997]) or relied largely on statis-tical methods to regroup country practices (see Levy-Yeyati and Sturzenegger[2002]) The Fund recognizing the limitations of its former strategy revised andupgraded the ofcial approach toward classifying exchange rate arrangements in1997 and again in 1999 Notably all these prior approaches to exchange rateregime classication whether or not they accept the countryrsquos declared regimehave been based solely on ofcial exchange rates
2 When we refer to multiple exchange rates in this context we are focusingon the cases where one or more of the rates is market-determined This is verydifferent from the cases where the multiple ofcial rates are all xed and simplyact as a differential tax on a variety of transactions Dual markets are typicallylegal whereas parallel markets may or may not be legal
2 QUARTERLY JOURNAL OF ECONOMICS
norm in the 1940s and the 1950s and in some cases these lasteduntil much later Our data lend strong support to the viewstressed by Bordo [1993] that Bretton Woods encompassed twovery different kinds of exchange rate arrangements in the pre-and postconvertibility periods and that the period of meaningfulexchange rate stability was quite short-lived In the developingworld such practices remained commonplace through the 1980sand 1990s and into the present
We show that market-determined dualparallel markets areimportant barometers of underlying monetary policy This maybe obvious in cases such as modern-day Myanmar where theparallel market premium at the beginning of 2003 exceeded 700percent As we show however the phenomenon is much moregeneral with the parallel market premium often serving as areliable guide to the direction of future ofcial exchange ratechanges Whereas dualparallel markets have been marginal oversome episodes they have been economically important in othersand there are many instances where only a few transactions takeplace at the ofcial rate To assess the importance of secondary(legal or illegal) parallel markets we collected data that allow usto estimate export misinvoicing practices in many cases goingback to 1948 These estimates show that leakages from the ofcialmarket were signicant in many of the episodes when there weredual or parallel markets
Second when one uses market-determined rates in place ofofcial rates the history of exchange rate policy begins to lookvery different For example it becomes obvious that de factooating was common during the early years of the Bretton Woodsera of ldquopeggedrdquo exchange rates Conversely many ldquooatsrdquo of thepost-1980s turn out to be (de facto) pegs crawling pegs or verynarrow bands Of countries listed in the ofcial IMF classicationas managed oating 53 percent turned out to have de facto pegscrawls or narrow bands to some anchor currency
Third next to pegs (which account for 33 percent of theobservations during 1970 ndash2001 (according to our new ldquoNaturalrdquoclassication) the most popular exchange rate regime over mod-ern history has been the crawling peg which accounted for over26 percent of the observations During 1990 to 2001 this was themost common type of arrangement in emerging Asia and WesternHemisphere (excluding Canada and the United States) makingup for about 36 and 42 percent of the observations respectively
Fourth our taxonomy introduces a new category freely fall-
3EXCHANGE RATE ARRANGEMENTS
ing or the cases where the twelve-month ination rate is equal toor exceeds 40 percent per annum3 It turns out to be a crowdedcategory indeed with about 12 12 percent of the observations inour sample occurring in the freely falling category As a resultldquofreely fallingrdquo is about three times as common as ldquofreely oat-ingrdquo which accounts for only 4 12 percent of the total observa-tions (In the ofcial classication freely oating accounts forover 30 percent of observations over the past decade) Our newfreely falling classication makes up 22 and 37 percent of theobservations respectively in Africa and Western Hemisphere(excluding Canada and the United States) during 1970 ndash2001 Inthe 1990s freely falling accounted for 41 percent of the observa-tions for the transition economies Given the distortions associ-ated with very high ination any xed versus exible exchangerate regime comparisons that do not break out the freely fallingepisodes are meaningless as we shall conrm
There are many important reasons to seek a better approachto classifying exchange rate regimes Certainly one is the recog-nition that contemporary thinking on the costs and benets ofalternative exchange rate arrangements has been profoundly in-uenced by the large number of studies on the empirical differ-ences in growth trade ination business cycles and commodityprice behavior Most have been based on the ofcial classicationsand all on ofcial exchange rates In light of the new evidence wecollect we conjecture that the inuential results in Baxter andStockman [1989]mdashthat there are no signicant differences inbusiness cycles across exchange arrangementsmdashmay be due tothe fact that the ofcial historical groupings of exchange ratearrangements are misleading
The paper proceeds as follows In the next section we presentevidence to establish the incidence and importance of dual ormultiple exchange rate practices In Section III we sketch ourmethodology for reclassifying exchange rate arrangements Sec-tion IV addresses some of the possible critiques to our approachcompares our results with the ldquoofcial historyrdquo and providesexamples of how our reclassication may reshape evidence on thelinks between exchange rate arrangements and various facets ofeconomic activity The nal section reiterates some of the main
3 We also include in the freely falling category the rst six months followingan exchange rate crisis (see the Appendix for details) but only for those caseswhere the crisis marked a transition from a peg or quasi-peg to a managed orindependent oat
4 QUARTERLY JOURNAL OF ECONOMICS
ndings while background material to this paper provides thedetailed country chronologies that underpin our analysis
II THE INCIDENCE AND IMPORTANCE OF DUAL AND MULTIPLE
EXCHANGE RATE ARRANGEMENTS
In this section we document the incidence of dual or parallelmarkets (legal or otherwise) and multiple exchange rate practicesduring post-World War II We then present evidence that themarket-determined exchange rate is a better indicator of theunderlying monetary policy than the ofcial exchange rate Fi-nally to provide a sense of the quantitative importance for eco-nomic activity of the dual or parallel market we present esti-mates of ldquoleakagesrdquo from the ofcial market Specically weprovide quantitative measures of export misinvoicing practices
We primarily use monthly data on ofcial and market-deter-mined exchange rates for the period 1946 ndash2001 In some in-stances the data for the market-determined rate is only availablefor a shorter period and the background material provides theparticulars on a country-by-country basis The pre-1999 market-determined exchange rate data come from various issues of PickrsquosCurrency Yearbook Pickrsquos Black Market Yearbooks and WorldCurrency Reports and the ofcial rate comes from the samesources and as well as the IMF The quotes are end-of-monthexchange rates and are not subject to revisions For the recentperiod (1999 ndash2001) the monthly data on market-determined ex-change rates come from the original country sources (ie thecentral banks) for those countries where there are active parallelmarkets for which data are available4 Since our coverage spansmore than 50 years it encompasses numerous cases of monetaryreforms involving changes in the units of account so the datawere spliced accordingly to ensure continuity
IIA On the Popularity of Dual and Multiple Exchange RatePractices
Figure I illustrates de facto and de jure nonunied exchangerate regimes The gure shows the incidence of exchange ratearrangements over 1950 ndash2001 with and without stripping out
4 These countries include Afghanistan Angola Argentina Belarus BelizeBolivia Burundi Congo (DCR) Dominican Republic Egypt Ghana Iran LibyaMacedonia Mauritania Myanmar Nigeria Pakistan Rwanda Tajikistan Turk-menistan Ukraine Uzbekistan Yemen Yugoslavia and Zimbabwe
5EXCHANGE RATE ARRANGEMENTS
cases of dual markets or multiple exchange rates The IMF clas-sication has been simplied into what it was back in the days ofBretton Woodsmdashnamely Pegs and Other5 The dark portions ofthe bars represent cases with unied exchange rates and thelightly shaded portion of each bar separates out the dual multi-ple or parallel cases In 1950 more than half (53 percent) of allarrangements involved two or more exchange rates Indeed theheyday of multiple exchange rate practices and active parallelmarkets was 1946 ndash1958 before the restoration of convertibilityin Europe Note also that according to the ofcial IMF classi-cation pegs reigned supreme in the early 1970s accounting forover 90 percent of all exchange rate arrangements In fact overhalf of these ldquopegsrdquo masked parallel markets that as we shallshow often exhibited quite different behavior
5 For a history of the evolution of the IMFrsquos classication strategy see theworking paper version of this paper Reinhart and Rogoff [2002]
FIGURE IThe Incidence of Dual or Multiple Exchange Rate Arrangements 1950ndash2001
Simplied IMF ClassicationSources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
Exchange rate arrangements classied as ldquoOtherrdquo include the IMFrsquos categoriesof limited exibility managed oating and independently oating
6 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
As a rst innovation we incorporate data on parallel anddual exchange rate markets which have been enormously impor-tant not only in developing countries but in virtually all theEuropean countries up until the late 1950s and sometimes wellbeyond We argue that any classication algorithm that fails todistinguish between unied rate systems (with one ofcial ex-change rate and no signicant ldquoblackrdquo or parallel market) and allothers is fundamentally awed Indeed in the vast majority ofmultiple exchange rate or dual systems the oating dual orparallel rate is not only a far better barometer of monetary policythan is the ofcial exchange rate it is often the most economicallymeaningful rate2 Very frequentlymdashroughly half the time forofcial pegsmdashwe nd that dualparallel rates have been used as aform of ldquoback doorrdquo oating albeit one usually accompanied byexchange controls The second novelty in our approach is that wedevelop extensive chronologies of the history of exchange ar-rangements and related factors such as exchange controls andcurrency reforms Together with a battery of descriptive statis-tics this allows us to draw a nuanced distinction between whatcountries declare as their ofcial de jure regime and their actualde facto exchange rate practices To capture the wide range ofarrangements our approach allows for fourteen categories ofexchange rate regimes ranging from no separate legal tender ora strict peg to a dysfunctional ldquofreely fallingrdquo or ldquohyperoatrdquo
Some highlights from our reclassication of exchange ratearrangements are as follows
First dual or multiple rates and parallel markets haveprevailed far more frequently than is commonly acknowledged In1950 45 percent of the countries in our sample had dual ormultiple rates many more had thriving parallel markets Amongthe industrialized economies dual or multiple rates were the
Previous studies have either extended the four-way ofcial classication into amore informative taxonomy (see Ghosh et al [1997]) or relied largely on statis-tical methods to regroup country practices (see Levy-Yeyati and Sturzenegger[2002]) The Fund recognizing the limitations of its former strategy revised andupgraded the ofcial approach toward classifying exchange rate arrangements in1997 and again in 1999 Notably all these prior approaches to exchange rateregime classication whether or not they accept the countryrsquos declared regimehave been based solely on ofcial exchange rates
2 When we refer to multiple exchange rates in this context we are focusingon the cases where one or more of the rates is market-determined This is verydifferent from the cases where the multiple ofcial rates are all xed and simplyact as a differential tax on a variety of transactions Dual markets are typicallylegal whereas parallel markets may or may not be legal
2 QUARTERLY JOURNAL OF ECONOMICS
norm in the 1940s and the 1950s and in some cases these lasteduntil much later Our data lend strong support to the viewstressed by Bordo [1993] that Bretton Woods encompassed twovery different kinds of exchange rate arrangements in the pre-and postconvertibility periods and that the period of meaningfulexchange rate stability was quite short-lived In the developingworld such practices remained commonplace through the 1980sand 1990s and into the present
We show that market-determined dualparallel markets areimportant barometers of underlying monetary policy This maybe obvious in cases such as modern-day Myanmar where theparallel market premium at the beginning of 2003 exceeded 700percent As we show however the phenomenon is much moregeneral with the parallel market premium often serving as areliable guide to the direction of future ofcial exchange ratechanges Whereas dualparallel markets have been marginal oversome episodes they have been economically important in othersand there are many instances where only a few transactions takeplace at the ofcial rate To assess the importance of secondary(legal or illegal) parallel markets we collected data that allow usto estimate export misinvoicing practices in many cases goingback to 1948 These estimates show that leakages from the ofcialmarket were signicant in many of the episodes when there weredual or parallel markets
Second when one uses market-determined rates in place ofofcial rates the history of exchange rate policy begins to lookvery different For example it becomes obvious that de factooating was common during the early years of the Bretton Woodsera of ldquopeggedrdquo exchange rates Conversely many ldquooatsrdquo of thepost-1980s turn out to be (de facto) pegs crawling pegs or verynarrow bands Of countries listed in the ofcial IMF classicationas managed oating 53 percent turned out to have de facto pegscrawls or narrow bands to some anchor currency
Third next to pegs (which account for 33 percent of theobservations during 1970 ndash2001 (according to our new ldquoNaturalrdquoclassication) the most popular exchange rate regime over mod-ern history has been the crawling peg which accounted for over26 percent of the observations During 1990 to 2001 this was themost common type of arrangement in emerging Asia and WesternHemisphere (excluding Canada and the United States) makingup for about 36 and 42 percent of the observations respectively
Fourth our taxonomy introduces a new category freely fall-
3EXCHANGE RATE ARRANGEMENTS
ing or the cases where the twelve-month ination rate is equal toor exceeds 40 percent per annum3 It turns out to be a crowdedcategory indeed with about 12 12 percent of the observations inour sample occurring in the freely falling category As a resultldquofreely fallingrdquo is about three times as common as ldquofreely oat-ingrdquo which accounts for only 4 12 percent of the total observa-tions (In the ofcial classication freely oating accounts forover 30 percent of observations over the past decade) Our newfreely falling classication makes up 22 and 37 percent of theobservations respectively in Africa and Western Hemisphere(excluding Canada and the United States) during 1970 ndash2001 Inthe 1990s freely falling accounted for 41 percent of the observa-tions for the transition economies Given the distortions associ-ated with very high ination any xed versus exible exchangerate regime comparisons that do not break out the freely fallingepisodes are meaningless as we shall conrm
There are many important reasons to seek a better approachto classifying exchange rate regimes Certainly one is the recog-nition that contemporary thinking on the costs and benets ofalternative exchange rate arrangements has been profoundly in-uenced by the large number of studies on the empirical differ-ences in growth trade ination business cycles and commodityprice behavior Most have been based on the ofcial classicationsand all on ofcial exchange rates In light of the new evidence wecollect we conjecture that the inuential results in Baxter andStockman [1989]mdashthat there are no signicant differences inbusiness cycles across exchange arrangementsmdashmay be due tothe fact that the ofcial historical groupings of exchange ratearrangements are misleading
The paper proceeds as follows In the next section we presentevidence to establish the incidence and importance of dual ormultiple exchange rate practices In Section III we sketch ourmethodology for reclassifying exchange rate arrangements Sec-tion IV addresses some of the possible critiques to our approachcompares our results with the ldquoofcial historyrdquo and providesexamples of how our reclassication may reshape evidence on thelinks between exchange rate arrangements and various facets ofeconomic activity The nal section reiterates some of the main
3 We also include in the freely falling category the rst six months followingan exchange rate crisis (see the Appendix for details) but only for those caseswhere the crisis marked a transition from a peg or quasi-peg to a managed orindependent oat
4 QUARTERLY JOURNAL OF ECONOMICS
ndings while background material to this paper provides thedetailed country chronologies that underpin our analysis
II THE INCIDENCE AND IMPORTANCE OF DUAL AND MULTIPLE
EXCHANGE RATE ARRANGEMENTS
In this section we document the incidence of dual or parallelmarkets (legal or otherwise) and multiple exchange rate practicesduring post-World War II We then present evidence that themarket-determined exchange rate is a better indicator of theunderlying monetary policy than the ofcial exchange rate Fi-nally to provide a sense of the quantitative importance for eco-nomic activity of the dual or parallel market we present esti-mates of ldquoleakagesrdquo from the ofcial market Specically weprovide quantitative measures of export misinvoicing practices
We primarily use monthly data on ofcial and market-deter-mined exchange rates for the period 1946 ndash2001 In some in-stances the data for the market-determined rate is only availablefor a shorter period and the background material provides theparticulars on a country-by-country basis The pre-1999 market-determined exchange rate data come from various issues of PickrsquosCurrency Yearbook Pickrsquos Black Market Yearbooks and WorldCurrency Reports and the ofcial rate comes from the samesources and as well as the IMF The quotes are end-of-monthexchange rates and are not subject to revisions For the recentperiod (1999 ndash2001) the monthly data on market-determined ex-change rates come from the original country sources (ie thecentral banks) for those countries where there are active parallelmarkets for which data are available4 Since our coverage spansmore than 50 years it encompasses numerous cases of monetaryreforms involving changes in the units of account so the datawere spliced accordingly to ensure continuity
IIA On the Popularity of Dual and Multiple Exchange RatePractices
Figure I illustrates de facto and de jure nonunied exchangerate regimes The gure shows the incidence of exchange ratearrangements over 1950 ndash2001 with and without stripping out
4 These countries include Afghanistan Angola Argentina Belarus BelizeBolivia Burundi Congo (DCR) Dominican Republic Egypt Ghana Iran LibyaMacedonia Mauritania Myanmar Nigeria Pakistan Rwanda Tajikistan Turk-menistan Ukraine Uzbekistan Yemen Yugoslavia and Zimbabwe
5EXCHANGE RATE ARRANGEMENTS
cases of dual markets or multiple exchange rates The IMF clas-sication has been simplied into what it was back in the days ofBretton Woodsmdashnamely Pegs and Other5 The dark portions ofthe bars represent cases with unied exchange rates and thelightly shaded portion of each bar separates out the dual multi-ple or parallel cases In 1950 more than half (53 percent) of allarrangements involved two or more exchange rates Indeed theheyday of multiple exchange rate practices and active parallelmarkets was 1946 ndash1958 before the restoration of convertibilityin Europe Note also that according to the ofcial IMF classi-cation pegs reigned supreme in the early 1970s accounting forover 90 percent of all exchange rate arrangements In fact overhalf of these ldquopegsrdquo masked parallel markets that as we shallshow often exhibited quite different behavior
5 For a history of the evolution of the IMFrsquos classication strategy see theworking paper version of this paper Reinhart and Rogoff [2002]
FIGURE IThe Incidence of Dual or Multiple Exchange Rate Arrangements 1950ndash2001
Simplied IMF ClassicationSources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
Exchange rate arrangements classied as ldquoOtherrdquo include the IMFrsquos categoriesof limited exibility managed oating and independently oating
6 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
norm in the 1940s and the 1950s and in some cases these lasteduntil much later Our data lend strong support to the viewstressed by Bordo [1993] that Bretton Woods encompassed twovery different kinds of exchange rate arrangements in the pre-and postconvertibility periods and that the period of meaningfulexchange rate stability was quite short-lived In the developingworld such practices remained commonplace through the 1980sand 1990s and into the present
We show that market-determined dualparallel markets areimportant barometers of underlying monetary policy This maybe obvious in cases such as modern-day Myanmar where theparallel market premium at the beginning of 2003 exceeded 700percent As we show however the phenomenon is much moregeneral with the parallel market premium often serving as areliable guide to the direction of future ofcial exchange ratechanges Whereas dualparallel markets have been marginal oversome episodes they have been economically important in othersand there are many instances where only a few transactions takeplace at the ofcial rate To assess the importance of secondary(legal or illegal) parallel markets we collected data that allow usto estimate export misinvoicing practices in many cases goingback to 1948 These estimates show that leakages from the ofcialmarket were signicant in many of the episodes when there weredual or parallel markets
Second when one uses market-determined rates in place ofofcial rates the history of exchange rate policy begins to lookvery different For example it becomes obvious that de factooating was common during the early years of the Bretton Woodsera of ldquopeggedrdquo exchange rates Conversely many ldquooatsrdquo of thepost-1980s turn out to be (de facto) pegs crawling pegs or verynarrow bands Of countries listed in the ofcial IMF classicationas managed oating 53 percent turned out to have de facto pegscrawls or narrow bands to some anchor currency
Third next to pegs (which account for 33 percent of theobservations during 1970 ndash2001 (according to our new ldquoNaturalrdquoclassication) the most popular exchange rate regime over mod-ern history has been the crawling peg which accounted for over26 percent of the observations During 1990 to 2001 this was themost common type of arrangement in emerging Asia and WesternHemisphere (excluding Canada and the United States) makingup for about 36 and 42 percent of the observations respectively
Fourth our taxonomy introduces a new category freely fall-
3EXCHANGE RATE ARRANGEMENTS
ing or the cases where the twelve-month ination rate is equal toor exceeds 40 percent per annum3 It turns out to be a crowdedcategory indeed with about 12 12 percent of the observations inour sample occurring in the freely falling category As a resultldquofreely fallingrdquo is about three times as common as ldquofreely oat-ingrdquo which accounts for only 4 12 percent of the total observa-tions (In the ofcial classication freely oating accounts forover 30 percent of observations over the past decade) Our newfreely falling classication makes up 22 and 37 percent of theobservations respectively in Africa and Western Hemisphere(excluding Canada and the United States) during 1970 ndash2001 Inthe 1990s freely falling accounted for 41 percent of the observa-tions for the transition economies Given the distortions associ-ated with very high ination any xed versus exible exchangerate regime comparisons that do not break out the freely fallingepisodes are meaningless as we shall conrm
There are many important reasons to seek a better approachto classifying exchange rate regimes Certainly one is the recog-nition that contemporary thinking on the costs and benets ofalternative exchange rate arrangements has been profoundly in-uenced by the large number of studies on the empirical differ-ences in growth trade ination business cycles and commodityprice behavior Most have been based on the ofcial classicationsand all on ofcial exchange rates In light of the new evidence wecollect we conjecture that the inuential results in Baxter andStockman [1989]mdashthat there are no signicant differences inbusiness cycles across exchange arrangementsmdashmay be due tothe fact that the ofcial historical groupings of exchange ratearrangements are misleading
The paper proceeds as follows In the next section we presentevidence to establish the incidence and importance of dual ormultiple exchange rate practices In Section III we sketch ourmethodology for reclassifying exchange rate arrangements Sec-tion IV addresses some of the possible critiques to our approachcompares our results with the ldquoofcial historyrdquo and providesexamples of how our reclassication may reshape evidence on thelinks between exchange rate arrangements and various facets ofeconomic activity The nal section reiterates some of the main
3 We also include in the freely falling category the rst six months followingan exchange rate crisis (see the Appendix for details) but only for those caseswhere the crisis marked a transition from a peg or quasi-peg to a managed orindependent oat
4 QUARTERLY JOURNAL OF ECONOMICS
ndings while background material to this paper provides thedetailed country chronologies that underpin our analysis
II THE INCIDENCE AND IMPORTANCE OF DUAL AND MULTIPLE
EXCHANGE RATE ARRANGEMENTS
In this section we document the incidence of dual or parallelmarkets (legal or otherwise) and multiple exchange rate practicesduring post-World War II We then present evidence that themarket-determined exchange rate is a better indicator of theunderlying monetary policy than the ofcial exchange rate Fi-nally to provide a sense of the quantitative importance for eco-nomic activity of the dual or parallel market we present esti-mates of ldquoleakagesrdquo from the ofcial market Specically weprovide quantitative measures of export misinvoicing practices
We primarily use monthly data on ofcial and market-deter-mined exchange rates for the period 1946 ndash2001 In some in-stances the data for the market-determined rate is only availablefor a shorter period and the background material provides theparticulars on a country-by-country basis The pre-1999 market-determined exchange rate data come from various issues of PickrsquosCurrency Yearbook Pickrsquos Black Market Yearbooks and WorldCurrency Reports and the ofcial rate comes from the samesources and as well as the IMF The quotes are end-of-monthexchange rates and are not subject to revisions For the recentperiod (1999 ndash2001) the monthly data on market-determined ex-change rates come from the original country sources (ie thecentral banks) for those countries where there are active parallelmarkets for which data are available4 Since our coverage spansmore than 50 years it encompasses numerous cases of monetaryreforms involving changes in the units of account so the datawere spliced accordingly to ensure continuity
IIA On the Popularity of Dual and Multiple Exchange RatePractices
Figure I illustrates de facto and de jure nonunied exchangerate regimes The gure shows the incidence of exchange ratearrangements over 1950 ndash2001 with and without stripping out
4 These countries include Afghanistan Angola Argentina Belarus BelizeBolivia Burundi Congo (DCR) Dominican Republic Egypt Ghana Iran LibyaMacedonia Mauritania Myanmar Nigeria Pakistan Rwanda Tajikistan Turk-menistan Ukraine Uzbekistan Yemen Yugoslavia and Zimbabwe
5EXCHANGE RATE ARRANGEMENTS
cases of dual markets or multiple exchange rates The IMF clas-sication has been simplied into what it was back in the days ofBretton Woodsmdashnamely Pegs and Other5 The dark portions ofthe bars represent cases with unied exchange rates and thelightly shaded portion of each bar separates out the dual multi-ple or parallel cases In 1950 more than half (53 percent) of allarrangements involved two or more exchange rates Indeed theheyday of multiple exchange rate practices and active parallelmarkets was 1946 ndash1958 before the restoration of convertibilityin Europe Note also that according to the ofcial IMF classi-cation pegs reigned supreme in the early 1970s accounting forover 90 percent of all exchange rate arrangements In fact overhalf of these ldquopegsrdquo masked parallel markets that as we shallshow often exhibited quite different behavior
5 For a history of the evolution of the IMFrsquos classication strategy see theworking paper version of this paper Reinhart and Rogoff [2002]
FIGURE IThe Incidence of Dual or Multiple Exchange Rate Arrangements 1950ndash2001
Simplied IMF ClassicationSources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
Exchange rate arrangements classied as ldquoOtherrdquo include the IMFrsquos categoriesof limited exibility managed oating and independently oating
6 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
ing or the cases where the twelve-month ination rate is equal toor exceeds 40 percent per annum3 It turns out to be a crowdedcategory indeed with about 12 12 percent of the observations inour sample occurring in the freely falling category As a resultldquofreely fallingrdquo is about three times as common as ldquofreely oat-ingrdquo which accounts for only 4 12 percent of the total observa-tions (In the ofcial classication freely oating accounts forover 30 percent of observations over the past decade) Our newfreely falling classication makes up 22 and 37 percent of theobservations respectively in Africa and Western Hemisphere(excluding Canada and the United States) during 1970 ndash2001 Inthe 1990s freely falling accounted for 41 percent of the observa-tions for the transition economies Given the distortions associ-ated with very high ination any xed versus exible exchangerate regime comparisons that do not break out the freely fallingepisodes are meaningless as we shall conrm
There are many important reasons to seek a better approachto classifying exchange rate regimes Certainly one is the recog-nition that contemporary thinking on the costs and benets ofalternative exchange rate arrangements has been profoundly in-uenced by the large number of studies on the empirical differ-ences in growth trade ination business cycles and commodityprice behavior Most have been based on the ofcial classicationsand all on ofcial exchange rates In light of the new evidence wecollect we conjecture that the inuential results in Baxter andStockman [1989]mdashthat there are no signicant differences inbusiness cycles across exchange arrangementsmdashmay be due tothe fact that the ofcial historical groupings of exchange ratearrangements are misleading
The paper proceeds as follows In the next section we presentevidence to establish the incidence and importance of dual ormultiple exchange rate practices In Section III we sketch ourmethodology for reclassifying exchange rate arrangements Sec-tion IV addresses some of the possible critiques to our approachcompares our results with the ldquoofcial historyrdquo and providesexamples of how our reclassication may reshape evidence on thelinks between exchange rate arrangements and various facets ofeconomic activity The nal section reiterates some of the main
3 We also include in the freely falling category the rst six months followingan exchange rate crisis (see the Appendix for details) but only for those caseswhere the crisis marked a transition from a peg or quasi-peg to a managed orindependent oat
4 QUARTERLY JOURNAL OF ECONOMICS
ndings while background material to this paper provides thedetailed country chronologies that underpin our analysis
II THE INCIDENCE AND IMPORTANCE OF DUAL AND MULTIPLE
EXCHANGE RATE ARRANGEMENTS
In this section we document the incidence of dual or parallelmarkets (legal or otherwise) and multiple exchange rate practicesduring post-World War II We then present evidence that themarket-determined exchange rate is a better indicator of theunderlying monetary policy than the ofcial exchange rate Fi-nally to provide a sense of the quantitative importance for eco-nomic activity of the dual or parallel market we present esti-mates of ldquoleakagesrdquo from the ofcial market Specically weprovide quantitative measures of export misinvoicing practices
We primarily use monthly data on ofcial and market-deter-mined exchange rates for the period 1946 ndash2001 In some in-stances the data for the market-determined rate is only availablefor a shorter period and the background material provides theparticulars on a country-by-country basis The pre-1999 market-determined exchange rate data come from various issues of PickrsquosCurrency Yearbook Pickrsquos Black Market Yearbooks and WorldCurrency Reports and the ofcial rate comes from the samesources and as well as the IMF The quotes are end-of-monthexchange rates and are not subject to revisions For the recentperiod (1999 ndash2001) the monthly data on market-determined ex-change rates come from the original country sources (ie thecentral banks) for those countries where there are active parallelmarkets for which data are available4 Since our coverage spansmore than 50 years it encompasses numerous cases of monetaryreforms involving changes in the units of account so the datawere spliced accordingly to ensure continuity
IIA On the Popularity of Dual and Multiple Exchange RatePractices
Figure I illustrates de facto and de jure nonunied exchangerate regimes The gure shows the incidence of exchange ratearrangements over 1950 ndash2001 with and without stripping out
4 These countries include Afghanistan Angola Argentina Belarus BelizeBolivia Burundi Congo (DCR) Dominican Republic Egypt Ghana Iran LibyaMacedonia Mauritania Myanmar Nigeria Pakistan Rwanda Tajikistan Turk-menistan Ukraine Uzbekistan Yemen Yugoslavia and Zimbabwe
5EXCHANGE RATE ARRANGEMENTS
cases of dual markets or multiple exchange rates The IMF clas-sication has been simplied into what it was back in the days ofBretton Woodsmdashnamely Pegs and Other5 The dark portions ofthe bars represent cases with unied exchange rates and thelightly shaded portion of each bar separates out the dual multi-ple or parallel cases In 1950 more than half (53 percent) of allarrangements involved two or more exchange rates Indeed theheyday of multiple exchange rate practices and active parallelmarkets was 1946 ndash1958 before the restoration of convertibilityin Europe Note also that according to the ofcial IMF classi-cation pegs reigned supreme in the early 1970s accounting forover 90 percent of all exchange rate arrangements In fact overhalf of these ldquopegsrdquo masked parallel markets that as we shallshow often exhibited quite different behavior
5 For a history of the evolution of the IMFrsquos classication strategy see theworking paper version of this paper Reinhart and Rogoff [2002]
FIGURE IThe Incidence of Dual or Multiple Exchange Rate Arrangements 1950ndash2001
Simplied IMF ClassicationSources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
Exchange rate arrangements classied as ldquoOtherrdquo include the IMFrsquos categoriesof limited exibility managed oating and independently oating
6 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
ndings while background material to this paper provides thedetailed country chronologies that underpin our analysis
II THE INCIDENCE AND IMPORTANCE OF DUAL AND MULTIPLE
EXCHANGE RATE ARRANGEMENTS
In this section we document the incidence of dual or parallelmarkets (legal or otherwise) and multiple exchange rate practicesduring post-World War II We then present evidence that themarket-determined exchange rate is a better indicator of theunderlying monetary policy than the ofcial exchange rate Fi-nally to provide a sense of the quantitative importance for eco-nomic activity of the dual or parallel market we present esti-mates of ldquoleakagesrdquo from the ofcial market Specically weprovide quantitative measures of export misinvoicing practices
We primarily use monthly data on ofcial and market-deter-mined exchange rates for the period 1946 ndash2001 In some in-stances the data for the market-determined rate is only availablefor a shorter period and the background material provides theparticulars on a country-by-country basis The pre-1999 market-determined exchange rate data come from various issues of PickrsquosCurrency Yearbook Pickrsquos Black Market Yearbooks and WorldCurrency Reports and the ofcial rate comes from the samesources and as well as the IMF The quotes are end-of-monthexchange rates and are not subject to revisions For the recentperiod (1999 ndash2001) the monthly data on market-determined ex-change rates come from the original country sources (ie thecentral banks) for those countries where there are active parallelmarkets for which data are available4 Since our coverage spansmore than 50 years it encompasses numerous cases of monetaryreforms involving changes in the units of account so the datawere spliced accordingly to ensure continuity
IIA On the Popularity of Dual and Multiple Exchange RatePractices
Figure I illustrates de facto and de jure nonunied exchangerate regimes The gure shows the incidence of exchange ratearrangements over 1950 ndash2001 with and without stripping out
4 These countries include Afghanistan Angola Argentina Belarus BelizeBolivia Burundi Congo (DCR) Dominican Republic Egypt Ghana Iran LibyaMacedonia Mauritania Myanmar Nigeria Pakistan Rwanda Tajikistan Turk-menistan Ukraine Uzbekistan Yemen Yugoslavia and Zimbabwe
5EXCHANGE RATE ARRANGEMENTS
cases of dual markets or multiple exchange rates The IMF clas-sication has been simplied into what it was back in the days ofBretton Woodsmdashnamely Pegs and Other5 The dark portions ofthe bars represent cases with unied exchange rates and thelightly shaded portion of each bar separates out the dual multi-ple or parallel cases In 1950 more than half (53 percent) of allarrangements involved two or more exchange rates Indeed theheyday of multiple exchange rate practices and active parallelmarkets was 1946 ndash1958 before the restoration of convertibilityin Europe Note also that according to the ofcial IMF classi-cation pegs reigned supreme in the early 1970s accounting forover 90 percent of all exchange rate arrangements In fact overhalf of these ldquopegsrdquo masked parallel markets that as we shallshow often exhibited quite different behavior
5 For a history of the evolution of the IMFrsquos classication strategy see theworking paper version of this paper Reinhart and Rogoff [2002]
FIGURE IThe Incidence of Dual or Multiple Exchange Rate Arrangements 1950ndash2001
Simplied IMF ClassicationSources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
Exchange rate arrangements classied as ldquoOtherrdquo include the IMFrsquos categoriesof limited exibility managed oating and independently oating
6 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
cases of dual markets or multiple exchange rates The IMF clas-sication has been simplied into what it was back in the days ofBretton Woodsmdashnamely Pegs and Other5 The dark portions ofthe bars represent cases with unied exchange rates and thelightly shaded portion of each bar separates out the dual multi-ple or parallel cases In 1950 more than half (53 percent) of allarrangements involved two or more exchange rates Indeed theheyday of multiple exchange rate practices and active parallelmarkets was 1946 ndash1958 before the restoration of convertibilityin Europe Note also that according to the ofcial IMF classi-cation pegs reigned supreme in the early 1970s accounting forover 90 percent of all exchange rate arrangements In fact overhalf of these ldquopegsrdquo masked parallel markets that as we shallshow often exhibited quite different behavior
5 For a history of the evolution of the IMFrsquos classication strategy see theworking paper version of this paper Reinhart and Rogoff [2002]
FIGURE IThe Incidence of Dual or Multiple Exchange Rate Arrangements 1950ndash2001
Simplied IMF ClassicationSources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
Exchange rate arrangements classied as ldquoOtherrdquo include the IMFrsquos categoriesof limited exibility managed oating and independently oating
6 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
IIB The Market-Determined Exchange Rate as an Indicator ofMonetary Policy
While the quality of data on market-determined rates islikely to vary across countries and time we nevertheless believethese data to be generally far better barometers of the underlyingmonetary policy than are ofcial exchange rates For instance ifthe laxity in monetary policy is not consistent with maintaining axed ofcial exchange rate one would expect that the market-determined rate starts depreciating ahead of the inevitable de-valuation of the ofcial rate When the ofcial realignment oc-cursmdashit is simply a validation of what had previously transpiredin the free market Indeed this is the pattern shown in the threepanels of Figure II for the cases of Bolivia Indonesia and Iranmdashmany more such cases are displayed in the gures that accom-pany the 153 country chronologies6 This pattern also emergesoften in the developed European economies and Japan in theyears following World War II
To illustrate more rigorously that the market-based ex-change rate is a better indicator of the monetary policy stancethan the ofcial rate we performed two exercises for each coun-try First we examined whether the market-determined ex-change rate systematically predicts realignments in the ofcialrate as suggested in Figure II To do so we regressed a currencycrash dummy on the parallel market premium lagged one to sixmonths for each of the developing countries in our sample7 If themarket exchange rate consistently anticipates devaluations of theofcial rate its coefcient should be positive and statisticallysignicant If in turn the ofcial exchange rate does not validatethe market rate then the coefcient on the lagged market ex-change rate will be negative or simply not signicant Table Isummarizes the results of the country-by-country time seriesprobit regressions In the overwhelming number of cases (97percent) the coefcient on the market-determined exchange rateis positive In about 81 percent of the cases the sign on thecoefcient was positive and statistically signicant Indeed for
6 See ldquoPart I The Country Chronologies and Chartbook Background Mate-rial to A Modern History of Exchange Rate Arrangements A Reinterpretationrdquo athttpwwwpuafumdedufacultypapersreinhartreinharthtm
7 Two denitions of currency crashes are used A severe currency crashrefers to a 25 percent or higher monthly depreciation which is at least 10 percenthigher than the previous monthrsquos depreciation The ldquomilderrdquo version represents a125 percent monthly depreciation which is at least 10 percent above the preced-ing monthrsquos depreciation see details in the Appendix
7EXCHANGE RATE ARRANGEMENTS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IIOfcial Exchange Rates Typically Validate the Changes in the Market RatesSources Pick and Sedillot [1971] International Currency Analysis World Cur-
rency Yearbook various issues
8 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
Western Hemisphere as a region the coefcient on the parallelpremium was signicant for all the countries in our sampleThese ndings are in line with those of Bahmani-OskooeeMiteza and Nasir [2002] who use panel annual data for 1973ndash1990 for 49 countries and employ a completely different approachTheir panel cointegration tests indicate that the ofcial rate willsystematically adjust to the market rate in the long run
Second we calculated pairwise correlations between ination(measured as the twelve-month change in the consumer priceindex) and the twelve-month percent change in the ofcial andmarket exchange rates six months earlier If the market rate isa better pulse of monetary policy it should be (a priori) moreclosely correlated with ination As shown in Table II we ndthat for the majority of cases (about three-quarters of the coun-tries) the changes in market-determined exchange rates havehigher correlations with ination than do changes in the ofcialrate8 An interesting exception to this pattern of higher correla-
8 Note that due to data limitations we use ofcial prices rather than blackmarket or ldquostreetrdquo prices to measure ination here Otherwise the dominance ofthe market-determined rates in this exercise would presumably be even morepronounced
TABLE IIS THE PARALLEL MARKET RATE A GOOD PREDICTOR OF CRASHES IN THE OFFICIAL
EXCHANGE RATESUMMARY OF THE PROBIT COUNTRY-BY-COUNTRY ESTIMATION
Regression DOt 5 a 1 b DPt2 i 1 ut ldquoMildrdquo crash
Percent of countries for whichb 0 971b 0 and signicanta 814b 0 29b 0 and signicanta 14
Sources Pickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and theauthorsrsquo calculations
DOt is a dummy variable that takes on the value of 1 when there is a realignment in the ofcial exchange ratealong the lines described below and 0 otherwise a and b are the intercept and slope coefcients respectively (ournull hypothesis is b 0) DPt2i is the twelve-monthchange in the parallel exchange rate lagged one to six months(the lags were allowed to vary country by country as there was no prior reason to restrict dynamics to be the samefor all countries) and ut is a random disturbance Two denitions of currency crashes are used in the spirit ofFrankel and Rose [1996] A ldquosevererdquo currency crash refers to a 25 percent or higher monthly depreciation whichis at least 10 percent higher than the previousmonthrsquos depreciationThe ldquomildrdquo version represents a 125 percentmonthly depreciation which is at least 10 percent above the preceding monthrsquos depreciation Since both deni-tions of crash yield similar results we report here only those for the more inclusive denition The regressionsample varies by country and is determined by data availability
a At the 10 percent condence level or higher
9EXCHANGE RATE ARRANGEMENTS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
tions between the market-determined exchange rate changes andination is for the industrial countries in the ldquoConvertible Bret-ton Woodsrdquo period (1959 ndash1973) an issue that merits furtherstudy
IIC How Important Are Parallel Markets
There are cases where the parallel (or secondary) exchangerate applies only to a few limited transactions An example is theldquoswitch poundrdquo in the United Kingdom during September 1950through April 19679 However it is not unusual for dual orparallel markets (legal or otherwise) to account for the lionrsquosshare of transactions with the ofcial rate being little more thansymbolic As Kiguel Lizondo and OrsquoConnell [1997] note theofcial rate typically diminishes in importance when the gapbetween the ofcial and market-determined rate widens
To provide a sense of the comparative relevance of the dual orparallel market we proceed along two complementary dimen-sions First we include a qualitative description in the country-specic chronologies (see background material) of what transac-tions take place in the ofcial market versus the secondary mar-ket Second we develop a quantitative measure of the potentialsize of the leakages into dual or parallel exchange markets10
9 For example while the United Kingdom ofcially had dual rates throughApril 1967 the secondary rate was so trivial (both in terms of the premium andthe volume of transactions it applied to) that it is classied as a peg in ourclassication scheme (see background material) In the next section we describehow our classication algorithm deals with these cases
10 For instance according to Claessens [1997] export underinvoicing hit ahistoric high in Mexico during 1982mdashthe crisis year in which the dual market was
TABLE IIINFLATION OFFICIAL AND MARKET-DETERMINED EXCHANGE RATES
COUNTRY-BY-COUNTRY PAIRWISE CORRELATIONS
Percent of countries for which the correlations of
The market-determined exchange rate and ination are higher than thecorrelations of the ofcial rate and ination 737
The market-determined exchange rate and ination are lower than thecorrelations of the ofcial rate and ination 263
Sources International Monetary Fund International Financial Statistics Pickrsquos Currency YearbookWorld Currency Report Pickrsquos Black Market Yearbook and the authorsrsquo calculations
The correlations reported are those of the twelve-monthpercent change in the consumer price index withthe twelve-month percent change in the relevant bilateral exchange rate lagged six months
10 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
Following Ghei Kiguel and OrsquoConnell [1997] we classify epi-sodes where there are dualparallel markets into three tiers ac-cording to the level (in percent) of the parallel market premiumlow (below 10 percent) moderate (10 percent or above but below50) and high (50 percent and above) For the episodes of dualparallel markets we provide information about which categoryeach episode falls in (by calculating the average premium for theduration of the episode) In addition to the information containedin the premium we constructed an extensive database on exportmisinvoicing or the difference between what a country reports asits exports and what other countries report as imports from thatcountry adjusted for shipping costs Historically there are tightlinks between capital ight export underinvoicing and the par-allel market premium11 As with the parallel market premiumwe divide the export misinvoicing estimates into three categories(as a percent of the value of total exports) low (less than 10percent of exports) moderate (10 to 15 percent of exports) andhigh (above 15 percent) For Europe Japan and the UnitedStates misinvoicing calculations start in 1948 while for theremaining countries these start in 1970 In the extensive back-ground material to this paper we show for each episode which ofthe three categories is applicable Finally we construct a score (1for Low 2 for Moderate and 3 for High) for both of these proxiesfor leakages The combined score on the estimated size of theleakages (these range from 2 to 6) is also reported12
Table III which shows the evolution of export misinvoicing(as a percent of the value of total exports) and the parallel marketpremium (in percent) across regions and through time provides ageneral avor of the size of potential leakages from the ofcialmarket According to our estimates of misinvoicing (top panel)the regional patterns show the largest leakages for the Caribbeanand non-CFA Sub-Saharan Africa 1970 ndash2001 with averages inthe 30 to 50 percent range The lowest estimates of misinvoicing(8 to 11 percent) are for Western Europe North America and the
introduced Similar statements can be made about other crisis episodes thatinvolved the introduction of exchange controls and the segmentation of markets
11 See Kiguel Lizondo and OrsquoConnell [1997] and the references containedtherein
12 See ldquoPart II Parallel Markets and Dual and Multiple Exchange RatePractices Background Material to A Modern History of Exchange Rate Arrange-ments A Reinterpretationrdquo at httpwwwpuafumdedufacultypapersrein-hartreinharthtm
11EXCHANGE RATE ARRANGEMENTS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
TABLE IIILEAKAGES EXPORT MISINVOICING AND THE PARALLEL MARKET PREMIUM
ABSOLUTE VALUE OF EXPORT MISINVOICING
(AS A PERCENT OF THE VALUE OF EXPORTS)
Descriptive statistics Mean absolute value (by decade)
Min Max St dev 48ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash01 70ndash01
World 70 398 84 128 109 99 247 221 260 244North Africa 25 599 103 72 83 161 109CFA 126 483 84 285 217 215 238Rest of Africa 163 2019 335 234 234 536 341Middle East and
Turkey 91 454 96 307 167 174 215Developing Asia
and Pacic 95 791 169 314 149 241 235Industrialized Asia 37 182 33 112 142 139 146 120 103 122Caribbean 97 1360 332 308 489 600 470Central and South
America 120 496 82 261 360 304 308Central and
Eastern Europe 25 500 183 466 154 74 221Western Europe 24 169 30 141 104 100 116 76 77 89North America 06 226 59 46 94 38 160 114 48 104
Monthly average parallel market premium(excluding freely falling episodes in percent)
Descriptive statistics Average (by decade)
Min Max St dev 46ndash49 50ndash59 60ndash69 70ndash79 80ndash89 90ndash98 46ndash98
World 116 2059 354 1378 567 381 313 578 526 541North Africa 212 1648 414 99 357 307 1086 620 536CFA 264 127 27 00 12 18 09Rest of Africa 17 3225 739 319 69 337 1137 1127 1077 710Middle East and
Turkey 51 4931 996 546 810 260 214 1465 1932 886Developing Asia
and Pacic 237 6601 950 1435 609 1689 447 431 121 729Industrialized Asia 269 8159 1076 3244 430 120 36 13 15 361Caribbean 2238 3000 428 296 302 568 536 423Central and South
America 30 7161 785 491 1330 164 186 748 84 510Western Europe 256 3475 486 1655 170 12 20 17 12 169North America 243 497 33 72 05 00 11 14 16 13
Sources International Monetary Fund Direction of Trade Statistics International Financial StatisticsPickrsquos Currency Yearbook World Currency Report Pickrsquos Black Market Yearbook and authorsrsquo calculations
To calculate export misinvoicing let XWi 5 imports from country i as reported by the rest of the world(CIF basis) Xi 5 exports to the world as reported by country i Z 5 imports CIF basisimports COB basisthen export misinvoicing 5 (XWiZ) 2 Xi The averages reported are absolute values as a percent of the valueof total exports The parallel premium is dened as 100 3 [(P 2 O)O)] where P and O are the parallel andofcial rates respectively The averages for the parallel premium are calculated for all the countries in oursample in each region as such it includes countries where rates are unied and the premium is zero or nil
12 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
CFA Franc Zone It is also noteworthy that although low by thestandards of other regions the export misinvoicing average in1970 ndash2001 for Western Europe is half of what it was in 1948 ndash1949 Yet these regional averages may understate the importanceof misinvoicing in some countries For example the maximumvalue for 1948 ndash2001 for Western Europe (169 percent) does notreect the fact that for Spain misinvoicing as a percent of thevalue of exports amounted to 36 percent in 1950 a comparablevalue to what we see in some of the developing regions
As to the regional average parallel market premium shownin the bottom panel of Table III all regions fall squarely in theModerate-to-High range (with the exception of North AmericaWestern Europe and CFA Africa) In the case of developing Asiathe averages are signicantly raised by Myanmar and Laos It isworth noting the averages for Europe and industrialized Asia inthe 1940s are comparable and even higher than those recordedfor many developing countries highlighting the importance ofacknowledging and accounting for dual markets during thisperiod
To sum in this section we have presented evidence that leadsus to conclude that parallel markets were both important asindicators of monetary policy and as representative of the pricesunderlying an important share of economic transactions It istherefore quite reasonable to draw heavily on the dual or parallelmarket data in classifying exchange rate regimes the task towhich we now turn
III THE ldquoNATURALrdquo CLASSIFICATION CODE A GUIDE
We would describe our classication scheme as a ldquoNaturalrdquosystem that relies on a broad variety of descriptive statistics andchronologies to group episodes into a much ner grid of regimesrather than the three or four buckets of other recent classicationstrategies13 The two most important new pieces of informationwe bring to bear are our extensive data on market-determineddual or parallel exchange rates and detailed country chronologiesThe data its sources and country coverage are described alongwith the chronologies that map the history of exchange ratearrangements for each country in the detailed background mate-
13 In biology a natural taxonomic scheme relies on the characteristics of aspecies to group them
13EXCHANGE RATE ARRANGEMENTS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
rial to this paper To verify and classify regimes we also rely ona variety of descriptive statistics based on exchange rate andination data from 1946 onwards the Appendix describes these
IIIA The Algorithm
Figure III is a schematic summarizing our Natural Classi-cation algorithm First we use the chronologies to sort out forseparate treatment countries with either ofcial dual or multiplerates or active parallel (black) markets14 Second if there is nodual or parallel market we check to see if there is an ofcialpreannounced arrangement such as a peg or band If there is weexamine summary statistics to verify the announced regime go-ing forward from the date of the announcement If the regime isveried (ie exchange rate behavior accords with the prean-nounced policy) it is then classied accordingly as a peg crawlingpeg etc If the announcement fails verication (by far the mostcommon outcome) we then seek a de facto statistical classica-tion using the algorithm described below and discussed ingreater detail in the Appendix
Third if there is no preannounced path for the exchangerate or if the announced regime cannot be veried by the dataand the twelve-month rate of ination is below 40 percent weclassify the regime by evaluating exchange rate behavior Asregards which exchange rate is used we consider a variety ofpotential anchor currencies including the US dollar deutschemark euro French franc UK pound yen Australian dollarItalian lira SDR South African rand and the Indian rupee Areading of the country chronologies makes plain that the relevantanchor currency varies not only across countries but sometimeswithin a country over time (For example many former Britishcolonies switched from pegging to the UK pound to pegging to theUS dollar)
Our volatility measure is based on a ve-year moving window(see the Appendix for details) so that the monthly exchange ratebehavior may be viewed as part of a larger continuous regime15
14 See background material posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
15 If the classication is based on exchange rate behavior in a particularyear it is more likely that one-time events (such as a one-time devaluation andrepeg) or an economic or political shock leads to labeling the year as a change inregime when in effect there is no change For example Levy-Yeyati and Stur-zenegger [2002] who classify regimes one year at a time (with no memory)classied all CFA zone countries as having an intermediate regime in 1994 when
14 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
these countries had a one-time devaluation in January of that year Our algorithmclassies them as having pegs throughout The ve-year window also makes itless likely that we classify as a peg an exchange rate that did not move simplybecause it was a tranquil year with no economic or political shocks It is far lessprobable that there are no shocks over a ve-year span
FIGURE IIIA Natural Exchange Rate Classication Algorithm
15EXCHANGE RATE ARRANGEMENTS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
We also examined the graphical evidence as a check on theclassication In practice the main reason for doing so is toseparate pegs from crawling pegs or bands and to sort the latterinto crawling and noncrawling bands
Fourth as we have already stressed a straightforward butfundamental departure from all previous classication schemes isthat we create a new separate category for countries whosetwelve-month rate of ination is above 40 percent These casesare labeled ldquofreely fallingrdquo16 If the country is in a hyperination(according to the classic Cagan [1956] denition of 50 percent ormore monthly ination) we categorize the exchange rate regimeas a ldquohyperoatrdquo a subspecies of freely falling In Figure IVbilateral exchange rates versus the US dollar are plotted for twocountries that have been classied by the IMF (and all previousclassication efforts) as oating over much of the postwar pe-riodmdashCanada and Argentina17 To us lumping the Canadianoat with that of Argentina during its hyperination seems at aminimum misleading As Figure IV illustrates oating regimeslook rather different from freely falling regimesmdashwitness theorders of magnitude difference in the scales between Canada (topof page) and Argentina (bottom) This difference is highlighted inthe middle panel which plots the Canadian dollar-US dollarexchange rate against Argentinarsquos scale from this perspective itlooks like a xed rate The exchange rate histories of other coun-tries that experienced chronic high ination boutsmdash even if thesedid not reach the hyperination stagemdashlook more similar to Ar-gentina in Figure IV than to Canada18 In our view regimesassociated with an utter lack of monetary control and the atten-dant very high ination should not be automatically lumpedunder the same exchange rate arrangement as low ination oat-ing regimes On these grounds freely falling needs to be treatedas a separate category much in the same way that Highly In-debted Poorest Countries (HIPC) are treated as a separate ldquotyperdquoof debtor
16 In the exceptional cases (usually the beginning of an ination stabiliza-tion plan) where despite ination over 40 percent the market rate neverthelessfollows a conrmed preannounced band or crawl the preannounced regime takesprecedence
17 For Argentina this of course refers to the period before the ConvertibilityPlan is introduced in April 1991 and for Canada the post-1962 period
18 Two-panel gures such as that shown for Chile (Figure V) for eachcountry in the sample are found in the background material alongside the coun-try-specic chronologies
16 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IVThe Essential Distinction between Freely Floating and Falling
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
17EXCHANGE RATE ARRANGEMENTS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
In step 5 we take up those residual regimes that were notclassied in steps 1 through 4 These regimes become candidatesfor ldquomanagedrdquo or ldquofreelyrdquo oating19 To distinguish between thetwo we perform some simple tests (see the Appendix) that look atthe likelihood the exchange rate will move within a narrow rangeas well as the mean absolute value of exchange rate changesWhen there are dual or parallel markets and the parallel marketpremium is consistently 10 percent or higher we apply steps 1through 5 to our data on parallel exchange rates and reclassifyaccordingly though in our ner grid20
IIIB Using the Chronologies
The 153 individual country chronologies are also a centralpoint of departure from all previous efforts to classify regimes Inthe rst instance the data are constructed by culling informationfrom annual issues of various secondary sources including PickrsquosCurrency Yearbook World Currency Yearbook Pickrsquos Black Mar-ket Yearbook International Financial Statistics the IMFrsquos An-nual Report on Exchange Rate Arrangements and Exchange Re-strictions and the United Nations Yearbook Constructing ourdata set required us to sort and interpret information for everyyear from every publication above Importantly we draw on na-tional sources to investigate apparent data errors or inconsisten-cies More generally we rely on the broader economics literatureto include pertinent information such as the distribution oftransactions among ofcial and parallel markets21
The chronologies allow us to date dual or multiple exchangerate episodes as well as to differentiate between preannouncedpegs crawling pegs and bands from their de facto counterpartsWe think it is important to distinguish between say de facto pegsor bands from announced pegs or bands because their propertiesare potentially different22 At the very least we want to providefuture researchers with the data needed to ask a variety ofquestions about the role of exchange rate arrangements The
19 Our classication of ldquofreely oatingrdquo is the analogue of ldquoindependentlyoatingrdquo in the ofcial classication
20 When the parallel market premium is consistently (ie all observationswithin the ve-year window) in single digits we nd that in nearly all these casesthe ofcial and parallel rates yield the same classication
21 See Marion [1994] for instance22 Policy-makers may not be indifferent between the two In theory at least
announcements of pegs bands and so on can act as a coordinating device whichby virtue of being more transparent could invite speculative attacks
18 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
chronologies also ag the dates for important turning points suchas when the exchange rate rst oated or when the anchorcurrency was changed
Table IV gives an example of one of our 153 chronologies (seebackground material) for the case of Chile The rst column givescritical dates Note that we extend our chronologies as far back aspossible (even though we can only classify from 1946 onwards) inthe case of Chile we go back to 1932
The second column lists how the arrangement is classiedPrimary classication refers to the classication according to ourNatural algorithm which may or may not correspond to theofcial IMF classication (shown in parentheses in the secondcolumn of Table IV) Secondary and tertiary classications aremeant only to provide supplemental information as appropriateSo for example from November 1952 until April 1956 Chilersquosination was above 40 percent and hence its primary classica-tion is freely fallingmdashthat is the only classication that mattersfor the purposes of the Natural algorithm For those interested inadditional detail however we also note in that column that themarket-determined exchange rate was a managed oat along thelines described in detail in the Appendix (secondary) and thatfurthermore Chile had multiple exchange rates (tertiary) Thisadditional information may be useful for example for research-ers who are not interested in treating the high ination casesseparately (as we have done here) In this case they would havesufcient information to place Chile in the 1952ndash1956 period inthe managed oat category Alternatively for those researcherswho wish to treat dual or multiple exchange rate practices as aseparate category altogether (say because these arrangementsusually involve capital controls) the second column (under sec-ondary or tertiary classication) provides the relevant informa-tion to do that sorting accordingly
As one can see although Chile unied rates on September1999 it previously had some form of dual or multiple ratesthroughout most of its history In these circumstances we reit-erate that our classication algorithm relies on the market-de-termined rather than the ofcial exchange rate23 Over some
23 The other Chronologies do not contain this information but the annualofcial IMF classication for the countries in the sample is posted at httpwwwpuafumdedufacultypapersreinhartreinharthtm
19EXCHANGE RATE ARRANGEMENTS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
AS
AM
PL
EC
HR
ON
OL
OG
YIN
TH
EN
AT
UR
AL
CL
ASS
IFIC
AT
ION
SC
HE
MEC
HIL
E1
932ndash
2001
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Sep
tem
ber
161
925ndash
Apr
il20
193
2P
egG
old
stan
dard
For
eign
exch
ange
cont
rols
are
intr
oduc
edon
July
301
931
Apr
il20
193
2ndash19
37D
ualm
arke
tP
ound
Ster
ling
isre
fere
nce
curr
ency
Sus
pens
ion
ofgo
ldst
anda
rd
1937
ndashFeb
ruar
y19
46M
anag
edo
atin
gM
ulti
ple
rate
sU
Sdo
llar
beco
mes
the
refe
renc
ecu
rren
cy
Mar
ch19
46ndashM
ay19
47F
reel
yfa
lling
Man
aged
oat
ing
Mul
tipl
era
tes
June
1947
ndashOct
ober
1952
Man
aged
oat
ing
Mul
tipl
era
tes
Nov
embe
r19
52ndashA
pril
161
956
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sA
pril
161
956ndash
Aug
ust
1957
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tR
ate
stru
ctur
eis
sim
pli
eda
nda
dual
mar
ket
iscr
eate
dS
epte
mbe
r19
57ndashJ
une
1958
Man
aged
oat
ing
Dua
lmar
ket
July
1958
ndashJan
uary
119
60F
reel
yfa
lling
Man
aged
oat
ing
Dua
lmar
ket
Janu
ary
119
60ndashJ
anua
ry15
196
2P
egto
US
dolla
rT
hees
cudo
repl
aces
the
peso
Ja
nuar
y15
196
2ndashN
ovem
ber
1964
Fre
ely
falli
ngM
anag
edo
atin
gM
ulti
ple
rate
sF
reel
yfa
lling
sinc
eA
pril
1962
D
ecem
ber
1964
ndashJun
e19
71M
anag
edo
atin
gM
ulti
ple
rate
s(P
eg)
July
1971
ndashJun
e29
197
6F
reel
yfa
lling
Mul
tipl
eex
chan
gera
tes
(Peg
thro
ugh
1973
-man
aged
oat
ing
afte
rwar
ds)
On
Sep
tem
ber
291
975
the
peso
repl
aced
the
escu
do
Oct
ober
1973
clas
sie
sas
ahy
per
oat
June
291
976ndash
Janu
ary
1978
Fre
ely
falli
ngC
raw
ling
peg
toU
Sdo
llar
(Man
aged
oat
ing)
Feb
ruar
y19
78ndashJ
une
1978
Pre
anno
unce
dcr
awlin
gpe
gto
US
dolla
rF
reel
yfa
lling
(Man
aged
oat
ing)
The
Tab
lita
Pla
n
July
1978
ndashJun
e30
197
9P
rean
noun
ced
craw
ling
peg
toU
Sdo
llar
(Peg
)T
heT
ablit
aP
lan
June
301
979ndash
June
151
982
Peg
toU
Sdo
llar
(Peg
)T
hese
cond
phas
eof
the
Tab
lita
Pla
nJu
ne15
198
2ndashD
ecem
ber
1982
Fre
ely
falli
ngM
anag
edo
atin
gD
ualm
arke
tJa
nuar
y19
83ndashD
ecem
ber
819
84M
anag
edo
atin
gD
ualm
arke
t(M
anag
edo
atin
g)P
aral
lelm
arke
tpr
emiu
mre
ache
s10
2pe
rcen
tin
earl
y19
83O
nM
arch
1983
the
inte
ntio
nto
follo
wa
PP
Pru
lew
asan
noun
ced
20 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
TA
BL
EIV
(CO
NT
INU
ED
)
Dat
eC
lass
ica
tion
prim
ary
seco
nda
ryt
erti
ary
(of
cial
IMF
clas
si
cati
onin
pare
nth
eses
)C
omm
ents
Dec
embe
r8
1984
ndashJan
uary
1988
Man
aged
oat
ing
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
leT
heof
cia
lrat
eis
kept
wit
hin
a6
2cr
awlin
gba
ndto
US
dolla
rF
ebru
ary
1988
ndashJan
uary
119
89D
efa
cto
craw
ling
band
arou
ndU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
PP
Pru
le 6
5ba
ndO
fci
alpr
eann
ounc
ed6
3cr
awlin
gba
ndto
US
dolla
rW
hile
the
ofc
ialr
ate
rem
ains
wit
hin
the
prea
nnou
nced
band
par
alle
lmar
ket
prem
ium
rem
ain
indo
uble
digi
ts
Janu
ary
119
89ndashJ
anua
ry22
199
2P
rean
noun
ced
craw
ling
band
arou
ndU
Sdo
llar
Dua
lm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dw
idth
is6
5
Janu
ary
221
992ndash
Janu
ary
201
997
De
fact
ocr
awlin
gba
ndar
ound
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)P
PP
rule
Ban
dis
65
The
reis
anof
cia
lpre
anno
unce
d6
10
craw
ling
band
toU
Sdo
llar
Par
alle
lpre
miu
mfa
llsbe
low
15pe
rcen
tan
din
tosi
ngle
digi
ts
Janu
ary
201
997ndash
June
251
998
De
fact
ocr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)O
fci
alpr
eann
ounc
edcr
awlin
g6
125
ba
ndto
US
dolla
rde
fact
oba
ndis
65
Ju
ne25
199
8ndashSe
ptem
ber
161
998
Pre
anno
unce
dcr
awlin
gba
ndto
US
dolla
rD
ualm
arke
t(M
anag
edo
atin
g)6
275
ba
nd
Sep
tem
ber
161
998ndash
Dec
embe
r22
199
8P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
63
5ba
nd
Dec
embe
r22
199
8ndashSe
ptem
ber
219
99P
rean
noun
ced
craw
ling
band
toU
Sdo
llar
Dua
lmar
ket
(Man
aged
oat
ing)
68
band
Sep
tem
ber
219
99ndashD
ecem
ber
2001
Man
aged
oat
ing
(Ind
epen
dent
lyo
atin
g)R
ates
are
uni
ed
Ref
eren
cecu
rren
cyis
the
US
dolla
rD
ata
avai
labi
lity
O
fci
alra
te1
900
1ndash20
011
2P
aral
lel
rate
194
61ndash
1998
12
21EXCHANGE RATE ARRANGEMENTS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
periods the discrepancy between the ofcial and parallel ratehowever proved to be small For example from January 1992onwards the parallel market premium remained in single digitsand our algorithm shows that it makes little difference whetherthe ofcial or parallel rate is used In these instances we leavethe notation in the second column that there are dual rates (forinformation purposes) but also note in the third column that thepremium is in single digits As noted Chile has also experiencedseveral periods where the twelve-month monthly ination ex-ceeded 40 percent Our algorithm automatically categorizes theseas freely falling exchange rate regimesmdashunless there is a prean-nounced peg crawling peg or narrow band that is veried as wasthe case when the Tablita program was introduced on February1978
The third column in our chronology gives further sundryinformation on the regimemdash eg the width of the announced andde facto bands etc For Chile which followed a crawling bandpolicy over many subperiods it is particularly interesting to notethe changes over time in the width of the bands The third columnalso includes information about developments in the parallelmarket premium and currency reform As an example of theformer we note that since 1992 the parallel premium slipped intosingle digits an example of the latter is given for Chile when thepeso replaced the escudo in 1975
The top panel of Figure V plots the path of the ofcial andmarket-determined exchange rate for Chile from 1946 It is evi-dent that through much of the period shown the arrangementwas one of a crawling peg or a crawling band with the rate ofcrawl varying through time and notably slowing as inationbegan to stabilize following the Tablita plan of the early 1980sThe bottom panel plots the parallel market premium (in percent)This pattern is representative of many other countries in oursample the premium skyrockets in the periods of economic andpolitical instability declines into single digits as credible policiesare put in place and capital controls are eased As we will discussin the next section the Chilean case is also illustrative in thatcrawling pegs or bands are quite common Figure VI whichshows the path of the exchange rate for the Philippines Indiaand Greece provides other examples of the plethora of crawlingpegs or bands in our sample
22 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VChile Ofcial and Market-Determined Exchange Rates and the Parallel
Market PremiumJanuary 1946ndashDecember 1998
Sources InternationalMonetary Fund Annual Report on Exchange Arrangementsand Exchange Restrictions and International Financial Statistics Pick and Sedillot[1971] International Currency Analysis World Currency Yearbook various issues
23EXCHANGE RATE ARRANGEMENTS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VIThe Prevalence of Crawling Pegs and Bands
Sources Pick and Sedillot [1971] International Currency Analysis World Cur-rency Yearbook various issues
24 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
IIIC Alternative Taxonomies Comparing the Basic Categories
Altogether our taxonomy of exchange rate arrangementsincludes the fourteen classications sketched in Table V (or f-teen if hyperoats are treated as a separate category) Of coursefourteen (or fteen) buckets are not exhaustive for example ifone wishes to distinguish between forward- and backward-look-ing crawls or bands along the lines of Cottarelli and Giannini[1998] Given that we are covering the entire post-World War IIperiod we did not have enough information to make that kind ofner distinction Conversely because we sometimes want to com-pare our classication regime with the coarser ofcial one wealso show how to collapse our fourteen types of arrangements intove broader categories see Table V where the least exiblearrangements are assigned the lowest values in our scale
TABLE VTHE FINE AND COARSE GRIDS OF THE NATURAL CLASSIFICATION SCHEME
Natural classication bucket
Numberassigned tocategory in
Finegrid
Coarsegrid
No separate legal tender 1 1Preannounced peg or currency board arrangement 2 1Preannounced horizontal band that is narrower than or equal
to 62 3 1De facto peg 4 1Preannounced crawling peg 5 2Preannounced crawling band that is narrower than or equal
to 62 6 2De facto crawling peg 7 2De facto crawling band that is narrower than or equal to 62 8 2Preannounced crawling band that is wider than 62 9 2De facto crawling band that is narrower than or equal to 65 10 3Noncrawling band that is narrower than or equal to 62a 11 3Managed oating 12 3Freely oating 13 4Freely falling (includes hyperoat) 14 5
Source The authorsa By contrast to the common crawling bands a noncrawling band refers to the relatively few cases that
allow for both a sustained appreciation and depreciation of the exchange rate over time While the degree ofexchange rate variability in these cases is modest at higher frequencies (ie monthly) lower frequencysymmetric adjustment is allowed for
The Appendix provides a detailed discussion of our classication algorithm
25EXCHANGE RATE ARRANGEMENTS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
In the ner grid we distinguish between preannounced pol-icies and the less transparent de facto regimes Since the formerinvolve an explicit announcement while the latter leave it tonancial market analysts to determine the implicit exchange ratepolicy in the ner classication we treat preannouncement asless exible than de facto We accordingly assign it a lower num-ber in our scale Those not interested in testing whether an-nouncements serve as a coordinating device (say to make aspeculative attack more likely) and only interested in sorting outthe degree of observed exchange rate exibility will prefer thecoarser grid However even in the coarse grid it is imperative totreat freely falling as a separate category
IV THE ldquoNATURALrdquo TAXONOMY CRITIQUES AND COMPARISONS
As the previous section described our classication strategyrelies importantly on the observed behavior of the market-deter-mined exchange rate In this section we rst address some poten-tial critiques of our approach including whether a countryrsquos in-ternational reserve behavior should affect its classication andwhether we may be mislabeling some regimes as pegs or crawlssimply due to the absence of shocks We then proceed to compareour results with the ldquoofcial historyrdquo and provide examples ofhow our reclassication may reshape some of the existing evi-dence on the links between exchange rate arrangements andvarious facets of economic activity
IVA The Trilogy Exchange Rates Monetary Policy andCapital Controls
To capture the nuances of any exchange rate arrangementone might also want information on the presence and effective-ness of capital controls the modalities of (sterilized or unsteril-ized) foreign exchange intervention and the extent to whichinterest rates (or other less conventional types of intervention)are used as a means to stabilize the exchange rate Since for thepurposes of universality our classication rests squarely on theunivariate time series behavior of the nominal exchange rates(combined with historical chronologies) in this subsection weaddress some of these limitations to our approach
Some studies have reclassied exchange rate arrangementsby also factoring in the behavior of foreign exchange reserves as
26 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
reported by the IMFrsquos International Financial Statistics24 How-ever as Calvo and Reinhart [2002] note using reserves hasserious limitations In Brazil and in over two dozen other coun-tries foreign exchange market intervention is frequently donethrough purchases and sales of domestic dollar-linked debt25
This debt is not reected in the widely used IFS reserve dataneither were the massive interventions of the Thai authorities inthe forward market during 1997 and in South Africa thereafterFurthermore as nancial liberalization has spread throughoutthe globe there has been a widespread switch from direct inter-vention in the foreign exchange market to the use of interest ratepolicy in the 1990s as a means to stabilize the exchange rate26
Picking up on this kind of policy intervention requires having thepolicy interest ratemdashthe equivalent of the federal funds rate forthe United Statesmdashfor each country Such data are very difcultto come by and none of the other efforts at reclassication havedealt with issue
Other issues arise in the context of the links between mone-tary capital controls and exchange rate policy In particularwhile xing the exchange rate (or having narrow bands or crawl-ing pegs or bands) largely denes monetary policy our two mostexible arrangement categories (managed or freely oating) donot Floating could be consistent with monetary targets interestrate targets or ination targeting the latter being a relativelyrecent phenomenon27 Since our study dates back to 1946 itspans a sea change in capital controls and monetary policy re-gimes and it is beyond the scope of this paper to subdivide themonetary policy framework for the most exible arrangements in
24 For instance the algorithm used by Levy-Yeyati and Sturzenegger [2002]also uses (besides the exchange rate) reserves and base money This gives rise tomany cases of what they refer to as ldquoone classication variable not availablerdquo Thismeans that their algorithm cannot provide a classication for the United Kingdom(where it is hard to imagine such data problems) until 1987 andmdashin the mostextreme of casesmdashsome developing countries cannot be classied for any year overtheir 1974ndash2000 sample
25 See Reinhart Rogoff and Savastano [2003] for a recent compilation ofdata on domestic dollar-linked debt
26 There are plenty of recent examples where interest rates were jacked upaggressively to fend off a sharp depreciation in the currency Perhaps one of themore obvious examples is in the wake of the Russian default in August 1998 whenmany emerging market currencies came under pressure and countries like Mexicoresponded by doubling interest rates (raising them to 40 percent) within a span ofa couple of weeks
27 Indeed several of the ination targeters in our sample (United KingdomCanada Sweden etc) are classied as managed oaters (However it must alsobe acknowledged that there are many different variants of ination targetingespecially in emerging markets)
27EXCHANGE RATE ARRANGEMENTS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
our grid Apart from exchange rate policy however our studysheds considerable light on the third leg of the trinitymdash capitalcontrols While measuring capital mobility has not been the goalof this paper our data consistently show that the parallel marketpremium dwindles into insignicance with capital market inte-gration providing a promising continuous measure of capitalmobility
IVB Exchange Rates and Real Shocks
Ideally one would like to distinguish between exchange ratestability arising from deliberate policy actions (whether its directforeign exchange market intervention or interest rate policy asdiscussed) and stability owing to the absence of economic orpolitical shocks In this subsection we provide evidence that if theexchange rate is stable and it is accordingly treated in our de jureapproach to classication it is typically not due to an absence ofshocks
Terms of trade shocks are a natural source of potentialshocks particularly for many developing countries Similarly thepresence (or absence) of shocks is likely to be reected in thevolatility of real GDP To investigate the incidence and size ofterms of trade shocks we constructed monthly terms of tradeseries for 172 countries over the period 1960 ndash200128 The termsof trade series is a geometric weighted average of commodityprices (xed weights based on the exports of 52 commodities)
Table VI presents a summary by region of the individualcountry ndings The rst column shows the share of commodi-ties in total exports while s Dtot denotes the variance of themonthly change in the terms of trade of the particular regionrelative to Australia Australia is our benchmark as it is both acountry that is a primary commodity exporter and has a oatingexchange rate that by some estimates approximates an optimalresponse to terms of trade shocks (see Chen and Rogoff [2003])The next three columns show the variance of the monthly changein the terms of trade of the region relative to Australia (s Dtot)exchange rate of the individual region relative to Australia (s De)and the variance of the annual change in real GDP of the regionrelative to Australia (s Dy) The last two columns show the
28 Table VI is based on the more extensive results in Reinhart Rogoff andSpilimbergo [2003]
28 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
variance of the exchange rate relative to the variance of the termsof trade (s De)(s Dtot) and output (s De)(s Dy) respectively
A priori adverse terms of trade shocks should be associatedwith depreciations and the converse for positive terms of tradeshocks greater volatility in the terms of trade should go hand-in-hand with greater volatility in the exchange rate (In Chen andRogoff [2003] there is greater volatility even under optimal pol-icy) Table VI reveals several empirical regularities (a) mostcountries (regions) have more variable terms of trade than Aus-traliamdashin some cases such as the Middle East and the Carib-bean as much as three or four times as variable (b) real GDP isalso commonly far more volatile than in Australia (c) most coun-triesrsquo exchange rates appear to be far more stable than Austra-liarsquos as evidenced by relatively lower variances for most of thegroups (d) following from the previous observations the last twocolumns show that for most of the country groupings that thevariance of exchange rate changes is lower than that of changesin the terms of trade or real GDP Taken together the implicationof these ndings is that if the exchange rate is not moving it is
TABLE VITERMS OF TRADE OUTPUT AND EXCHANGE RATE VARIABILITY
VARIANCE RATIOS (NORMALIZED TO AUSTRALIA AND EXCLUDES FREELY
FALLING EPISODES)
Region Share s Dtots
Des
Dys De
s Dtots Des Dy
North Africa 051 329 093 254 064 023Rest of Africa (excluding CFA) 056 292 287 250 129 138Middle East 060 415 095 348 033 050Development AsiaPacic 034 202 085 240 054 044Industrialized Asia 018 082 097 115 123 086Caribbean 050 415 067 240 020 035Central America 062 302 049 211 021 028South America 063 203 108 215 066 052Central East Europe 024 060 103 151 166 078Western Europe 018 175 084 125 076 056North America 033 164 060 112 047 054
Source Reinhart Rogoff and Spilimbergo [2003] and sources cited thereinThe variable denitions are as follows Share 5 share of primary commodities to total exports the next
three columns show the variance of the monthly change in the terms of trade of the region relative toAustralia (s Dtot) the variance of the monthly change in the exchange rate of the individual region relativeto Australia (s De) and the variance of the annual change in real GDP of the region relative to Australia (sDy) the last two columns show the variance of the exchange rate relative to the variance of the terms of trade(s De)(s Dtot) and output (s De)(s Dy) respectively
29EXCHANGE RATE ARRANGEMENTS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
not for lack of shocks Of course terms of trade are only one classof shocks that can cause movement in the exchange rate Thusconsidering other kinds of shocksmdashpolitical and economic domesticand internationalmdashwould only reinforce the results presented here
IVC Fact and Fiction Natural and Articial
We are now prepared to contrast the ofcial view of thehistory of exchange rate regimes with the view that emerges fromemploying our alternative methodology To facilitate compari-sons we will focus mainly on the coarse grid version of theNatural system
Figure VII highlights some of the key differences between theNatural and IMF classications The dark portions of the barsdenote the cases where there is overlap between the IMF and theNatural classication29 The white bar shows the cases where theIMF labels the regime in one way (say a peg in 1970 ndash1973) andthe Natural labels it differently Finally the striped portions ofthe bars indicate the cases where the Natural classication labelsthe regime in one way (say freely falling 1991ndash2001) and theIMF labels differently (say freely oating) As shown in FigureVII according to our Natural classication system about 40percent of all regimes in 1950 were pegs (since many countrieshad dualparallel rates that did not qualify as pegs) Figure VIIalso makes plain that some of the ldquopegsrdquo in our classication werenot considered pegs under the ofcial classication in turn ouralgorithm rejects almost half of the ofcial pegs as true pegs Ourreclassication of the early postwar years impacts not only ondeveloping countries but on industrialized countries as wellnearly all the European countries had active parallel marketsafter World War II
A second reason why our scheme shows fewer pegs is that theIMFrsquos pre-1997 scheme allowed countries to declare their regimesas ldquopegged to an undisclosed basket of currenciesrdquo This notablynontransparent practice was especially popular during the 1980sand it was also under this that a great deal of managed oatingfreely oating and freely falling actually took place
For the period 1974 ndash1990 the ofcial classication hasroughly 60 percent of all regimes as pegs our classication hasonly half as many Again as we see in Figure VII this comparison
29 Specically both classications assigned the regime for a particularcountry in a given particular year to the same category
30 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
understates the differences since some of our pegs are not ofcialpegs and vice versa For the years 1974 ndash1990 and 1991ndash2001one can see two major trends First ldquofreely fallingrdquo continues tobe a signicant category accounting for 12 percent of all regimesfrom 1974 ndash1990 and 13 percent of all regimes from 1991ndash2001For the transition economies in the 1990s over 40 percent of theobservations are in the freely falling category Of course what weare reporting in Figure VII is the incidence of each regimeClearly future research could use GDP weights andmdashgiven that
FIGURE VIIComparison of Exchange Rate Arrangements According to the IMF Ofcial and
Natural Classications 1950ndash2001Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The dark bars show the overlap between the IMF and Natural classication(ie for that particular year the IMF and Natural classications coincide) thewhite bars show the cases where the IMF classication labeled the regime in oneway (say a peg in 1974ndash1990) and the Natural classication labeled it differentlythe striped bars indicate the cases where the Natural classication labeled theregime in one way (say freely falling) and the IMF labeled it differently (sayfreely oating)
31EXCHANGE RATE ARRANGEMENTS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
low-income countries are disproportionately represented in thefreely falling categorymdashthis would reveal a lower importance tothis category30
Second the Natural classication scheme reveals a bunchingto the middle in terms of exchange rate exibility when com-pared with the ofcial monetary history of the world Limitedexibilitymdashwhich under the Natural classication is dominatedby de facto crawling pegsmdashbecomes notably more importantFrom being a very small class under the ofcial scheme theNatural classication algorithm elevates limited exibility to thesecond most important grouping over the past decade just behindpegs Another startling difference is the reduced importance offreely oating According to the ofcial classication more than30 percent of countries were independently oating during 1991ndash2001 According to the Natural classication less than 10 percentwere freely oating This is partly a manifestation of what Calvoand Reinhart [2002] term ldquofear of oatingrdquo but equally becausewe assign high ination oats (including ones that are ofciallyldquopegsrdquo) to our new freely falling category Indeed more countrieshad freely falling exchange rates than had freely oating ex-change rates
The contrast between the IMF and Natural classicationsystems becomes even more striking when one sees just howsmall the overlap is between the two classications country bycountry and year by year As shown in Table VII if the IMFdesignation of the regime is a peg (1970 ndash2001) there is a 44percent probability that our algorithm will place it into a moreexible arrangement If the ofcial regime is a oat there is a 31percent chance we will categorize it as a peg or limited exibilityIf the ofcial regime is a managed oat there is a 53 percentchance our algorithm will categorize it as a peg or limited exi-bility Whether the ofcial regime is a oat or peg it is virtuallya coin toss whether the Natural algorithm will yield the sameresult The bottom of the table gives the pairwise correlationbetween the two classications with the ofcial classicationrunning from 1 (peg) to 4 (independently oating) and the Nat-ural classication running from 1 (peg) to 5 (freely falling) Thesimple correlation coefcient is only 042 As one can conrm from
30 GDP weights and population weights would of course present verydifferent pictures For example the United States and Japan alone would increasethe worldrsquos share of oaters if it were GDP weights while weight by populationwould increase the weight of xers by China alone
32 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
the chronologies the greatest overlap occurs in the classicationof the G3 currencies and of the limited exibility European ar-rangements Elsewhere and especially in developing countriesthe two classications differ signicantly as we shall see
IVD The Pegs That Float
Figure VIII plots the parallel market premium since January1946 in percent for Africa Asia Europe and Western HemisphereAs is evident from the Figure VIII for all the regions except Europeit would be difcult to make the case that the breakdown of BrettonWoods was a singular event let alone a sea change31 For thedeveloping world the levels of pre- and post-1973 volatilities in themarket-determined exchange rate as revealed by the parallel mar-ket premium are remarkably similar Note that for all regions weexclude the freely falling episodes that would signicantly increasethe volatility but also distort the scale To give a avor of thecross-country variation within region and across time the dashedline plots the regional average plus one standard deviation (calcu-lated across countries and shown as a ve-year moving average)
As regards Europe the story told by Figure VIII is consistentwith the characterization of the Bretton Woods system as a periodof when true exchange rate stability was remarkably short-livedFrom 1946 until the arrival of the late 1950s while Europe wasnot oating in the modern sensemdashas most currencies were not
31 We plot the premium rather than the market-determined rate as itallows us to aggregate across countries in comparable units (percent)
TABLE VIIFLOATING PEGS AND PEGGED FLOATS REVISITING THE PAST 1970ndash2001
Conditional probability that the regime isIn
percent
ldquoOtherrdquo according to NCa conditional on being classied ldquoPegrdquo by IMF 445ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoManaged Floatingrdquo by IMF 532ldquoPegrdquo or ldquoLimited Flexibilityrdquo according to NC conditional on being
classied ldquoIndependently Floatingrdquo by IMF 315Pairwise correlation between IMF and NC classications 420
Sources The authorsrsquo calculationsa NC refers to the Natural Classication ldquoOtherrdquo according to NC includes limited exibility managed
oating freely oating and freely falling
33EXCHANGE RATE ARRANGEMENTS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
FIGURE VIIIAverage Monthly Parallel Market Premium 1946ndash1998
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average monthly parallel market premium whilethe dashed line shows the ve-year moving average of plus one standard devia-tion The regional averages are calculated excluding the freely falling episodes
34 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
convertiblemdashit had some variant of de facto oating under theguise of pegged ofcial exchange rates Each time ofcial ratesare realigned the story had already unfolded in the parallelmarket (as shown earlier in Figure II) While the volatility of thegap between the ofcial rate and the market exchange rate is notquite in the order of magnitude observed in the developing worldthe volatility of the parallel rate is quite similar to the volatilityof todayrsquos managed or freely oating exchange rates32
There are many cases that illustrate clearly that littlechanged before and after the breakup of Bretton Woods33
Clearly more careful statistical testing is required to make cate-gorical statements about when a structural break took place butit is obvious from the gures that whatever break might havetaken place hardly lives up to the usual image of the move fromxed to exible rates
IVE The Floats That Peg
Figure IX provides a general avor of how exchange rateexibility has evolved over time and across regions The gureplots ve-year moving averages of the probability that themonthly percent change in the exchange rate remains within a 2percent band for Africa Asia Europe and Western Hemisphere(excluding only the United States) Hence under a pegged ar-rangement assuming no adjustments to the parity these proba-bilities should equal 100 percent As before we exclude the freelyfalling episodes For comparison purposes the gures plot theunweighted regional averages against the unweighted averagesfor the ldquocommitted oatersrdquo (The committed oaters include thefollowing exchange rates against the dollar Yen DM (euro)Australian dollar and the UK pound) The dashed lines whichshow plusminus one standard deviation around the regionalaverages highlight the differences between the group of oatersand the regional averages
It is evident for all regions (this applies the least to Africa)that the monthly percent variation in the exchange rate has
32 See Bordo [1993] on Bretton Woods and Bordo [2003] on a historicalperspective on the evolution of exchange rate arrangements
33 The country-by-country gures in ldquoThe Country Chronologies and Chart-book Background Material to A Modern History of Exchange Rate ArrangementsA Reinterpretationrdquo at httpwwwpuafumdedufacultypapersreinhartreinharthtm are particularly revealing in this regard
35EXCHANGE RATE ARRANGEMENTS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
FIGURE IXAbsolute Monthly Percent Change in the Exchange Rate Percent ofObservations within a 62 Percent Band (ve-year moving average)
Sources International Monetary Fund Annual Report on Exchange Arrange-ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The solid line represents the average for the group while the dashed lines showplusminus one standard deviation The regional averages are calculated exclud-ing the freely falling episodes
36 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
typically been kept to a minimummdashthere is a great deal ofsmoothing of exchange rate uctuations in all regions when com-pared with the usual monthly variations of the committed oat-ers The smoothing is most evident in Asia where the indexhovers around 90 percent for most of the period versus 60ndash70percent for the oaters Hence over time the nature of theclassication problem has evolved from labeling something as apeg when it is not to labeling something as oating when thedegree of exchange rate exibility has in fact been very limited
IVF Does the Exchange Rate Regime Matter
The question of whether the exchange rate arrangementmatters for various facets of economic activity has indeed been afar-reaching issue over the years in the literature on interna-tional trade and nance and is beyond the scope of this paper Inthis subsection we present a few simple exercises that do notspeak to possible causal patterns between exchange rate regimesand economic performance but are meant as illustrative of thepotential usefulness of our classication First consider TableVIII which separates dualparallel markets from all the otherregimes where the ldquoexchange rate is unitaryrdquo to employ thelanguage of the IMF The top row shows average ination ratesand real per capita GDP growth for the period 1970 ndash2001 for dualarrangements separately from all other regimes This two-waysplit drastically alters the picture presented by the IMFrsquos classi-cation in the top and fourth rows of Table IX which does not
TABLE VIIIINFLATION AND PER CAPITA REAL GDP GROWTH
A COMPARISON OF DUAL (OR MULTIPLE) AND UNIFIED EXCHANGE RATE SYSTEMS1970ndash2001
RegimeAverage annual
ination rateAverage per capitareal GDP growth
Unied exchange rate 198 18Dual (or multiple) exchange rates 1625 08
Sources International Monetary Fund Annual Report on Exchange Arrangements and Exchange Re-strictions and International Financial Statistics Pick and Sedillot [1971] International Currency AnalysisWorld Currency Yearbook various issues
The averages for the two regime types (unied and dual) are calculated on a country-by-country andyear-by-yearbasis Thus if a country has a unied exchange rate for most of the year the observation for thatyear is included in the averages for unied rates if in the following year that same country introduces a dualmarket (or multiple rate) for most of the year the observation for that year is included in the average for dualrates This treatment allows us to deal with transitions across regime types over time
37EXCHANGE RATE ARRANGEMENTS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
treat dual markets as a separate category Dual (or multiple)exchange rate episodes are associated with an average inationrate of 163 percent versus 20 percent for unied exchange mar-ketsmdashgrowth is one percentage point lower for dual arrange-ments The explanation for this gap between the outcomes shownin Table VIII and the IMFrsquos in Table IX is twofold First 62percent of the freely falling cases during 1970 ndash2001 were associ-ated with parallel markets or dual or multiple exchange ratesSecond the high ination cases classied by the IMF as freelyoating were moved to the freely falling category Natural classi-cation Again we caution against overinterpreting the results inTable VIII as evidence of causality as exchange controls and dualmarkets are often introduced amid political and economic cri-sesmdashas the recent controls in Argentina (2001) and Venezuela(2003) attest
As Table IX highlights according to the IMF only limitedexibility cases record moderate ination On the other handfreely oating cases record the best ination performance (9percent) in the Natural classication Freely falling regimes ex-hibit an average annual ination rate 443 percent versus anination average in the 9 to 17 percent range for the othercategories (Table IX)
TABLE IXDO CLASSIFICATIONS MATTER
GROWTH INFLATION AND TRADE ACROSS REGIMES 1970ndash2001
Classication scheme PegLimited
exibilityManagedoating
Freelyoating
Freelyfalling
Average annual ination rateIMF Ofcial 388 53 748 1739 naNatural 159 101 165 94 4433
Average annual per capita real GDP growthIMF Ofcial 14 22 19 05 naNatural 19 24 16 23 225
Exports plus imports as a percent of GDPIMF Ofcial 699 810 658 606 naNatural 787 803 612 449 571
Source International Monetary Fund World Economic OutlookAn na denotes not available The averages for each regime type (peg limited exibility etc) are
calculated on a country-by-countryand year-by-year basis Thus if a country has a pegged exchange rate formost of the year the observation for that year is included in the averages for pegs if in the following year thatsame country has a managed oat for most of the year the observation for that year is included in the averagefor managed oats This treatment allows us to deal with transitions across regime types over time
38 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
The contrast is also signicant both in terms of the level ofper capita GDP (Figure X) and per capita growth (Figure XI andTable IX) Freely falling has the lowest per capita income (US$3476) of any categorymdash highlighting that the earlier parallel tothe HIPC debtor is an apt onemdashwhile freely oating has thehighest (US $13602) In the ofcial IMF classication limitedexibility which was almost entirely comprised of Europeancountries shows the largest per capita income
Growth is negative for the freely falling cases (225 percent)versus growth rates in the 16ndash24 percent range for the othercategories Once freely falling is a separate category the differ-ences between our other classications pale relative to the differ-ences between freely falling and all others (Table VIII) In theofcial IMF classication freely oating shows a meager averagegrowth rate of 05 percent for the independently oating casesFor the Natural classication the average growth rate quadru-ples for the oaters to 23 percent Clearly this exercise high-lights the importance of treating the freely falling episodesseparately
FIGURE XPPP Adjusted GDP per Capita across Regime Types 1970ndash2001
(averaging over all regions)
39EXCHANGE RATE ARRANGEMENTS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
V CONCLUDING REMARKS
According to our Natural classication across all countriesfor 1970 ndash2001 45 percent of the observations ofcially labeled asa ldquopegrdquo should in fact have been classied as limited exibilitymanaged or freely oatingmdash or worse ldquofreely fallingrdquo Post-Bretton Woods a new type of misclassication problem emergedand the odds of being ofcially labeled a ldquomanaged oatrdquo whenthere was a de facto peg or crawling peg were about 53 percentWe thus nd that the ofcial and other histories of exchangerate arrangements can be profoundly misleading as a strikingnumber of pegs are much better described as oats and viceversa
These misclassication problems may cloud our view of his-tory along some basic dimensions Using the IMFrsquos classication
FIGURE XIReal per Capita GDP Growth across Regime Types 1970ndash2001
(averaging over all regions)Sources International Monetary Fund Annual Report on Exchange Arrange-
ments and Exchange Restrictions and International Financial Statistics Pick andSedillot [1971] International Currency Analysis World Currency Yearbook vari-ous issues
The averages for each regime type (peg limited exibility etc) are calculatedon a country-by-country and year-by-year basis Thus if a country has a peggedexchange rate for most of the year the observation for that year is included in theaverages for pegs if in the following year that same country has a managed oatfor most of the year the observation for that year is included in the average formanaged oats This treatment allows us to deal with transitions across regimetypes over time
40 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
for the period 1970 to 2001 for instance one would conclude thata freely oating exchange rate is not a very attractive optionmdashitproduces an average annual ination rate of 174 percent and apaltry average per capita growth rate of 05 percent This is theworst performance of any arrangement Our classication pre-sents a very different picture free oats deliver an average in-ation that is less than 10 percent (the lowest of any exchangerate arrangement) and an average per capita growth rate of 23percent Equally importantly we nd that unied exchange rateregimes vastly outperform dual or multiple exchange rate ar-rangements although one cannot necessarily interpret these dif-ferences as causal While we have focused in this paper on theexchange rate arrangement classication issue the country his-tories and data provided in this paper may well have conse-quences for theory and empirics going forward especially theissue of accounting for dual an parallel markets
In her classic history of the IMF de Vries [1969] looked backat the early years of the Bretton Woods regime and noted
Multiple exchange rates were one of the rst problems that faced theFund in 1946 and have probably been its most common problem in theeld of exchange rates An impressive number and diversity of coun-tries in the last twenty years have experimented with one form oranother of what the Fund has called multiple currency practices atleast for a few if not most of their transactions The problem ofmultiple rates then never seems entirely at an end
Thirty-four years have passed since this history was written andmultiple exchange rate practices are showing no signs of becom-ing passe On December 2001 Argentina suspended convertibilityand in so doing segmented the market for foreign exchangewhile on February 7 2003 Venezuela introduced strict new ex-change controlsmdashde facto creating a multiple exchange rate sys-tem Some things never change
APPENDIX THE DETAILS OF THE ldquoNATURALrdquo CLASSIFICATION
This appendix describes the details of our classication algo-rithm which is outlined in Section III of the paper We concen-trate on the description of the ne grid as shown in Table V
A Exchange Rate Flexibility Indices and Probability Analysis
Our judgment about the appropriate exchange rate classi-cation is shaped importantly by the time-series of several mea-
41EXCHANGE RATE ARRANGEMENTS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
sures of exchange rate variability based on monthly observationsand averaged over two-year and ve-year rolling windows Therst of these measures is the absolute percent change in themonthly nominal exchange rate We prefer the mean absolutechange to the variance to minimize the impact of outliers Theseoutliers arise when for example there are long periods in whichthe exchange rate is xed but nonetheless subject to rare butlarge devaluations
To assess whether exchange rate changes are kept within aband we calculate the probabilities that the exchange rate re-mains within a plusminus 1 2 and 5 percent-wide band over anygiven period Two percent seems a reasonable cutoff to distin-guish between the limited exibility cases and more exible ar-rangements as even in the Exchange Rate Mechanism arrange-ment in Europe 62 14 bands were allowed As with the meanabsolute deviation these probabilities are calculated over two-year and ve-year rolling windows Unless otherwise noted in thechronologies we use the ve-year rolling windows as our primarymeasure for the reasons discussed in Section III of the paperThese rolling probabilities are especially useful to detect implicitunannounced pegs and bands
B De Jure and de Facto Pegs and Bands
Where the chronologies show the authorities explicitly an-nouncing a peg we shortcut the de facto dating scheme describedbelow and zero in on the date announced as the start of the pegWe then conrm (or not) the peg by examining the mean absolutemonthly change over the period following the announcement Thechronologies we develop which give the day month and yearwhen a peg becomes operative are essential to our algorithmThere are two circumstances where we need to go beyond simplyverifying the announced peg The rst case is where our chronol-ogies indicate that the peg applies only to an ofcial rate and thatthere is an active parallel (ofcial or illegal) market As shown inFigure III in these cases we apply the same battery of tests to theparallel market exchange rate as we do to the ofcial rate in aunied market Second there are the cases where the ofcialpolicy is a peg to an undisclosed basket of currencies In thesecases we verify if the ldquobasketrdquo peg is really a de facto peg to a singledominant currency (or to the SDR) If no dominant currency can beidentied we do not label the episode as a peg Potentially of course
42 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
we may be missing some de facto basket pegs though in practicethis is almost certainly not a major issue
We now describe our approach toward detecting de factopegs If there is no ofcially announced peg we test for a ldquodefactordquo peg in two ways First we examine the monthly absolutepercent changes If the absolute monthly percent change in theexchange rate is equal to zero for four consecutive months ormore that episode is classied (for however long its lasts) as a defacto peg if there are no dual or multiple exchange rates Thisallows us to identify short-lived de facto pegs as well as those witha longer duration For instance this lter allowed us to identifythe Philippinesrsquo de facto peg to the US dollar during 1995ndash1997 inthe run-up to the Asian crisis as well as the numerous Europeande facto pegs to the DM well ahead of the introduction of the euroSecond we compute the probability that the monthly exchangerate change remains within a 1 percent band over a rollingve-year period34
P~e 1
where e is the monthly absolute percentage change in the ex-change rate If this probability is 80 percent or higher then theregime is classied as a de facto peg or crawling peg over theentire ve-year period If the exchange rate has no drift it isclassied as a xed parity if a positive drift is present it islabeled a crawling peg and if the exchange rate also goesthrough periods of both appreciation and depreciation it isdubbed a ldquononcrawlingrdquo peg Our choice of an 80 percent thresh-old is not accidental but rather we chose this value because itappears to do a very good job at detecting regimes one would wantto label as pegs without drawing in a signicant number of ldquofalsepositivesrdquo
Our approach regarding preannounced and de facto bandsfollows exactly the same process as that of detecting prean-nounced and de facto pegs we simply replace the 61 band witha 62 band in the algorithm If a band is announced and thechronologies show a unied exchange market we label the epi-sode as a band unless it had already been identied as a de factopeg by the criteria described earlier But importantly we alsoverify whether the announced and de facto bands coincide espe-
34 There are a handful of cases where a two-year window is used In suchinstances it is noted in the chronologies
43EXCHANGE RATE ARRANGEMENTS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
cially as there are numerous cases where the announced (de jure)band is much wider than the de facto band35 To detect such caseswe calculate the probability that the monthly exchange rate changeremains within a 62 band over a rolling ve-year period
P~e 2
If this probability is 80 percent or higher then the regime isclassied as a de facto narrow horizontal crawling or noncrawl-ing band (which allows for both a sustained appreciation anddepreciation) over the period through which it remains continu-ously above the 80 percent threshold
In the case where the preannounced bands are wide (mean-ing equal to or greater than 65) we also verify 65 bands Thespecics for each case are discussed in the country chronologiesFor instance as shown earlier in Table IV in the case of Chile wefound that the de facto band during 1992ndash1998 was narrower(65) than that which was announced at the time (610 and6125) In the case of Libya which had an announced 77 per-cent wide band along a xed central parity pegged to the SDR overthe March 1986 ndashDecember 2001 we detected a 65 crawlingband to the US dollar
C Freely Falling
As we emphasize in the text there are situations almostinvariably due to high ination or hyperination in which thereare mega-depreciations in the exchange rate on a routine andsustained basis We have argued that it is inappropriate andmisleading to lump these casesmdashwhich is what all previous clas-sications (IMF or otherwise) domdashwith oating rate regimes Welabel episodes freely falling on the basis of two criteria Firstperiods where the twelve-month rate of ination equals or ex-ceeds 40 percent are classied as freely falling unless they havebeen identied as some form of preannounced peg or prean-nounced narrow band by the above criteria36 The 40 percent
35 Mexicorsquos exchange rate policy prior to the December 1994 crisis is one ofnumerous examples of this pattern Despite the fact that the band was wideningover time as the oor of the band was xed and the ceiling was crawling the pesoremained virtually pegged to the US dollar for extended periods of time
36 It is critical that the peg criteria supersede the high ination criteria inthe classication strategy since historically a majority of ination stabilizationefforts have used the exchange rate as the nominal anchor and in many of theseepisodes ination rates at the outset of the peg were well above our 40 percentthreshold
44 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
ination threshold is not entirely arbitrary as it has been iden-tied as an important benchmark in the literature on the deter-minants of growth (see Easterly [2001]) As a special subcategoryof freely falling we dub as hyperoats those episodes that meetCaganrsquos [1956] classic denition of hyperination (50 percent ormore ination per month)
A second situation where we classify an exchange rate re-gime as freely falling are the six months immediately following acurrency crisismdashbut only for those cases where the crisis marks atransition from a xed or quasi-xed regime to a managed orindependently oating regime37 Such episodes are typicallycharacterized by exchange rate overshooting This is anothersituation where a large change in the exchange rate does not oweto a deliberate policy it is the reection of a loss of credibility andrecurring speculative attacks To date these crisis episodes wefollow a variant of the approach suggested by Frankel and Rose[1996] Namely any month where the depreciation exceeds orequals 12 12 percent and also exceeds the preceding monthrsquosdepreciation by at least 10 percent is identied as a crisis38 Tomake sure that this approach yields plausible crisis dates wesupplement the analysis with our extensive country chronologieswhich also shed light on balance of payments difculties39 Sinceas a rule freely falling is not typically an explicit arrangement ofchoice our chronologies also provide for all the freely fallingcases the underlying de jure or de facto arrangement (for exam-ple dual markets independently oating etc)
D Managed and Freely Floating
Our approach toward identifying managed and freely oat-ing episodes is basically to create these classes out of the residualpool of episodes that after comprehensive application of ouralgorithm have not been identied as an explicit or implicit pegor some form of band and that are not included in the freely
37 This rules out cases where there was a devaluation and a repeg and caseswhere the large exchange rate swing occurred in the context of an already oatingrate
38 Frankel and Rose [1996] do not date the specic month of the crisis butthe year their criteria call for a 25 percent (or higher) depreciation over the year
39 For instance the Thai crisis of July 1997 does not meet the modiedFrankel-Rose criteria While the depreciation in July exceeded that of the preced-ing month by more than 10 percent the depreciation of the Thai Baht in thatmonth did not exceed 25 percent For these cases we rely on the chronologies ofevents
45EXCHANGE RATE ARRANGEMENTS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
falling category To proxy the degree of exchange rate exibilityunder freely oating and managed oats we construct a compos-ite statistic
eP~e 1
where the numerator is the mean absolute monthly percentchange in the exchange rate over a rolling ve-year period whilethe denominator ags the likelihood of small changes For de jureor de facto pegs this index will be very low (close to or equal tozero) while for the freely falling cases it will be very large Asnoted we only focus on this index for those countries and periodswhich are candidates for freely or managed oating We tabulatethe frequency distribution of our index for the currencies that aremost transparently oating these include US dollarDM-euroUS dollaryen US dollarUK pound US dollarAustralian dollarand US dollarNew Zealand dollar beginning on the date in whichthe oat was announced We pool the observations (the ratio forrolling ve-year averages) for all the oaters So for examplesince Brazil oated the real in January 1999 we would calculatethe ratio only from that date forward If Brazilrsquos ratio falls insidethe 99 percent condence interval (the null hypothesis is freelyoating and hence the rejection region is located at the lower tailof the distribution of the oaterrsquos group) the episode is charac-terized as freely oating If that ratio falls in the lower 1 percenttail the null hypothesis of freely oating is rejected in favor of thealternative hypothesis of managed oat It is important to notethat managed by this denition does not necessarily imply ac-tive or frequent foreign exchange market interventionmdashit refersto the fact that for whatever reason our composite exchange ratevariability index eP(e 1) does not behave like the indicesfor the freely oaters
E Dual or Multiple Exchange Rate Regimes and ParallelMarkets
Dual rates are essentially a hybrid arrangement There arecases or periods in which the premium is nil and stable so that theofcial rate is representative of the underlying monetary policyThe ofcial exchange rate could be pegged crawling or main-tained within some bands or in a few cases allowed to oat Butthere are countless episodes where the divergence between theofcial and parallel rate is so large that the picture is incompletewithout knowledge of what the parallel market rate is doing The
46 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
country chronologies are critical in identifying these episodes Inthe cases where dual or multiple rates are present or parallelmarkets are active we focus on the market-determined ratesinstead of the ofcial exchange rates As shown in Figure III wesubject the market-determined exchange rate (dual multiple orparallel) to the battery of tests described above40 This particularcategory will especially reshape how we view the 1940s throughthe 1960s where about half the cases in the sample involved dualmarkets
UNIVERSITY OF MARYLAND COLLEGE PARK
HARVARD UNIVERSITY
REFERENCES
Bahmani-Oskooee Mohsen Ilir Miteza and A B M Nasir ldquoThe Long-RunRelationship between Black Market and Ofcial Exchange Rates Evidencefrom Panel Cointegrationrdquo Economics Letters LXXVI (2002) 397ndash404
Baxter Marianne and Alan Stockman ldquoBusiness Cycle and Exchange RateRegime Some International Evidencerdquo Journal of Monetary EconomicsXXIII (1989) 377ndash 400
Bordo Michael ldquoThe Bretton Woods International Monetary System A HistoricalOverviewrdquo in A Retrospective on the Bretton Woods System Michael Bordoand Barry Eichengreen eds (Chicago IL University of Chicago Press 1993)pp 3ndash98
mdashmdash ldquoExchange Rate Regimes in Historical Perspectiverdquo National Bureau ofEconomic Research Working Paper No 9654 2003
Cagan Philip ldquoThe Monetary Dynamics of Hyperinationrdquo in Studies in theQuantity Theory of Money Milton Friedman ed (Chicago IL University ofChicago Press 1956) pp 25ndash117
Calvo Guillermo A and Carmen M Reinhart ldquoFear of Floatingrdquo QuarterlyJournal of Economics CXVII (2002) 379ndash 408
Chen Yu-chin and Kenneth S Rogoff ldquoCommodity Currenciesrdquo Journal ofInternational Economics VX (2003) 133ndash160
Claessens Stijn ldquoEstimates of Capital Flight and Its Behaviorrdquo Revista deAnalisis Economico XII (1997) 3ndash34
Cotarelli C and C Giannini ldquoCredibility Without Rules Monetary Frameworksin the Post Bretton-Woods Erardquo IMF Occasional Paper No 154 (WashingtonDC International Monetary Fund 1998)
de Vries Margaret G ldquoMultiple Exchange Ratesrdquo in The International MonetaryFund 1945ndash1965 Margaret de Vries and J Keith Horseeld eds (Washing-ton DC International Monetary Fund 1969) pp 122ndash151
Easterly William The Elusive Quest for Growth (Cambridge MA MIT Press2001)
Frankel Jeffrey A and Andrew K Rose ldquoCurrency Crashes in Emerging Mar-kets An Empirical Treatmentrdquo Journal of International Economics XXXXI(1996) 351ndash368
Ghei Nita Miguel A Kiguel and Stephen A OrsquoConnell ldquoParallel Exchange Ratesin Developing Countries Lessons from Eight Case Studiesrdquo in Parallel Ex-change Rates in Developing Countries Miguel Kiguel J Saul Lizondo andStephen OrsquoConnell eds (New York NY Saint Martinrsquos Press 1997) pp17ndash76
40 There are a few such cases in the sample where only government trans-actions take place at the ofcial rate
47EXCHANGE RATE ARRANGEMENTS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS
Ghosh Atish Anne-Marie Gulde Jonathan Ostry and Holger Wolfe ldquoDoes theNominal Exchange Rate Regime Matterrdquo National Bureau of Economic Re-search Working Paper No 5874 1997
International Currency Analysis World Currency Yearbook (New York NY In-ternational Currency Analysis 1983ndash1998) various issues
International Monetary Fund Annual Report on Exchange Restrictions (Wash-ington DC International Monetary Fund 1949ndash1978) various issues
International Monetary Fund Annual Report on Exchange Arrangements andExchange Restriction (Washington DC International Monetary Fund 1979ndash2001) various issues
Kiguel Miguel J Saul Lizondo and Stephen A OrsquoConnell eds Parallel Ex-change Rates in Developing Countries (New York NY Saint Martinrsquos Press1997)
Levy-Yeyati Eduardo and Federico Sturzenegger ldquoClassifying Exchange RateRegimes Deeds versus Wordsrdquo mimeo Universidad Torcuato Di Tella 2002
Marion Nancy P ldquoDual Exchange Rates in Europe and Latin Americardquo WorldBank Economic Review VIII (1994) 213ndash245
Pick Franz World Currency Reports (New York NY Pick Publishing Corpora-tion 1945ndash1955) various issues
mdashmdash Black Market Yearbook (New York NY Pick Publishing Corporation 1951ndash1955) various issues
mdashmdash Pickrsquos Currency Yearbook (New York NY Pick Publishing Corporation1955ndash1982) various issues
mdashmdash World Currency Reports (New York NY International Currency AnalysisInc 1983ndash1998) various issues
Pick Franz and Rene Sedillot All the Monies of the World A Chronicle ofCurrency Values (New York NY Pick Publishing Corporation 1971)
Reinhart Carmen M and Kenneth S Rogoff ldquoA Modern History of ExchangeRate Arrangements A Reinterpretationrdquo National Bureau of Economic Re-search Working Paper No 8963 2001
Reinhart Carmen M and Kenneth S Rogoff ldquoParts I and II Background Ma-terial to a Modern History of Exchange Rate Arrangements A Reinterpreta-tionrdquo mimeo International Monetary Fund Washington DC 2003 at httpwwwpuafumdedufacultypapersreinhartreinharthtm
Reinhart Carmen M Kenneth S Rogoff and Miguel A Savastano ldquoAddicted toDollarsrdquo National Bureau of Economic Research Working Paper No 100152003
Reinhart Carmen M Kenneth S Rogoff and Antonio Spilimbergo ldquoWhen HardShocks Hit Soft Pegsrdquo mimeo International Monetary Fund WashingtonDC 2003
United Nations United Nations Yearbook (New York United Nations 1946ndash1960) various issues
48 QUARTERLY JOURNAL OF ECONOMICS