occasional papers of the international monetary fund · 2019-12-11 · occasional papers of the...

Occasional Papers of the International Monetary Fund

* 1 . International Capital Markets: Recent Developments and Short-Term Prospects, by a Staff TeamHeaded by R.C. Williams, Exchange and Trade Relations Department. 1980.

2. Economic Stabilization and Growth in Portugal, by Hans O. Schmitt. 1981.

*3. External Indebtedness of Developing Countries, by a Staff Team Headed by Bahram Nowzad andRichard C. Williams. 1981.

*4. World Economic Outlook: A Survey by the Staff of the International Monetary Fund. 1981.

5. Trade Policy Developments in Industrial Countries, by S.J. Anjaria, Z. Iqbal, L.L. Perez, andW.S. Tseng. 1981.

6. The Multilateral System of Payments: Keynes, Convertibility, and the International MonetaryFund's Articles of Agreement, by Joseph Gold. 1981.

7. International Capital Markets: Recent Developments and Short-Term Prospects, 1981, by a StaffTeam Headed by Richard C. Williams, with G.G. Johnson. 1981.

8. Taxation in Sub-Saharan Africa. Part I: Tax Policy and Administration in Sub-Saharan Africa, byCarlos A. Aguirre, Peter S. Griffith, and M. Zuhtu Yucelik. Part II: A Statistical Evaluation ofTaxation in Sub-Saharan Africa, by Vito Tanzi. 1981.

9. World Economic Outlook: A Survey by the Staff of the International Monetary Fund. 1982.

10. International Comparisons of Government Expenditure, by Alan A. Tait and Peter S. Heller.1982.

11. Payments Arrangements and the Expansion of Trade in Eastern and Southern Africa, byShailendra J. Anjaria, Sena Eken, and John F. Laker. 1982.

12. Effects of Slowdown in Industrial Countries on Growth in Non-Oil Developing Countries, byMorris Goldstein and Mohsin S. Khan. 1982.

13. Currency Convertibility in the Economic Community of West African States, by John B.McLenaghan, Saleh M. Nsouli, and Klaus-Walter Riechel. 1982.

14. International Capital Markets: Developments and Prospects, 1982, by a Staff Team Headed by

Richard C. Williams, with G.G. Johnson. 1982.

15. Hungary: An Economic Survey, by a Staff Team Headed by Patrick de Fontenay, 1982.

16. Developments in International Trade Policy, by S.J. Anjaria, Z. Iqbal, N. Kirmani, andL.L. Perez. 1982.

17. Aspects of the International Banking Safety Net, by G.G. Johnson, with Richard K. Abrams.

1983.

18. Oil Exporters' Economic Development in an Interdependent World, by Jahangir Amuzegar. 1983.

19. The European Monetary System: The Experience, 1979-82, by Horst Ungerer, with Owen Evans

and Peter Nyberg. 1983.

20. Alternatives to the Central Bank in the Developing World, by Charles Collyns. 1983.


*Out of print

(Continued on inside back cover)

©International Monetary Fund. Not for Redistribution

Occasional Paper No. 41

Fund-Supported Adjustment Programsand Economic Growth

By Mohsin S. Khan and Malcolm D. Knight

International Monetary FundWashington, D.C.

November 1985


® 1985 International Monetary Fund

International Standard Serial Number: ISSN 0251-6365

International Standard Book Number: ISBN 0-939934-55-8

Price: US$7.50(US$4.50 to university libraries, faculty members,

and students)

Address orders to:External Relations Department, Attention Publications

International Monetary Fund, Washington, D.C. 20431


Contents

Page

PREFATORY NOTE v

I Introduction 1

II Analyzing the Effects of Fund Programs on Economic Growth 2ObjectivesPolicy Content

Description of Fund ProgramsChoice of Policy Instruments

Issues relating to GrowthShort-Term IssuesLong-Term Issues

Empirical Analysis

III Effects of Specific Policy Measures 9Policies to Restrain Demand

Monetary PolicyFiscal Policy

Policies to Stimulate SupplyImproving Resource AllocationIncreasing Capacity Output

Exchange Rate Policies

IV Effects of Policy Packages 18MethodologyEmpirical Evidence

V Simulations of the Growth Effects of Programs 21

VI Conclusions 24

APPENDIX Description of the Simulation Model 26

REFERENCES 31

TABLES

Section

III 1. Short-Run Effect on the Growth Rate of Output of a 10 PercentChange in the Growth of the Money Supply or DomesticCredit 11

2. Effect of Changes in Domestic Credit on Real Private Investment 123. Estimates of Long-Run Price Elasticities of Supply of Selected

Primary Commodities 13

iii


CONTENTS

4. Estimates of the Effects of Increases in Factors of Production onthe Growth of Real GDP 15

5. Effects of a 10 Percent Devaluation on the Growth of Real GDP 17

Appendix

6. Specification of the Model7. Values of Parameters

2729

CHART

1. Effect on Growth Rate of Demand-Management and Supply-SidePolicies 22

iv

The following symbols have been used throughout this paper:

... to indicate that data are not available;

— to indicate that the figure is zero or less than half the final digit shown, orthat the item does not exist;

- between years or months (e.g., 1979-81 or January-June) to indicate theyears or months covered, including the beginning and ending years ormonths;

/ between years (e.g., 1980/81) to indicate a crop or fiscal (financial)year.

"Billion" means a thousand million.

Minor discrepancies between constituent figures and totals are due to rounding.


Prefatory Note

This is the first of a group of papers dealing with various aspects of Fund-supportedadjustment programs. The other two, The Global Effects of Fund-SupportedAdjustment Programs by Morris Goldstein and Fund-Supported Programs, FiscalPolicy, and Income Distribution by the Fiscal Affairs Department, will also bepublished in the Fund's Occasional Paper Series. The three papers were the basis for adiscussion at a seminar held in the Fund's Executive Board in September 1985, andthe published versions reflect the comments and suggestions made at that time byExecutive Directors. The views expressed in the present study are the authors' alone,however, and should not necessarily be attributed to Executive Directors or to theFund.

The present paper was prepared in the Research Department of the InternationalMonetary Fund. Its authors are Mohsin S. Khan, Advisor, and Malcolm D. Knight,Chief of the External Adjustment Division. The paper was edited by David D.Driscoll of the Fund's External Relations Department.

V


This page intentionally left blank


I Introduction

The Fund's Articles of Agreement make it clear (Arti-cle I) that promoting the growth of output and trade is aprimary objective of economic policy and that eliminat-ing payments disequilibria should be sought in accor-dance with this objective. Fund-supported adjustmentprograms consequently have to be designed to achieve aviable balance of payments within the context of im-proved long-term growth performance and price stabil-ity.1 Nevertheless, Fund policies and programs havecome under mounting criticism in recent years in thepress, as well as in certain academic circles, for failing toencourage economic growth.2 Indeed, it has been fre-quently argued that rather than fostering the growth ofoutput, Fund programs tend to cause a slowdown ineconomic activity, increased unemployment, and a gen-eral worsening of living standards.

This paper addresses some aspects of this criticism andbrings to bear the empirical evidence available in theliterature on the subject, emphasizing the results fordeveloping countries. The paper focuses exclusively onexisting evidence; no attempt is made to provide any newempirical results. The evidence is largely indirect, sincein most cases the studies from which it is obtained werenot necessarily undertaken with the express intention ofilluminating the strengths and weaknesses of Fund pro-grams. The conclusions drawn from this body of empir-ical evidence accordingly do not provide definitive an-swers to the criticism cited above and are in any eventtentative. The survey does, however, furnish informationabout the current state of empirical knowledge of theissue and identifies gaps that need to be filled if thevalidity of the criticisms of Fund programs is to beproperly assessed.

The plan of this paper is as follows. Section II defines

2This is only one of the many criticisms made of the Fund. For areview of the broad lines of the criticisms that occur most often inappraisals of the Fund and its policies, see Nowzad (1981).

the approach taken to analyze the empirical evidence onFund programs and economic growth. It discusses thecircumstances in which member countries typically seekFund assistance, the specific objectives of Fund pro-grams, the policy content of a typical program, and thecriteria that affect the choice between demand-side andstructural, or supply-side, measures in programs. It thendescribes issues that arise with respect to the effect ofprograms on growth, the empirical approach necessary toexamine these issues properly, the type of evidenceavailable, and finally the approach adopted in this paperin interpreting this evidence.

The empirical evidence that can be used to infer theeffects of Fund programs on growth is examined in thenext two sections of the paper. Section III looks at theevidence produced by econometric models on the effectsof individual policy measures. These measures includemonetary policy, fiscal policy, supply-side policies, andexchange rate policy. Section IV, on the other hand,surveys the evidence on the effects of complete policypackages based on cross-country analyses undertakenboth within the Fund—in connection with reviews ofprograms—as well as outside the institution. As thesetwo sections emphasize, both the model-based and cross-country approaches involve different problems; further-more they do not yield results that are totally consistentwith each other. An attempt is made in Section V toreconcile the conflicting evidence produced by the twoapproaches by using illustrative simulation experiments.These simulations also indicate how alternative combina-tions of demand-side and supply-side measures can beexpected to influence the rate of growth of output in theshort term. Finally, Section VI summarizes the conclu-sions that may legitimately be drawn from the empiricalevidence available at present and points out where addi-tional efforts would likely yield a better understanding ofthe relationship between Fund programs and economicgrowth. The model used for the simulation analysis iscontained in the Appendix.

1See Guitian (1981).

1


II Analyzing the Effects of Fund Programs onEconomic Growth

Objectives

In analyzing the effects of Fund-supported adjustmentprograms on the level or rate of growth of output, it iscrucial first to consider the circumstances in which suchprograms are introduced. Typically the need for a stabili-zation program, whether supported by the Fund or other-wise, arises when a country experiences an imbalancebetween aggregate domestic demand (absorption) andaggregate supply, which is reflected in a worsening of itsexternal payments position. While it is true that suchexternal factors as an exogenous deterioration of theterms of trade or an increase in foreign interest rates canbe responsible for the basic demand-supply imbalances,often these imbalances can be traced to inappropriatedomestic policies that expand aggregate domestic de-mand too rapidly relative to the productive potential ofthe economy and seriously distort relative prices.3 Ifforeign financing is available, the relative expansion ofdomestic demand can persist for extended periods—albeit at the cost of a widening current account deficit, aloss of international competitiveness owing to rapid do-mestic inflation, an inefficient allocation of resourcesbecause of the distortions in relative prices, and a heavierforeign debt burden.

Clearly, this disequilibrium cannot continueindefinitely, as the country steadily loses internationalcompetitiveness and eventually creditworthiness. In theabsence of appropriate policy action, a cessation offoreign financing would impose adjustment on the coun-try, and this forced adjustment is likely to be very disrup-tive. The basic objective of the Fund in these circum-stances is to provide for a more orderly adjustment of theimbalance between absorption and aggregate supply soas to achieve a viable balance of payments positionwithin a reasonable period of time. A viable balance ofpayments has two aspects. First, it implies that the bal-ance of payments problems will not merely be sup-pressed but eradicated, and second, that the im-

provement in the country's external position will bedurable.

The Fund's task is, in the first instance, to ensure thatforeign financing attains a level consistent with the coun-try's present and future debt-servicing capacity. Thismay involve setting limits on foreign borrowing or, ashas been more evident in recent years, ensuring that therequisite inflow of foreign capital is in fact forthcomingto fill the financing gap.4 The permissible rate of foreignborrowing defines the necessary degree of adjustment ofthe imbalances in the economy. To achieve the requiredadjustment, the Fund designs a stabilization program thatincludes measures to restore a sustainable balance be-tween aggregate demand and supply and simultaneouslyto expand the production of tradables, thereby easing thebalance of payments constraint.

Policy Content

Description of Fund Programs

Although stand-by arrangements with the Fund areoften viewed as synonymous with devaluation and do-mestic credit restraint, Fund programs are in fact com-plex packages of policy measures geared to the particularcircumstances of the country.5 More important, thechoice of policies and the nature of the policy mix inprograms result from extensive negotiations between thecountry authorities and the Fund. Aside from monetaryand exchange rate policies, a typical Fund program callsfor fiscal measures, such as reductions in governmentexpenditures and increases in taxation, increases in do-mestic interest rates and producer prices to realisticlevels, policies to raise investment and improve itsefficiency, trade liberalization, and wage restraint. Con-siderable overlap among these various policies does not

3See Khan and Knight (1983) for a discussion of the respective rolesof foreign and domestic factors in the current account outcomes ofnon-oil developing countries during the 1970s.

4For example, during 1983 lending by the Fund exceeded $12billion and the Fund helped to secure over $21 billion in additionalbank lending to countries with programs.

5For a discussion of the types of policies usually contained in Fundprograms, see Crockett (1981) and Guitian (1981).

2


preclude the convenience of grouping them into the fol-lowing three categories: demand-side policies, supply-side policies, and policies to improve international com-petitiveness.

Demand-side policies are measures that influence theaggregate level or rate of growth of domestic demandand absorption. Such policies include the whole range offiscal, monetary, and domestic credit measures associ-ated with traditional macroeconomic policy. Althoughthese policies also affect production and supply, it isuseful at this level of abstraction to label policies thatprimarily affect aggregate absorption as "demand-oriented" policies.

Supply-side policies are intended to increase thevolume of goods and services supplied by the domesticeconomy at any given level of domestic demand. Suchsupply-oriented policies can be divided broadly into twogroups. First, there are policies designed to increasecurrent output by improving the efficiency with whichfactors of production, such as capital and labor, areutilized and allocated among competing uses. This cate-gory includes measures to reduce distortions caused byprice rigidities, monopolies, taxes, subsidies, and traderestrictions. The second group encompasses policies de-signed to raise the long-run rate of growth of capacityoutput. Under this heading are incentives for domesticsaving and investment. Also important are policies de-signed to increase the inflow of foreign savings, whetherin the form of private lending, foreign direct investment,or increased development assistance. These two groupsof supply-side policies are obviously interrelated, sincepolicies that increase current output may, by themselves,lead to a larger flow of saving and investment and ahigher rate of growth of capacity output.

Policies to improve international competitiveness con-tain elements of both demand-side and supply-side poli-cies, since they are based on combinations of measures(such as devaluation cum wage restraint) intended toaffect the program country's real exchange rate. Improv-ing competitiveness is why considerable importance of-ten attaches to the role of exchange rate policies instabilization programs. In general, to the extent that acombination of policies alters the real exchange rate, itwill affect both real domestic absorption and the incen-tive to produce tradable goods.

In summary, it would be misleading to suggest thatFund programs rely exclusively on one or two policyinstruments directed solely at restraining domestic de-mand. As the above discussion has indicated, much moreis involved in the design of Fund programs than a me-chanical application of the simple monetary approach tothe balance of payments, supplemented perhaps by anexchange rate change.6 From this standpoint, it may be

Policy Content

noted that many alternative policies proposed by critics,particularly those relating to the supply side, alreadyform an integral part of Fund programs. For example, inone of the few concrete expositions of an alternativestabilization strategy, Diaz-Alejandro (1984) sets out apolicy package whose general characteristics are in mostrespects indistinguishable from a typical Fund program.7

Considering the case of a country experiencing highinflation and facing an unsustainable balance of pay-ments position, the author proposes a policy packageincluding fiscal and monetary restraint, wage guidelines,increases in domestic interest rates to positive real levels,gradual tariff reductions, the introduction of incentivesfor import substitution and export promotion, measuresto maintain and expand investment, and the adoption of acrawling-peg exchange rate regime, with emphasisplaced in the direction of "undervaluation" of the realexchange rate to support export promotion and importliberalization. Other critics of Fund policies, such asTaylor (1981), have further argued for the use of controlson imports and capital flows as a substitute for demand-management policies and exchange rate action on thegrounds that these are less costly alternatives.

If the guidelines for stabilization suggested by Diaz-Alejandro (1984) are taken as representative of programstypically proposed by the critics of the Fund, there islittle to take issue with. The basic difference, if any,arises from some critics' proposal to use controls as apolicy to correct balance of payments problems. Fundpolicy leans heavily in the direction of eliminating con-trols and restrictions on trade and payments; neverthe-less, the Fund has on occasion accepted the temporaryuse of import controls and export subsidies and, as aninterim arrangement, the continuation of dual exchangemarkets. What the Fund has consistently opposed is theintroduction of new restrictions, as well as the intensifi-cation on a permanent basis of existing restrictions andother distortions in the trade and payments system.8

Although a theoretical case can be made for controls andrestrictions in the short run, in practice it has proveddifficult to manage such systems efficiently and effec-tively over time. Furthermore, such policies, by in-troducing rigidities in the economy and creating incen-tives for the inefficient use of resources and forms ofproduction, can turn out in the long run to be counter-productive in the long run and damaging to the growthpotential of the economy.

6Diaz-Alejandro (1984), for example, characterizes Fund programsin this way.

7 The "real economy" approach advocated by Killick and others(1984) is another example of a specific set of proposals for adjustment.This approach emphasizes structural policies at the sectoral level andargues for adjustment to be spread over a longer period (5 years) thanis considered by the Fund. Other alternative programs either tend to bevery general, with little elaboration of the policies that should beadopted, or are restricted to particular countries and specific episodesof adjustment.

8 The use of controls on trade and payments also is inconsistent withthe objectives of the Fund as set forth in the Articles of Agreement.

3


II • ANALYZING THE EFFECTS OF FUND PROGRAMS ON ECONOMIC GROWTH

Choice of Policy Instruments

A crucial concern that arises in the design of adjust-ment programs is how much emphasis should be placedon supply-side policies relative to demand-side policies.As the need for a stabilization program typically reflectsexcess demand, all programs must involve some degreeof restraint of aggregate domestic demand. This does notmean, however, that adjustment should be based exclu-sively on reducing absorption—the imbalance could inprinciple also be eliminated through expanding domesticsupply. In fact, demand-side and supply-side policies areclosely interrelated. Policies designed to achieve a highergrowth rate in the medium term generally require anincrease in the rate of productive investment, while de-mand-management policies require a reduction in thesavings-investment gap. The policy package, therefore,must be designed to reduce the level of aggregate domes-tic demand and simultaneously to cause a shift in itscomposition away from current consumption and towardfixed capital formation.

Notwithstanding the difficulties of implementing sup-ply-side policies as part of an adjustment program, theFund has stressed their importance in improvingefficiency and the long-term rate of growth. The firstdifficulty is that many types of supply-side measuresimprove output only after a significant delay.9 For exam-ple, investment programs designed to raise the rate ofgrowth of capacity output take time to come to fruition.Steps to create improved incentives for production andexports by eliminating distortions in the structure ofrelative prices take time to exert beneficial effects, par-ticularly if labor and capital are not very mobile amongdifferent activities. In this case, major changes in thepattern of resource allocation may necessitate a longperiod of adjustment during which some factors of pro-duction remain unemployed. As these examples suggest,most supply-side measures—even those designed to alteroutput by improving price incentives rather than by in-creasing investment—may exert beneficial effects onlyover the longer term. These examples also suggest thatthe extent to which supply-side policies can be empha-sized relative to demand-side policies in a program de-pends on the length of the period over which adjustmentis intended to achieve its objective. This horizon is obvi-ously constrained by the amount and duration of foreignfinancing that a Fund program can make available to thecountry through the Fund's own resources, privatebanks, and other sources.

A second constraint on the use of certain supply-sidemeasures is the possibility that they may affect the polit-ical and social objectives of governments. Many govern-ment policies that create distortions (that is, deviations of

9Of course, situations may arise in which the supply response isfairly rapid, that is, if there exists substantial excess capacity in theeconomy.

prices from marginal costs) are designed to achieve ob-jectives other than economic efficiency and may havebeen implemented with full knowledge of their likelyadverse effect on resource allocation. Such policies mayinclude food subsidies, employment programs, restric-tions on imports of certain categories of goods and ser-vices, and capital controls. Changes in such policiesoften have a strong impact on equity as well as economicefficiency, and the Fund is enjoined to respect the viewsof sovereign governments in these matters, although itcan, and frequently does, render advice on the budgetarycosts of such policies.

Even if these difficulties with supply-side policies aresomehow overcome, this does not imply that demand-side policies can be dispensed with. Supply-side policiesby themselves do not guarantee an improvement in thebalance of payments because, other things being equal,aggregate demand can rise beyond sustainable levelseven with increasing aggregate supply, unless it is re-strained from doing so. Consequently, stabilization pro-grams have to use both sets of policies, and the decisionon the relative emphasis that is to be placed on demandand supply measures in Fund programs is based on anumber of criteria. These include:

(i) The nature, magnitude, and likely duration of theexternal payments imbalance. For example, if adeterioration in the country's external terms oftrade causes the balance of payments deficit, theappropriate response would include supply-sidemeasures designed to change the basic structureof production in the economy. In other words, anadverse external development may alter the mixbetween demand and supply measures. By con-trast, if the initial disequilibrium is the result ofexcess aggregate domestic demand, owingperhaps to excessively expansionary fiscal anddomestic credit policies, then the response wouldnormally rely more heavily on demand restraintthan on supply-side measures.

(ii) The initial level of the country's external indebt-edness and the amount of additional financingthat can be expected. These conditions determinethe length of time over which the adjustmentprocess can take place.

(iii) The nature and importance of the constraints fac-ing the government in pursuing policies that haveimportant social and political implications.

Issues Relating to Growth

The general criticism that Fund programs are in somesense inimical to growth can be decomposed into twospecific criticisms. The first is that a number of policyrecommendations contained in Fund programs—particularly those relating to the restraint of aggregatedomestic demand and to altering the exchange rate—are

4


Issues Relating to Growth

thought to exert an adverse effect on economic growth,employment, and the level of activity. The second is thatthe policy recommendations are also unduly harsh,deflating the economy more than is in fact needed tosecure the objectives of the stabilization effort.

In addressing these criticisms, a distinction needs to bedrawn between the short-term issues and the longer-termissues relating to the effects of Fund programs ongrowth.

Short-Term Issues

If the initial problem is excess aggregate domesticdemand, then, in order to achieve the objectives of theadjustment program, absorption must be reduced in theshort run. Although this reduction in absorption can beperceived as representing a decline in living standards, itshould not be regarded as a "cost" of the program, sinceabsorption is merely being brought back into line withavailability of resources. The real issue is how the reduc-tion in absorption—whether brought about by appropri-ate demand-side policies or by exchange rate action—will influence the level and rate of growth of output orreal income. In theory, situations can be conceived inwhich the reduction in absorption can be achieved with-out diminishing output. This result would be possible,for example, if all of the adjustment is entirely confinedto the current account of the balance of payments and isachieved through some combination of an increase inreceipts and a decrease in payments. In practice, how-ever, this extreme result is unlikely to occur, and thereduction in absorption necessary to achieve the objec-tives will generally be accompanied by some fall in thegrowth of output, particularly if inflation has becomeingrained in the system. This decline in the growth rate isa necessary part of the adjustment to eliminate underly-ing imbalances in the economy. In other words, theadjustment aims at leading the economy onto a morestable and sustainable growth path from a higher butunsustainable path that generally accompanies the sup-ply-demand imbalances. The critical question, of course,concerns the size and duration of the short-run effects ofpolicies designed to reduce absorption.

The second line of criticism mentioned above ques-tions the overall design of Fund programs and is moredifficult to assess objectively. Fund programs are notintended to reduce a country's absorption of goods andservices below the level that can be financed (out ofcurrent savings and capital inflows) on a sustainablebasis. Any reductions in absorption and growth that gobeyond the levels necessary to achieve the objectives ofthe program can be fairly viewed as the true "costs" of aprogram. Since the "necessary" reduction in absorptionand the consequent decline in growth, however, are notmeasurable precisely, such a notion of costs is difficult toquantify. To estimate these costs would require, in the

first instance, a clear idea of the quantitative relation-ships between policy instruments and the levels andgrowth rates of absorption and output. Second, it wouldbe necessary to undertake a careful study of specificstand-by arrangements to determine the amount of ad-justment that would be required to achieve the target, andthen set the policies accordingly. Third, the effects of agiven Fund program would have to be compared with theoutcomes for growth resulting from an alternative feas-ible set of policies that would achieve the same objec-tives. 10 This last point involves estimating an unobservedhypothetical outcome, which as yet has proved to be afairly intractable problem and which will be consideredlater.

Long-Term Issues

A longer-term aspect of the relation between Fundprograms and growth also needs to be considered in anysystematic analysis of the subject. Even if it was deter-mined that stabilization programs reduce output in theshort run, this deficiency could be outweighed by thelong-term benefits resulting from the adoption of suitableadjustment policies. Indeed, it is a basic premise of Fundprograms that balance of payments recovery does notconflict with economic growth when the time-horizon ofboth objectives is properly specified to be the mediumterm.

This view is based on a number of considerations.First, even if a reduction in absorption impairs growthover the short run, to the extent that a Fund programsucceeds in avoiding the drastic cut in absorption thataccompanies a complete loss of creditor support, theprogram can be said to protect the growth of the econ-omy currently and in the future. Second, the supply-side,or structural, policies in adjustment programs are in-tended to enhance the productive potential of the econ-omy by improving the allocation of resources andstimulating domestic savings and investment. If suchpolicies are successful, they diminish any inescapableimpact upon growth of measures that focus on reducingabsorption. Furthermore, by raising the capacity of thecountry to service debt in the future, these structuralpolicies allow for a higher level of sustainable capitalinflows, and thus a higher rate of economic growth in thelong run. Lastly, financial stability resulting from a suc-cessful stabilization program can have a beneficial effecton the state of confidence in the economy. Thisconfidence can encourage both domestically financedand foreign-financed investment, leading to gains in em-ployment, productivity, and output.

10This could also include comparing a Fund program with noprogram or no change in policy.

5



Empirical Analysis

Against this background, a comprehensive empiricalanalysis of the effects of Fund-supported adjustment pro-grams would have to consider at least the followingquestions: (a) What are the short-run effects of Fund-supported adjustment programs on the level or rate ofgrowth of output? (b) Are these effects larger than theyneed be to achieve the principal objectives of the pro-gram and, in particular, are they significantly larger thanthe effects that would result from an alternative set ofpolicies? (c) What are the effects of Fund-supportedadjustment programs on the long-run rate of growth ofthe economy, and how do these relate to the short-runeffects?

Unfortunately, empirical evidence is available only onthe first of these questions. The studies that can be usedto examine whether adjustment programs of the typesupported by the Fund have a short-run effect (negativeor positive) on growth can be divided into two broadcategories. First, certain studies use time-series data todetermine the effects of specific policies. These can beviewed as "model-based" studies, in that they rely onsome theoretical notion of the channels by whichchanges in individual policies are likely to affect thegrowth rate in the short run and then proceed to test theresulting relationship through the use of standardeconometric methods. Such time-series studies yield nodirect evidence on the effects of Fund-supported pro-grams per se, as they do not necessarily pertain to coun-tries that have had programs with the Fund, to the timeperiods over which programs were operative, to theactual policies adopted in the context of programs, nor tothe circumstances typically prevailing when the pro-grams were introduced. At the same time, however, theresults reported in these studies can be used to infer theeffects of programs, if these programs include the sametypes of policies. For example, if there is evidence in theempirical literature that devaluation or restrictive mone-tary policies exert an adverse impact on output in theshort run, it can be inferred that, other things beingequal, a Fund-supported program that relied exclusivelyon these measures would also initially reduce growth.

The second group of studies uses cross-country analy-sis to examine how economic growth is affected by aFund-supported program, taken as a complete package ofmeasures. Such studies have been undertaken, bothwithin the Fund and outside, to test the overall effec-tiveness of programs. These cross-country studies pro-vide more direct evidence than the time-series studies onthe question of the effect of Fund programs, since theyfocus explicitly on the experiences of countries withprograms. Furthermore, they are wider in scope, as theyevaluate the impact of different combinations of policies

rather than looking only at individual policies. On theother hand, such studies cannot determine how much agiven outcome is due to the policies adopted rather thanto other (non-program) factors. 11

Comparing the quantitative effects of Fund programswith the corresponding effects of some feasible set ofalternative policies—that is, providing answers to thesecond question—is regarded as perhaps the most appro-priate approach to judging the costs of programs.12 Forexample, if a typical Fund program includes suchpolicies as a reduction in the rate of credit expansion, alowering of the fiscal deficit, and devaluation, and if theeffects of such policies on growth are known, then theeffects of this combination of policies can be comparedwith some alternative combination that yields the sametargets for current account, overall balance of payments,and inflation. Clearly, counterfactual experiments of thistype would determine whether Fund programs are some-how more deflationary than necessary.

The "comparison of policies" approach to judging theeffects of programs, however, runs into a number ofdifficulties. This approach requires, first of all, detailedempirical knowledge of the links between the variouspolicy instruments included in a typical Fund programand the ultimate objectives for the balance of payments,inflation, and the growth rate. The state of the art ineconometric modeling has made significant advances butis as yet still very much in its infancy, particularly in thecase of developing countries. The models that are avail-able at present do not come close to capturing the fullcomplexity of the relevant economic relationships.

Even if an appropriately specified model incorporatingthe full range of measures in a Fund program was avail-able, a number of additional issues would still have to befaced. First, to the extent that expectations are alteredwhen a Fund program is adopted, the effects would notdepend solely on the magnitude of the announced policychanges and the coefficients relating policies to ultimateobjectives based on past historical experience. For exam-ple, an announced reduction in monetary growth couldlead to an immediate fall in inflation if it caused domesticresidents to revise downward their expectations of thefuture rate of inflation. Conversely, if the policy changeis not viewed as credible, inflation could persist for muchlonger than the underlying estimated relationship wouldindicate, and the adverse effects of an anti-inflation pro-gram on output and absorption would be correspondinglylarger. It is for this reason that the Fund staff tries toensure that the overall package of measures comprising aFund adjustment program will be viewed as credible byboth domestic residents and foreign creditors.

11This and other problems with the cross-country approach will bediscussed in Section IV.

12See Guitian (1981).

6


Second, the comparison of policies using model simu-lations would have to start by taking explicit account ofthe disequilibria that the country faced at the time thestabilization program was introduced. However, this isno easy task for two reasons. If the analysis is begunfrom a position of disequilibrium, it would be difficult todistinguish changes that occurred as a result of discre-tionary policy actions from those that would have oc-curred in any case, owing to the automatic processes thattend to adjust the balance of payments and the domesticinflation rate in an open economy. Moreover, the timepath of the variables during the transition period, andindeed the transition period itself, are not independent ofthe initial conditions. Consequently, it would be hard todetermine whether the behavior of a variable duringtransition is due to the policies adopted or simply to thepoint from which it started.

Third, the alternative sets of policies against which tocompare a given Fund program must be precisely de-fined. These alternatives can presumably differ in themix of policies consistent with the same objectives (moreemphasis on demand-management or on structural poli-cies) or in terms of policy instruments (controls on im-ports versus devaluation). These alternative programsmust, however, have the following features to makethem suitably comparable with Fund programs. To beginwith, the alternative policies have to be defined in thecontext of a common set of constraints, such as the stateof the international environment and the availability offoreign financing, including financing provided by theFund. It would be pointless to compare Fund programswith alternatives that assumed a lower level of foreigninterest rates and a larger volume of foreign financing.Moreover, the time frame of the adjustment would haveto be the same as that faced by the Fund. Comparingprograms that differ in terms of the period over whichadjustment is to take place would clearly be inappro-priate. Furthermore, the alternative package would haveto be acceptable to the country. Since Fund programs arethe outcome of detailed negotiations between countryauthorities and the Fund, they reflect the political feasi-bility of the proposed set of policy measures. Finally, thealternative policies would have to be put forward at thesame level of specificity as those included in Fund pro-grams. For example, if export promotion is to receivegreater weight, the exact methods for achieving thisobjective have to be specified. As mentioned above, fewif any alternative programs proposed in the literaturehave such features.

Given such problems, it is easy to see why no empir-ical studies are available that undertake the relevant com-parisons between Fund programs and alternative pro-grams. Consequently, there is no obvious way of deter-mining whether or not Fund programs are "too harsh."

Empirical Analysis

Of course, performing counterfactual experiments withexisting models is possible, but these experiments haveto be kept simple. For example, one can analyze theeffects of certain policies (credit restraint, reduced gov-ernment expenditures, devaluation) or compare the ef-fects of a package including such measures with a pack-age that also includes more supply-oriented policies.13

Because of the fairly simplistic structure of the availablemodels and the narrow range of policy measures consid-ered, however, these comparisons of policies can at bestbe only illustrative.

Little empirical evidence exists on the long-run effectsof Fund programs, and none at all on the effects ofvarious combinations of stabilization policies on eco-nomic development (question (c)). Even the informalevidence that is available is ambivalent on the relation-ship between financial stability and economic devel-opment. While many developing countries haveachieved both price stability and high growth by adopt-ing prudent financial policies, a number have alsomanaged to combine high rates of inflation with stronggrowth performance for extended periods of time. (Atthe same time, of course, many countries have simulta-neously experienced high inflation and low rates ofgrowth.) Some indirect evidence, however, can bebrought to bear from studies of the growth process indeveloping countries. For example, if the structural poli-cies contained in Fund programs raise the level of invest-ment and increase its efficiency, as they are intended todo, this will presumably result in a higher rate of growthin the long run. At the same time, investment is only onefactor in determining growth, and little is known aboutthe effects of stabilization policies on the other factors,such as the growth of the labor force and its productivity,development of skills, and technical progress.

For the reasons outlined above, any study of the quan-titative effects of Fund programs has to be severelylimited in scope. At this stage the only question on whichthere is some empirical evidence is whether policiesdesigned to reduce absorption and increase supply, in-cluding exchange rate policies, have a significant effecton the growth rate in the short run, namely question (a).Accordingly, the next three sections of this paper surveythe recent empirical literature that has a bearing on thisquestion. The time-series and cross-country studies thatprovide the empirical evidence are examined individu-ally, and the results are then combined to draw someconclusions about whether Fund programs have a sys-tematic adverse effect on growth. By considering onlythese two types of studies, the paper interprets "empiricalevidence" in a fairly narrow and specific way. In particu-lar, the present survey excludes case studies of the expe-

13See Section V.

7



rience of individual countries with stand-by programs,14 because the case studies do not in general subject thebecause of the difficulties in trying to generalize from the basic propositions to any kind of formal statistical test-evidence on a particular country at a particular time and ing. This survey will obviously not resolve the con-

troversy associated with Fund programs and economic14See, for example, the papers contained in Cline and Weintraub growth; it is meant to represent only a first Step toward

(1981) and Williamson (1983). such a resolution.

8


III Effects of Specific Policy Measures

This section examines the empirical evidence avail-able from studies that use the time-series approach indetermining the growth effects of individual policiestypically found in Fund programs. The evidence isgrouped according to types of policies, that is, policies torestrain aggregate demand, policies to stimulate aggre-gate supply, and exchange rate policies.

While all the studies considered here deal with devel-oping countries, they do not focus specifically on thesubset of program countries as such.15 In addition, thepolicies examined in the empirical literature are not al-ways identical to those pursued in the context of Fundprograms, although they are often quite similar. Finally,the time-series studies typically consider policies indi-vidually, even though, as mentioned previously, Fundprograms involve the simultaneous implementation of anumber of policy measures. It is quite likely that themeasures included in a Fund program may have effectson the rate of growth that are quite different when poli-cies are pursued jointly rather than independently.

The common methodology adopted in the time-seriesstudies is first to outline a model that relates the level orrate of growth of output to certain policy instruments(and perhaps a few other variables). This relationship isthen tested econometrically by using data for an indi-vidual country or group of countries. The effects ofpolicy changes on growth are determined either directlyfrom the values of the estimated coefficients or by per-forming simulation experiments with the estimatedmodel. This model-oriented approach has the distinctadvantage of being able to isolate the effects of a changein a particular policy, which, as will be discussed inSection IV, is not possible in cross-country studies. Atthe same time, however, it is difficult to account fully forall the complex linkages between the policy variablesand ultimate objectives, such as economic growth. Thus,the specifications that are used are often quite simplisticand can serve only as approximations to the "true" em-pirical relationships.

Policies to Restrain Demand

The two main instruments for controlling aggregatedemand are monetary (or domestic credit) policy andgovernment tax and expenditure policies.16

Monetary Policy

Virtually all Fund programs involve restrictive mone-tary policies, specifically ceilings on the rate of domesticcredit expansion, whether by the banking system as awhole or by the central bank.17 Consequently, the rela-tionship between economic growth and monetary or do-mestic credit expansion is crucial in judging the effectsof Fund programs.

Despite the attention it receives in both the theoreticaland empirical literature, the size of the effect of changesin the rate of domestic credit expansion on economicgrowth is still a matter of considerable controversy. Thesimple version of the monetary approach to the balanceof payments suggests, for example, that in the long runin a small open economy operating under a fixed ex-change rate regime, a reduction in domestic credit will becompletely offset by international reserve flows that re-store the money stock to the level desired by the public.Consequently, this policy would have no long-run effecton the level of output relative to its trend or on the rate ofgrowth of output. It is clear, however, that during theadjustment process a decline in the growth of domesticcredit may be associated with a reduction in capacityutilization and a possible rise in unemployment, sinceprices are not completely flexible downward. The es-timated size and duration of the deflationary effectcreated by a restrictive monetary policy depends on anumber of factors, such as (1) the speed with which theinitial credit restriction is offset by international reservemovements (an effect that depends on the responsivenessof the current account and the degree of capital mobility);(2) the response of domestic inflation to the excess de-mand for real money balances created by the credit

15Countries with Fund-supported programs are often included in thesamples used by the various empirical studies, however, even thoughthe use of a Fund program is not the main criterion for choice ofcountry.

16The demand-side effects of exchange rate policies are treatedseparately later in this paper.

17Credit policies may also include the placing of subceilings on theextension of credit to the government. These, however, are generallyregarded as an aspect of fiscal restraint in Fund programs.

9


III • EFFECTS OF SPECIFIC POLICY MEASURES

restraint policy; (3) the extent to which the excess de-mand for money reduces aggregate demand (which de-pends partly on the degree of excess demand for outputin the economy);18 and (4) the effect on private invest-ment of a rise in the cost, or a reduction in the availabil-ity, of credit. As these factors can interact in complexways, the net outcome is clearly an empirical question.

In order to judge the effect of a contractionary mone-tary policy (defined as a reduction in the growth of eitherdomestic credit or the supply of money) on the growth ofoutput in developing countries, some recent empiricalevidence has been assembled. Although the group ofstudies surveyed here is selective, it is nevertheless rea-sonably representative of the empirical work undertakenon this subject during the last five years or so. The onlymajor exclusion would be a number of large structuralmacroeconomic models for individual countries. Table 1summarizes the first-year effects, derived from thestudies surveyed, on the rate of growth of real output(gross domestic product or gross national product) of aonce-for-all change of 10 percentage points in the rate ofgrowth of either money or domestic credit, holding ev-erything else constant.

Most empirical results reported in Table 1 suggest thata monetary contraction does indeed tend to exert a defla-tionary effect on domestic output in the short run, al-though in a number of cases the effect appears to be quitesmall.19 The studies listed in Table 1 indicate that onaverage a 10 percentage point reduction in the growth ofmoney or domestic credit would reduce the rate ofgrowth of output by less than 1 percentage point over oneyear.20 This result suggests that even the rule of thumbproposed by Hanson (1980)—that in developing coun-tries a 10 percentage point change in the rate of growth ofthe money supply would alter output (relative to its long-run trend) in the same direction by about 1 percentagepoint— seems to be somewhat of an overestimate. By farthe largest estimated deflationary effect is found by vanWijnbergen (1982) in a study on Korea.21 VanWijnbergen's results, however, are disputed by Lipschitz(1984), who, using a different type of model for the sameeconomy, estimates that the reduction in the growth ofoutput would amount to only 0.5 percentage point. Asidefrom van Wijnbergen (1982), the quantitative estimatesacross the various studies are remarkably similar.22 The

18In general, the larger the excess aggregate demand in the econ-omy, the smaller are the effects on growth.

19It should be noted, however, that some stabilization programshave involved large initial reductions in monetary growth, so that theeffect on output of tighter monetary policy can in practice still be quitesubstantial.

20The median value of the estimates in Table 1 is 0.8 percent.21Van Wijnbergen's estimate, based on a neostructuralist model

associated with the writings of Taylor (1981) and others, is roughlyfour times as large as most of the results given in Table 1.

22The average of the estimates is 0.8 percent with a standarddeviation of 0.4 percent.

uniformity and apparent robustness of the results areinteresting, given that the studies use quite differentmodels, methods of estimation, sample periods, and geo-graphical coverage.

It is also important to point out that the effects ofmonetary restraint on growth occur only in the short run.In the studies examined, the largest estimated impacttypically takes place in the first year, and growth starts topick up fairly soon thereafter. Generally, the total effectlasts for about two to three years. Most of the empiricalmodels listed in Table 1 either assume or find that areduction in monetary growth or domestic credit expan-sion exerts no long-run effect on the rate of economicgrowth.23

Since monetary policy is likely to affect growthmainly through its impact on domestic investment, fur-ther indirect empirical evidence on the effect of changesin domestic credit on output can be deduced from studiesof investment behavior. A consensus has emerged inrecent years that, in contrast to the case in industrialcountries, one of the principal constraints on investmentin developing countries is the availability of financialresources, rather than their cost. Even when adjusted forrisk, the rates of return on capital investment in thesecountries are typically higher than real interest rates onloanable funds, which are often kept artificially low bygovernments for a variety of reasons. In such cases itwould be unusual to find investors undertaking capitalformation up to the point where the anticipated marginalproduct of capital is just equal to its service cost, as isassumed in theoretical models of investment. Indeed, theadministrative control of interest rates at low real levelslikely results in a chronic excess demand for capital, withsome investments with low rates of return perhaps re-ceiving priority over other higher-yielding investments.

In circumstances in which the amount of financing islimited and the price mechanism does not operatesmoothly as an allocative device, it is generally morerealistic to assume that the flow of private investment in adeveloping country is constrained mainly by the avail-ability of bank financing. If this is the case, domesticinterest rates will influence private investment only indi-rectly through the effect of an increase in the real returnon financial assets in stimulating a larger volume offinancial savings by the domestic private sector. Thus, anincrease in real credit to the private sector will generallyencourage private investment. Although bank loans arerelatively short term, borrowers can normally roll themover, thereby effectively lengthening the maturity of thedebt sufficiently to correspond to the investment period.Since the control of total bank credit generally representsthe main instrument of monetary policy in developingcountries, the authorities can influence the rate at which

23This means that the assumption of the long-run neutrality ofmoney is either imposed on the model or found to be satisfied.

10


Policies to Restrain Demand

Table 1. Short-Run Effect on the Growth Rate of Output of a 10 Percent Change in the Growth of the MoneySupply or Domestic Credit

StudyPolicy Variable

(Rate of Growth)Effect on Growth

of Output1

Aghevli andKhan (1980)

Barro (1979)

Blejer andFernandez (1980)

Blejer andKhan (1984)

Edwards (1983a)

Edwards (1983b)

Edwards (1983b)

Hanson (1980)

Khan andKnight (1981)

Leiderman (1984)

Lipschitz (1984)

van Wijnbergen (1982)

8 developingcountries

Brazil, Colombia,and Mexico

Mexico


5 LatinAmericancountries





Colombia andMexico

Korea

Korea

Domestic credit

Money

Domestic credit

Credit to theprivate sector

Money

Money

Domestic credit

Money

Domestic credit

Money

Money

Money

0.82

0.92

0.4

0.5

1.72

0.82

1.22

1.02

0.5

0.2

0.73

4.03

1Effect on real GDP or GNP (in percent) over one year.2Simple average of individual country effects.3Cumulated effect over four quarters.

private investors achieve their desired level of invest-ment by varying the flow of domestic credit and itsallocation between the public and private sectors.

The results obtained by three recent empirical studiesfor the estimated effects of variations in domestic crediton real private investment are shown in Table 2.

The results in Table 2 confirm the hypothesis that indeveloping countries credit extended by the banking sys-tem can have a sizable impact on real private capitalformation. Since considerable evidence now exists onthe relationship between growth and investment in devel-oping countries, this result would imply a connectionbetween domestic credit changes and growth through theeffect on private investment. Two of the three studies inTable 2, however, that is, Blejer and Khan (1984) andTun Wai and Wong (1982), use credit to the privatesector (in real terms), rather than total domestic credit, asthe policy variable. To the extent that Fund programsattempt to ensure an adequate flow of credit to the privatesector, the results in Table 2 probably overstate theeffects of changes in total domestic credit.

Fiscal Policy

Direct evidence on the relationship between changesin government spending or taxes and economic growth indeveloping countries is quite scarce, especially whencompared with evidence on the links between output anddomestic credit discussed above. In standard Keynesianmodels a reduction in government expenditure or anincrease in taxation is expected to have a multiplier effecton the level of real income, at least in the short run.While this proposition is well known, remarkably fewstudies introduce fiscal variables directly into a growthmodel for developing countries, and those that have doneso have not generally found the effect to be statisticallysignificant.24 The lack of positive results is probably a

24For a test of the hypothesis that an unanticipated change ingovernment expenditures affects real output, see Khan and Knight(1981), page 13. The empirical tests conducted, using a model thatalready included the effects of monetary policy, produced inconclu-sive (i. e. , statistically insignificant) results. No empirical studies nowavailable find a direct statistical role for taxes in the determination ofoutput in developing countries.

11

Countries



Table 2. Effect of Changes in Domestic Credit on Real Private Investment

Study

Blejer and Khan (1984)

Tun Wai and Wong (1982)

Fry (1984)

1Change in real private investment (in dollar terms) with respect to a $1.00 change (in real terms) in the policy variable.2Since the dependent variable is defined as the ratio of total real investment to real GDP, this estimate is not strictly comparable to the other two in

this table.

Countries

24 developing countries

5 developing countries

14 Asian countries

Policy Variable

Credit to the private sector

Credit to the private sector

Total domestic credit

Effect on Real PrivateInvestment1

0.21

0.36

0.072

reflection of the fact that the relation between fiscalvariables and the level of output (or the rate of capacityutilization) in developing countries is more complicatedthan basic Keynesian macroeconomic theory would sug-gest. Consequently, a more intensive investigation of therelationship between government spending and taxation,savings, investment, and the growth rate seems to beneeded.25

The effects of fiscal deficits on growth also turn out tobe difficult to establish empirically because of the link-age between fiscal policy and monetary policy, which isgenerally much tighter in developing countries than inindustrial countries. This link occurs because changes inthe money supply are, by definition, equal to changes incredit to the government, changes in credit to the privatesector, and variations in international reserves. If domes-tic financial markets are underdeveloped, so that thegovernment has to rely on bank credit for its financingneeds, there is a close correspondence between the fiscaldeficit and changes in the supply of domestic credit,unless the authorities are prepared to allow the privatesector to be crowded out of the credit markets. Anawareness of this close linkage between fiscal deficitsand money supply changes is crucial to understandingthe limitations on the use of monetary and fiscal policiesas independent policy instruments in developing coun-tries.26 As such, in models that include the rate of do-mestic credit expansion or the growth of money, empir-ical tests tend to suggest that fiscal variables have only arelatively modest independent role.

Other than through the demand side, fiscal policy caninfluence output through the effects of public sector in-vestment on private investment. In developing countries,in contrast perhaps to industrial countries, the relation-ship between public and private investment takes on agreater importance because of the larger role played bythe government in the overall process of capital forma-tion. There is, however, considerable uncertainty as to

25A discussion of the empirical literature on the effects of taxationon labor supply, savings, and investment in developing countries iscontained in Ebrill (1984).

2 6 The importance of this linkage is also reflected in the inclusion inFund programs of subceilings on credit extended to the government orthe consolidated public sector. Such subceilings serve the dual func-tion of controlling both the public sector deficit and the growth of totaldomestic credit.

whether, on balance, public sector investment raises orlowers private investment. In broad terms, public sectorinvestment can cause crowding out if it uses scarcephysical and financial resources that would otherwise beavailable to the private sector or if it produces marketableoutput that competes with private output. Furthermore,the financing of public sector investment, whetherthrough taxes, issuance of debt instruments, or inflation,can lower resources available to the private sector andthus depress private investment activity.27

Clearly, public investment to maintain or expand in-frastructure and the provision of public goods can also becomplementary to private investment. Public investmentof this type can enhance the possibilities for privateinvestment, raise the productivity of capital, stimulateprivate output by increasing the demand for inputs andancillary services, and augment overall resource avail-ability by expanding aggregate output and savings. Ulti-mately, the effect of public investment on private invest-ment will depend on the relative strength of these variouseffects, and there is no a priori reason to believe that theyare necessarily substitutes or complements.

Specific evidence on the relationship between publicsector investment and private investment is not easy toobtain, owing to the many difficulties, both conceptualand data-related, involved in modeling private invest-ment behavior in developing countries. Nevertheless, afew advances have recently been made in this direction.The studies by Sundararajan and Thakur (1980) and TunWai and Wong (1982) derive models of investment be-havior in which public sector investment enters as anindependent explanatory variable. The results for theeffect of public investment on private investment are notconclusive, however, in either of these studies.28 Blejerand Khan obtain somewhat better results when a distinc-

2 7 Thi s is not to suggest that all public investment will necessarilycrowd out the private sector. A number of public sector investments donot affect private investment. Certain investments are undertaken bythe public sector which the private sector cannot undertake for reasonsof scale or financing difficulties. In addition, a number of public sectorprojects may be partly financed through concessional foreign lendingthat would not reduce the resources available to the private sector.

2 8Sundararajan and Thakur (1980) found the coefficient of thepublic sector capital stock in the private investment equation to bestatistically insignificant in both the countries (India and Korea) theystudied. The coefficient measuring the relationship between public andprivate investment was statistically significant only in one country(Greece) of the five studied by Tun Wai and Wong (1982).

12


tion is made between infrastructural and other types ofpublic investment.29 Using various proxies to representthese two types of public investment, Blejer and Khandraw several conclusions for the countries in their sam-ple. First, a $1.00 increase in real infrastructural publicinvestment would increase real private investment byabout $0.25. Second, an equivalent increase in otherforms of public investment would reduce real privateinvestment by nearly $0.30. These results are consistentwith the hypothesis that infrastructural investment iscomplementary to private investment, while increases inother types of government investment tend to crowd outthe private sector.

The issue of whether a contractionary fiscal policytaking the form of a cut in real public sector investmentwill reduce or expand private capital formation is cer-tainly far from settled. Although the direction of theeffect may be uncertain, it is apparent that, by varyingthe level and composition of public investment, the gov-ernment can alter private investment and influence thegrowth rate of the economy over the longer term. Assuch, in the course of reducing the fiscal deficit, it wouldbe necessary for the government to weigh carefully theshort-term consequences of cuts in current spendingagainst the longer-term effects that would occur if thereductions fell more heavily on investment expenditures.

Policies to Stimulate Supply

Supply-side policies may be described generally undertwo headings: policies to improve the efficiency of re-source allocation and policies to increase the level or rateof growth of capacity output in the economy.30

Improving Resource Allocation

Unlike measures to stimulate savings and investment,policies focusing on improving the efficiency of resourceallocation—principally through the elimination of distor-tions—are designed to increase the overall level of out-put that can be produced from an economy's given stockof resources. Nevertheless, attempts to eliminate distor-tions present a number of practical difficulties. As dis-cussed in Section II, the removal of distortions mayinitially cause unemployment and in some cases mayeven reduce welfare. Furthermore, standard second-bestconsiderations suggest that, if a country has a number ofdistortions, the removal of only some of them will notnecessarily result in the desired gain in efficiency.

Generally, the main sources of distortions in prices aremonopolies and other forms of imperfect competition,public sector pricing policies and government price con-

Policies to Stimulate Supply

trols, various types of taxes and subsidies, and tariffs andquotas. By their nature, distortions tend to be bothmicroeconomic and country specific. Nevertheless, twosources of inefficiency have macroeconomic significanceand have gained importance in recent years. First, thereare inefficiencies related to energy pricing policies. Theincreases in world energy prices during the past decadewere often accompanied by attempts by country authori-ties to limit corresponding increases in domestic energyprices. By now, however, it is widely accepted thatcountries should systematically pass through higherprices of oil and petroleum products to final users; other-wise, the government would have to absorb in the budgetthe cost of any subsidies. Furthermore, a policy of sub-sidizing energy tends to slow down the shift to lessenergy-intensive production techniques and patterns ofconsumption. Thus, the experience of energy pricemovements illustrates the adverse effects on resourceallocation that can result from failure to set domesticprices at their international opportunity cost.

A second source of inefficiency, particularly in pri-mary producing countries, derives from distortions asso-ciated with agricultural pricing policies, which oftencause the prices of agricultural commodities to deviatefrom prices in competitive markets. Such policies have astrong impact on the level and allocation of agriculturalproduction, because they act as a tax on output andexports. It is possible to get an idea of the effect ofraising prices on agricultural supply from estimated priceelasticities of supply of selected primary commodities.Such estimates have been provided by UNCTAD (1974)and more recently for the sub-Saharan African countriesby Bond (1983). These estimates, shown in Table 3,indicate that the price responsiveness of primary com-modities is not as small as might have been thought.Indeed, there are some commodities, such as cocoa,coffee, and rubber, where the long-run price elasticity issurprisingly high. By and large, these results suggest thatpricing policies to increase the return to producers wouldtend to stimulate the output of major agricultural com-modities, particularly in the longer term.

Table 3. Estimates of Long-Run Price Elasticities ofSupply of Selected Primary Commodities

Commodity UNCTAD (1974) Bond (1983)1

29The results are based on a pooled sample of 24 developingcountries for the period 1971-79.

30This section is based largely on the discussion contained in Khanand Knight (1982).

CocoaCoffeeCerealsGroundnutsPalm kernelsPalm oilCottonSisalRubberTobaccoTimber

0.200.200.40—

0.300.300.400.500.30—

0.50

0.791.33-0.520.250.490.500.620.990.65-

1Median value of estimates reported in Table 2, page 710.

13



Increasing Capacity Output

The rate at which the aggregate potential supply ofoutput can be expanded depends, among other things, ondecisions about the proportion of current real output to beinvested in productive capital rather than consumed, aswell as on the nature and quality of the capital stockbeing added. Once the decision to increase the econo-my's total potential output is taken, the appropriate sup-ply-side policies are those that favor private saving and(if private sector investment is being emphasized) thosethat increase the attractiveness of private capital forma-tion. Furthermore, when the authorities choose to expandproductive capacity within the public sector, it is neces-sary to ensure that the policies they adopt do not in turncreate disincentives that induce a fall in private sectorfixed capital formation.

Policies to foster the expansion of savings in programssupported by the Fund focus primarily on increasing thereturn on savings. A large number of developing coun-tries impose ceilings and other restrictions on the no-minal interest rates offered on savings deposits by thebanking system. Under inflationary conditions, theseceilings may imply low or negative real interest rates onfinancial savings. If financial savings are interest sensi-tive, adjusting interest rates to more realistic levelswould clearly be called for; this adjustment would pre-sumably stimulate flows of both domestic and foreignsavings. The increase in savings would raise domesticinvestment and thereby capacity growth. As such, theeffectiveness of Fund policies in raising capacity outputwould depend on (i) the degree of interest sensitivity ofsavings and (ii) the effect of increased investment on thegrowth rate.

(i) Effect of interest rate policies on savings

Despite the amount of research expended on the inter-est responsiveness of savings in general, and in devel-oping countries in particular, it is still uncertain whetheran increase in interest rates will, on balance, raise thesavings rate.31 Empirical work on savings behavior indeveloping countries, and particularly on the relationbetween savings and interest rates, has been handicappedby severe limitations of data. For example, savings dataas a rule are calculated as a residual item in developingcountries, either by taking the difference between grossnational product (GNP) and consumption expenditure, orby subtracting the current account deficit (less net factorincome from abroad) from gross domestic investment. Ineither case, the data on aggregate savings can be subjectto substantial measurement errors. Furthermore, as men-tioned previously, since nominal interest rates in many

developing countries are regulated, they often exhibitlittle or no variation for extended periods. Suffice it tosay that these two factors have made it difficult to em-ploy standard empirical methodology in analyzing thissubject.

Recently, however, estimation of the effect of interestrate changes on savings behavior has improved mod-estly. Essentially this improvement has been the conse-quence of researchers focusing their attention on theresponse of real savings to variations in real rather thannominal interest rates. From a theoretical perspective,this is clearly a more sensible approach and has in addi-tion the practical advantage that, while nominal interestrates may be relatively constant over time, real interestrates fluctuate widely as inflation rates vary. Fry (1980,1984), for example, finds that savings rates in a numberof Asian countries are positively related to real interestrates. The most recent estimates (Fry (1984)) show thatfor a pooled sample of 14 Asian countries the coefficientmeasuring the effect of the real interest rate on savingsdeposits was between 0.05 and 0.08, depending on thespecific model in question. This would imply that a10 percent increase in the real interest rate would, otherthings being equal, raise the ratio of savings to grossnational product by a little less than 1 percent. Theseresults for Asian countries are supported by McDonald(1983), who found that real interest rates played asignificant role in the determination of real savings in 12Latin American countries. The average short-run elastic-ity of savings with respect to real interest rates was about0.2, and the corresponding long-run elasticity was of theorder of 0.3.3 2 While the dispute on the interest respon-siveness of savings is far from being settled,33 theserecent studies lend a certain amount of support to thehypothesis that savers in developing countries are likelyto modify their behavior as the (real) rate of return onsavings changes.

Aside from the issue of the responsiveness of domesticsavings, some observers have expressed doubt that theflow of savings from the rest of the world will exhibitsignificant interest elasticity. Appropriate exchange rateand interest rate policies in the context of a comprehen-sive stabilization program can influence capital inflowsby creating a climate of confidence in the economy. Theresulting increase in foreign savings, whether throughcapital inflows or remittances, can be far in excess ofwhat would be implied by estimated interest rate elas-ticities. Such evidence as is available suggests that afinancial reform involving the removal of ceilings ondomestic interest rates can indeed have a strong effect onthe capital account of the balance of payments. For

31Of course, this controversy is not exclusive to developing coun-tries; there is just as much uncertainty about the relationship in indus-trial countries.

32These values are close to Fry's (1984) estimates when the latterare converted to elasticities.

33Giovannini (1983), for example, finds little evidence on thisquestion for a group of Asian countries similar to those studied by Fry(1980, 1984).

14



example, the relaxation of interest rate ceilings in Argen-tina, Chile, Korea, and Uruguay was followed by suchlarge capital inflows that the authorities experienced con-siderable difficulty in maintaining stability in the growthof domestic liquidity.34 This phenomenon of excessiveinflow of capital, however, represents more a case forexercising caution in undertaking a financial reform thanan argument against the basic policy.

In summary, while it is evident from the empiricalstudies reviewed here that the direct response of domes-tic private savings to variations in real interest rates isweak, this does not imply that policies to raise interestrates to their market-clearing levels are irrelevant. Evenif increases in real interest rates have no immediate largeimpact on private savings, they still represent an impor-tant aspect of adjustment policies for at least three rea-sons. First, such a policy discourages domestic invest-ment projects that will not yield an adequate rate ofreturn over the longer run. Second, other things beingequal, raising domestic real interest rates tends to in-crease the supply of foreign savings, and thus total sav-ings could in fact rise. Third, eliminating distortionsimposed by interest rate ceilings tends to increase theflow of savings intermediated through the domesticfinancial system, and this improves the overall efficiencyof the savings-investment process.

Table 4. Estimates of the Effects of Increases inFactors of Production on the Growth of Real GDP

Elasticity of Growth withRespect to Changes in:

Study

Robinson (1971)

Michalopoulos and Jay (1973)

Balassa (1978)

Tyler (1981)

Blejer and Khan (1984)

1Defined as the ratio of investment2Rate of growth.

Capital1

0.17

0.24 (domestic)0.12 (foreign)

0.16 (domestic)0.24 (foreign)

0.25

0.18

to GDP.

Labor2

0.54

0.60...

0.92...

0.98

0.58

The estimated effects of a change in the ratio of invest-ment to GDP or GNP on the rate of growth of real outputare quite similar across studies, despite differences in thesamples and periods of estimation. On average, the esti-mates in Table 4 are consistent with the view that a1 percent increase in the ratio of investment to incomewould, other things being equal, raise the overall growthrate by about 0.2 percentage point.36

(ii) Investment and growth

Policies designed to raise the long-run rate of growthof per capita output in the economy must necessarilyinvolve measures to increase the rate of capital forma-tion.35 Assuming that such measures are implementedand are successful in raising investment, what would betheir impact on growth? This question can be addressedby formulating a growth model in which the rate ofgrowth of output is related to increases in the variousfactors of production—such as the capital stock (of bothdomestic and foreign origin) and the labor force, as wellas technical progress and the use of imported inputs—and then estimating the resulting model with either time-series or cross-section data. The expected positive rela-tion between economic growth and investment indeveloping countries has been documented in a numberof studies. Some of the empirical evidence on the elastic-ity of output growth with respect to capital formation andgrowth of the labor force is shown in Table 4.

34This experience essentially forms the basis for Diaz-Alejandro's(1984) argument for capital controls (on both inflows and outflows)during a stabilization program.

35Of course, together with increases in physical capital, there mustbe a corresponding improvement in the efficiency of investment. Otherfactors, such as improvements in human capital (education, skills, andhealth), increases in the labor force, and technological developments,are also important for growth.


It is worthwhile for several reasons to devote separateconsideration to the use of the exchange rate as a stabili-zation tool. First, as discussed in Section II, imbalancesthat require stabilization frequently result from the lossof international competitiveness caused by an overvaluedexchange rate. Second, since it is simultaneously anexpenditure-switching and expenditure-reducing policy,devaluation affects both domestic absorption and domes-tic supply, and thus contains elements of both demand-side and supply-side policies. Finally, it is probably fairto say that, of all policy measures recommended by theFund, devaluation has provoked the most criticism. Oneextreme criticism is that devaluation not only fails toimprove the current account of the balance of payments,but also induces stagflation in the process. Even if such apolicy is effective in changing trade flows, it is stillregarded as too costly in comparison with alternativepolicies to improve the balance of payments (Taylor(1981)).

The discussion on exchange rate policies here con-cerns itself solely with whether devaluation has a con-tractionary effect on output in the short run. No attempt

36These estimates should, however, be treated with caution. Be-cause capital endowments differ among countries, estimates of themarginal productivity of capital based on cross-section data may besomewhat misleading.

15



is made to assess either the effectiveness of devaluationin correcting external imbalances or the inflationary con-sequences of the policy. Although important, both thesefactors are not directly related to the growth issue ad-dressed here.37 Furthermore, in the studies reviewed, thepolicy of devaluation is basically considered in isolation.Since devaluation is seldom, if ever, undertaken on itsown, but is usually accompanied by a number of otherpolicies, the measured effects obtained from variousstudies are at best only suggestive of orders of mag-nitude.

The basic demand- and supply-side aspects of devalu-ation have been extensively discussed in the literature.38

Consider, for example, a situation in which excess realdomestic demand is reflected in a current account deficit.A devaluation increases the level of foreign prices meas-ured in domestic currency and thus the price of tradablegoods relative to nontraded goods in the domestic econ-omy. On the demand side, the effect of a devaluation ondomestic absorption is unambiguously negative: themain demand-side effects are a reduction in private sec-tor real wealth and expenditure, owing to the impact ofthe rise in the overall price level on the real value ofprivate sector financial assets, and on real wages andother factor incomes whose nominal values do not riseproportionately with the devaluation. For these reasons,devaluation decreases domestic demand and, looked atfrom the point of view of current absorption, appears tobe contractionary.

On the supply side, however, the effects of the devalu-ation frequently tend to move the productive sector in theopposite direction. If the prices of domestic factors ofproduction rise less than proportionately to the domestic-currency price of final output in the short run, devalua-tion will have a stimulative impact on aggregate sup-ply.39 Thus both the aggregate demand and aggregatesupply effects of a devaluation work toward reducing theexcess demand in the economy and the current accountdeficit. Whether total output rises or falls during thisprocess obviously depends on whether the contractionaryeffects on aggregate domestic demand are outweighed bythe supply-stimulating aspects of this policy. This de-pends, among other things, on the relative sizes of theprice elasticities of imports and exports and on the rela-tive shares of tradable and nontradable goods in totalproduction. In general, output will decline if the tradeelasticities are small and the structure of production isweighted more toward tradables than toward non-tradables.40

The arguments put forward to support the view thatdevaluation exerts an overall adverse effect on growthtend to rely on special assumptions that may be importantin particular developing countries but have been foundnot to be applicable in all cases. For example, Diaz-Alejandro (1965) assumes that devaluation redistributesincome to groups with a relatively low marginal propen-sity to consume and that the consequent reduction inaggregate domestic demand has a depressing effect ondomestic supply, which more than offsets the increase inthe country's exports. Other writers postulate that theaggregate supply function is backward bending in theshort run, either because distortions in the domesticcredit market cause a credit crunch or because there isoverindexation and nominal wages rise more than pro-portionately to the change in the exchange rate. Asargued by Krugman and Taylor (1978), devaluation canalso increase the domestic-currency price of importedinputs, and if the demand for them is inelastic, totalproduction would decline. Finally, Cooper (1971),Dornbusch (1981), and Hanson (1983) have shown that,as a general rule, devaluation will be contractionary ifthe elasticities of import demand and export supply arelow or if the initial trade deficit is large.41

Keeping in mind these possibilities, however, it wouldnormally be expected that, as long as devaluation suc-ceeds in altering the real exchange rate by raising productprices in domestic currency relative to factor incomes, itshould exert a stimulative effect to the extent that theshort-run marginal cost curves of the relevant industriesare upward sloping. Naturally, the longer a real ex-change rate change persists, the larger are the gains to beachieved. In addition, if the wealth and distributionaleffects of devaluation stimulate savings and investment,a long-run gain of increased potential output will also berealized.

In general, whether a devaluation exerts a net expan-sionary or contractionary effect on domestic output andemployment depends on the relative strengths of theeffect on demand and supply described above and on thetime period in question. In the short run the demandfactors may outweigh the supply factors, but this may bereversed in the medium and long term. Clearly, theseissues are relevant to the use of exchange rate policies inadjustment programs, and two types of empirical evi-dence can be brought to bear on the question. First, onecan simply ascertain whether the price elasticities ofimports and exports are in fact as low as some of thewriters referred to above assume. The empirical esti-mates reported by Khan (1974, Tables 1 and 2), for

37In the long run, as mentioned in Section II, correcting the externalimbalance may certainly set the stage for more rapid growth later. Inthe present context, however, the focus is exclusively on the short run.

38See, for example, Guitian (1976) and Dornbusch (1981).39For a discussion of the supply-side aspects of devaluation, see

Nashashibi (1980) and Khan and Knight (1982).40See Guitian (1976).

41Devaluation could also lead to a "liquidity squeeze" if domesticfirms have significant foreign liabilities whose domestic-currencyvalue rose with the devaluation. At the same time, however, thewealth of residents holding foreign assets would also increase. Theseeffects may well be offsetting in the aggregate.

16



example, indicate that the Marshall-Lerner conditions fordevaluation to be successful in improving the trade bal-ance were satisfied in 13 of the 15 developing countriesincluded in his sample. This evidence at least contradictsthe presumption that developing countries are necessarilycharacterized by low trade elasticities.

Second, certain studies look directly at this issuewithin the framework of models that explicitly take intoaccount both the demand and supply effects of devalua-tion. In Table 5, a representative set of these models isused to show the results of a 10 percent devaluation onthe rate of growth of output in the first year. The disper-sion of the estimates depends primarily on the underlyingvalues of the supply-price elasticities, and generally theresults reported in Table 5 indicate that the relative-price,or supply, effects outweigh the negative demand-sideeffects. As a result, output growth would be higher aftera devaluation rather than lower.

The main conclusion that follows from this analysis isthat the direction and magnitude of the growth effects ofexchange rate changes depend crucially on such issues asthe extent and duration of the real exchange rate change,the structure of production, and the responses of tradeflows to relative price changes. To the extent that devalu-ation affects the sectoral distribution of income, it maynot be completely costless to some sectors. On the otherhand, no strong empirical evidence supported the propo-sition that devaluation necessarily reduces the growth ofreal output even in the short term.

If devaluation is precluded as a policy measure on thegrounds of potential output costs, it is valid to ask whatalternatives could be used to restore the internationalcompetitiveness of the economy and improve the exter-nal payments position. One obvious possibility is tocompensate with other policies in the package, and if theperiod of adjustment is limited, this may involve tighterdemand management than would be necessary if devalu-ation is a feasible option. A second possibility, proposedby certain critics of the policy of devaluation in devel-oping countries, would be to impose some type of con-trols on imports and provide subsidies to exports. Asmentioned previously, controls can be successful in theshort run in certain circumstances, but there is substantialevidence now that countries that have adopted outward-looking development strategies have experienced moresatisfactory economic growth, employment, and eco-nomic efficiency.42 These outward-oriented strategieshave typically been characterized by the provision ofincentives for export production, the encouragement ofimport competition for domestically produced goods,and the use of the exchange rate and related policies tomaintain the real exchange rate at a level consistent withthe objectives of external balance and export promotion.Therefore, the use of controls to achieve balance ofpayments objectives appears much less attractive onceaccount is taken of the probable longer-term effects ongrowth and efficiency.

42See, for example, Bhagwati and Srinivasan (1979), Balassa(1980), and Krueger and others (1981).

Table 5. Effects of a 10 Percent Devaluation on the Growth of Real GDP1

Study CountryEffects on Growth

of Real GDP2

Gylfason andSchmid (1982)

Nugent and Glezakos (1982)

Khan and Knight (1982)

Branson (1985)

Taylor and Rosensweig (1984)

BrazilIndiaPakistanPhilippinesTurkey

16 Latin Americancountries


Kenya

Thailand

- 0 . 51.14.24.53.4

0.4

-0 .5

- 0 . 9 to - 1 . 4 3

1.0 to 3.34

1Over a period of one year.2In percent.3Depending on the degree of wage indexation: the lower value corresponds to zero indexation and the higher value to full indexation.4Corresponding to assumed low and high export price elasticities, respectively.

17


IV Effects of Policy Packages

An alternative empirical approach to assessing theeffects of Fund programs has been adopted in what aretermed "cross-country" studies of stand-by and extendedarrangements. These studies focus on the effectivenessof the entire policy package rather than concentrating onone or a few policies, as is done in the empirical studiesdiscussed in the previous section. Studies of this typehave been undertaken periodically within the Fund aspart of the general evaluation of experience with pro-grams,43 and recently also by writers outside the Fundlooking at various aspects of Fund-supported stabiliza-tion programs.44 This section first describes the method-ology adopted by these cross-country studies, and thensummarizes the evidence that they provide on the growthissue.

Methodology

Since it is theoretically and empirically difficult to linkall the policy measures contained in a typical Fund pro-gram to the ultimate targets of balance of payments,inflation, and growth, the cross-country approach ofassessing the overall effects appears to have certain ad-vantages. In this framework the precise nature of theunderlying economic relationships is not made explicit,and attention is directed solely at determining whetherFund programs have been "effective" in some sense inachieving the broad objectives for which they were for-mulated. While, on the face of it, this would seem to be asensible and pragmatic approach, judging the effec-tiveness of stabilization programs turns out to be muchless straightforward than it at first appears.45 In general,one would wish to determine the effectiveness of pro-grams by comparing the actual behavior of certain keymacroeconomic variables in the program country withthe outcomes that would have been observed in theabsence of the program, that is, to estimate the "counter-factual." This criterion could be further extended tocover comparisons between actual outcomes and those

43For example, Reichmann and Stillson (1978), Donovan (1981,1982), and Kelly (1982).

44See, for example, Connors (1979), Killick (1984), and Gylfason(1983).

45SeeGuitian (1981).

that would have been achieved under some alternative setof policy measures.

Unfortunately, as pointed out in Section II, criteriabased on the determination of the counterfactual involvea great deal of subjectivity and are exceedingly difficultto employ in practice.46 Consequently, the typical cross-country study compares the behavior of one or moremacroeconomic variables (the current account, the over-all balance of payments, inflation, or the growth rate ofoutput) in the period just prior to the implementation ofthe program with their behavior after the program isintroduced. This "before-after" approach, while basic-ally objective, is nevertheless inadequate as an estimatorof the independent effects of programs. Specifically, thisapproach requires the strict ceteris paribus assumptionthat all other non-program-related determinants of mac-roeconomic outcomes are unchanged, a condition that isunlikely to be satisfied in any realistic situation. It isobvious that if nonprogram factors affecting the out-comes change between the period before the programand the period during or after it, this before-after methodwill yield misleading results. A further difficulty inapplying the before-after approach is that it does notdistinguish between sustainable and unsustainablegrowth. For example, a program may be undertakenfollowing a sharp increase in aggregate domesticdemand, stimulated by an overly expansionary fiscalpolicy, that has induced an unsustainable spurt in thegrowth rate. In this case a Fund program that simplyrestored real growth to its sustainable long-run pathwould be erroneously recorded as imposing an adverse"low-growth" cost on the program country. The recogni-tion of this problem has led some writers to supplementthe before-after calculations for program countries with asimilar comparison for a reference group or "controlgroup" of nonprogram countries.47 This control groupmethod implicitly assumes that the performance of theprogram countries would have been the same as that ofthe reference group of nonprogram countries in the abs-ence of a program, so that any differences between the

46Guitian (1981), for example, refers to this type of comparison as aconjectural or judgmental standard of measurement of the effects ofprograms.

47For example, Donovan (1982) and Gylfason (1983).

18


Empirical Evidence

two are due to the program. Even though the bias inprogram estimates created by ignoring nonprogramdeterminants of macroeconomic outcomes is reduced,other potentially serious errors are introduced to theextent that program countries turn out to differ systemat-ically from the control group. Since it is likely thatcountries coming to the Fund for support are in asignificantly worse economic situation than other devel-oping countries, the problem associated with using con-trol groups is a very real one.

Such problems as these have led to considerable doubtabout the usefulness of the cross-country approach. Ofcourse, similar criticisms can be made of studies dealingwith individual policies, but in this case these problemscan be taken care of, at least in principle, by performingcontrolled experiments of alternative policy scenariosunder differing external conditions. Thus, the advantageof cross-country studies in being able to deal with theoverall package has to be set against the drawback thatsuch studies may well be unable to give a fair picture ofthe effectiveness of programs. A further disadvantage ofthe cross-country approach is its inability to distinguishbetween different types of programs (for example, thosethat put more stress on demand management than onsupply-side measures); nor can it isolate the effects ofspecific policy measures. While some studies have at-tempted to make such distinctions (Donovan (1981),Kelly (1982)), the results in fact show only the effects ofthe total package and not those of individual policies.

Empirical Evidence

Keeping in mind the above-mentioned caveats, it isnevertheless interesting to see what empirical evidence isprovided by selected cross-country studies on the rela-tionship between Fund programs and the growth rate. Itturns out that the broad conclusions of these studies aremore consistent than those of the time-series studiesreviewed earlier in this paper.

One of the first studies undertaken within the Fundusing this type of approach was that of Reichmann andStillson (1978), who examined the impact of 70 stand-byarrangements implemented over the ten-year period1963-72. Reichmann and Stillson used the before-aftercriterion by examining the growth rates of real GDP forone-year periods before and after each program and con-cluded that, on balance, Fund programs exerted no per-ceptible adverse effects on growth rates as comparedwith the period immediately preceding their implementa-tion. In certain cases, growth did decline after the incep-tion of the program relative to the previous year's rate ofgrowth, but this result was matched by a number ofinstances where the growth rate rose.

It has been argued that one-year comparisons are inap-propriate, since the effects of programs could be spreadout over a number of years. The evidence provided in the

study by Kelly (1982), although focusing on the specificissue of fiscal adjustment in Fund programs, has a bear-ing on this timing issue. Kelly compared the three-yearaverage of the rates of growth beginning with the pro-gram year with the three-year average prior to that year.For the 48 upper credit tranche stand-by arrangementsimplemented over the period 1971-79 that were exam-ined, in 25 cases a decline in the average growth rateoccurred, but in the remaining 23 there was an increase.The conclusions reached by Reichmann and Stillson(1978) apparently held up for a later period and were alsoindependent of the length of the period chosen for thecomparison.

Donovan (1981, 1982) has extended the analysis byperforming the before-after tests as well as by using acontrol-group comparison to attempt to capture the non-program effects on the rates of growth. In both studies,the control group was defined as all non-oil developingcountries (including the program countries), and the ba-sic argument was that in the absence of a program theprogram countries' experiences would have been similarto those of non-oil developing countries as a group. Inthe first of the two studies, covering 12 programs overthe period 1970-76, Donovan (1981) concluded, on thebasis of both criteria for judging effectiveness, that therewas no evidence on average of Fund programs beingassociated with any systematic downward bias in growthrates. The second of the two studies, which extended theperiod through 1980 and increased the number of stand-by and extended arrangements to 75, supported the ear-lier conclusion. Briefly, Donovan (1982) found not onlythat in the short run (one year) the growth performancesof countries with programs was similar to that of all non-oil developing countries but also, what is more impor-tant, that when three-year averages of the data wereused, the relative performance of program countries wasin fact better.

There are also problems with the specific criteria em-ployed by Donovan (1981, 1982). For example, theprogram countries and the control group are clearly notindependent when the latter includes some of the former.Furthermore, it has been argued that Fund programs canhave an effect on nonprogram countries, and thus thelatter cannot serve as a satisfactory control group. Never-theless, it is interesting to observe that both the before-after and control-group tests yielded the same qualitativeconclusions.

While the cross-country approach has becomeidentified with assessments of programs by the Funditself, a number of such studies have also been underta-ken outside the institution. For example, the OverseasDevelopment Institute (ODI) study referred to in Killick(1984) investigated the effects of 38 programs imple-mented between 1974 and 1979 and, using variants ofthe before-after tests, found that growth rates werelargely unaffected. These results confirmed the conclu-sion reached by Connors (1979) in an examination of

19


IV • EFFECTS OF POLICY PACKAGES

growth rates one year before and after the adoption ofprograms in 17 countries during 1973-77 that Fund pro-grams had neither a negative nor a positive effect ongrowth. In another study, Gylfason (1983) went beyondprevious efforts in this area and specified a model outlin-ing the transmission mechanism between the policy in-struments, principally credit policy, and the ultimatetargets. The actual statistical tests, however, whichcovered 32 stand-by arrangements during 1977-79, tookthe same form as the other studies and obtained the sameconclusion that the adverse effects on growth, if any,were small. Finally, additional cross-country evidencewith some bearing on the issue of stabilization programsand growth is contained in Harberger and Edwards(1982). These authors examined the experiences of coun-tries undertaking disinflation policies, which specificallyimplied reducing the rate of inflation from 50 percent orhigher to half that rate over a period of seven years. Forthe nine cases of substantial disinflation that theystudied, Harberger and Edwards found only limited sup-port for the notion of a trade-off between stabilization

and growth. In three cases the growth rate was lowerunder the disinflation policy, as compared with the rateexperienced during the preceding inflation. In the othersix cases, the rate of growth of real GDP turned out to behigher by an average of about 3 percentage points.

In summary, most cross-country studies show an abs-ence of overwhelming negative effects and in fact indi-cate that in a substantial number of instances growthperformance turned out to be better in the course of anadjustment program than it had been prior to its imple-mentation. There is some apparent inconsistency be-tween the results described in this section and thoseemerging from the time-series studies, particularlystudies dealing with specific demand-management mea-sures, that were examined in Section III. The latterindicate that certain policies of the type normally in-cluded in Fund programs may in fact reduce growth,while the cross-country studies show no clear pattern. Anattempt is made in the next section to reconcile thediffering conclusions reached by the two types ofstudies.

20


V Simulations of the Growth Effects of Programs

There are several reasons why cross-country and time-series studies can yield conflicting results for the effectsof Fund programs on economic growth. First, the con-flict may simply arise from differences in the methodol-ogy and empirical criteria employed by the two types ofstudies. The time-series studies, for example, are basedon formal statistical tests, while the cross-country studiesrely on a less formal and more judgmental approach todetermining the effects of programs. Comparing the re-sults across studies that employ different criteria fortesting the effects of policies can, as such, be quitedifficult. Second, the time period over which the effectson growth are measured can be an issue. While the time-series evidence in Section III is restricted to comparisonsover a one-year period, the cross-country studies oftenlook at longer (three-year) periods. This difference intime horizons for the tests could also explain part of theinconsistency that arises between the results of thesestudies.

Third (and this is perhaps the most important factor),while the time-series studies are concerned with asses-sing a single policy, and mainly a demand-oriented pol-icy, the cross-country studies examine the effects of thewhole package of policies implemented in the course of aFund program. Whereas demand-management policiesby themselves may cause the rate of growth of output todecline, other policies, particularly those stressing sup-ply-side aspects, could work toward improving thegrowth picture. Furthermore, the adoption of a Fund-supported program may alter expectations, so that theeffects on the ultimate objectives cannot be preciselydetermined a priori. Expectations are inherently non-quantifiable and thus cannot be easily incorporated intothe time-series framework. Nor do these time-seriesstudies reflect the positive growth effects of increasedinflows of capital that may result both directly and indi-rectly from the implementation of a Fund-supported ad-justment program, unless they focus explicitly on thisquestion. Since cross-country studies concentrate solelyon comparing the outcome of programs (with the histor-ical pattern or with some control group performance) andattribute all changes that occur to the program, theyautomatically incorporate the effects of all factors, in-cluding expectations and increased external finance, thathave a bearing on the outcome.

Clearly, it would be useful if the main advantages ofthe two approaches could be combined in some way. Inother words, it would be worthwhile to incorporate arange of policies into the analysis, while at the same timeabstracting from factors that influence growth but areexogenous to the program. One way to do so would be toexpand the models underlying the time-series studies tohandle a variety of both demand-side and supply-sidemeasures and to use the resulting model to performcontrolled experiments corresponding to alternative pol-icy combinations. This type of simulation exercise mighthelp to reconcile the observed inconsistency betweenresults and could also provide insight on the main issueof the growth effects of Fund programs.

For this purpose, however, it is necessary to have athand a structural model that incorporates the relationsbetween various policies and certain macroeconomicvariables, including in particular the level or rate ofgrowth of output. No single model can generally coverthe whole range of policy measures contained in a typicalFund program, although some small-scale models incor-porate certain of the major policy instruments. One suchmodel, developed by Khan and Knight (1981), satisfiesthe requirement of being able to handle several policiessimultaneously, particularly those involving the controlof aggregate demand and the exchange rate, and ofassessing their effects on the main macroeconomic vari-ables—growth, inflation, and the balance of payments.Furthermore, the parameters embedded in this model,estimated from a sample of 29 developing countries, arebroadly consistent with those obtained in the studiescited in Section III. For present purposes, a slightlyexpanded version of this model, including certain ex-plicit supply-side aspects, is used. With this model,described in the Appendix, the effects of different com-binations of policies on growth can be studied by per-forming hypothetical simulation experiments.

The simulation experiments conducted here are, ofcourse, purely illustrative and are not intended to reflectall the complexities surrounding a program. Formalmodels of any type are clearly unable to analyze allquestions relating to Fund programs, and in particularthey do not capture the complex ways in which policyvariables are related to ultimate objectives. The modelused here is highly aggregative and thus focuses only on

21


V • SIMULATIONS OF THE GROWTH EFFECTS OF PROGRAMS

what are considered the most important macroeconomicrelationships. Also, while expectations are included,they are treated in a very simple fashion. Finally, al-though the parameters of the model reflect empiricalestimates, the changes in policy and the combination ofpolicies studied here are entirely arbitrary. These variousreservations should be kept in mind in considering thesimulation results.

The simulations conducted with this model start withthe assumption that the authorities wish to achieve an(arbitrarily defined) increase in the stock of internationalreserves in a period of one year. To hit this target, it ispossible to use demand-management policies alone orsome combination of demand-side and supply-side mea-sures. Since the time horizon is restricted to one year,supply-side measures alone cannot be used, as they tendto operate with a significant lag. This lag in the effect ofsupply-side policies is built explicitly into the model.

Specifically, the simulations trace out the effects onthe growth of real GDP of the following:48

a. A set of demand-management policies, defined as aonce-for-all 10 percentage point reduction in therates of growth of nominal domestic credit andnominal government expenditures, and a 10 per-cent devaluation.49 Since prices do not adjust im-mediately in the model, these policies translate intoreal changes in the short run.

b. The above demand-management policies, com-bined with a set of supply-side policies that wouldraise the rate of growth of capacity output by0.5 percentage point a year over a period of fouryears. In the context of the model this requires anincrease in the investment-income ratio of about2-3 percentage points a year for the same four-yearperiod.50 No attempt, however, has been made tospecify exactly the measures that would producethis result. As discussed in Section III, the increasein the investment-income ratio would have to beachieved by a combination of supply-side measuresgeared to the productive structure of the countryunder consideration.51

The results of the simulations are presented in Chart 1.Assume that the economy is initially growing at somearbitrary constant rate (equal to 5 percent a year) andthat a stabilization program is introduced in the third

48The simulation experiments are similar to those contained in Khanand Knight (1981, 1982).

49While, as mentioned previously, devaluation has both demand-side and supply-side features, for expositional convenience it is treatedhere as a demand-management policy.

50This calculation is made using an aggregate growth model es-timated by Blejer and Khan (1984) for 24 developing countries. Seealso Table 4 in Section III.

51Basically, this would involve an increase in domestic financialsavings, brought about, say, through raising domestic interest rates,and an expansion in the flow of real bank credit to the private sector.

Chart 1. Effect on Growth Rate of Demand-Management and Supply-Side Policies

Rate of Growth (Percent)

period (year).52 If the policy package consisted solely ofdemand-side measures (Simulation a), the rate of growthwould decline at the beginning of the program, as thetighter credit and fiscal policies reduced aggregate de-mand. The expansionary effect of devaluation isinsufficient to offset these developments, and altogetherthe growth rate in this simulation falls by about 1.5 per-centage points in the first year, but then starts to rise asinflation declines, raising real domestic credit and realgovernment expenditures. The general improvement per-sists for about two years, and eventually the growth rateapproaches its original level. The international reservestarget is achieved with a pure demand-oriented policypackage, but at the price of a transitory drop in thegrowth rate.

The output costs can be reduced significantly if appro-priate supply-side measures are introduced simulta-neously with the demand-management package. Assum-ing that these supply-side policies raise investment andthereby the economy's trend growth rate of capacityoutput (in the present illustration by 0.5 percentage pointa year), the actual growth rate would also start to rise.53

52For simplicity, it has been assumed that the initial growth rate isconstant. The analysis would not be changed if instead the initialconditions involved a declining (or increasing) rate of growth ofoutput.

53In this model there is one-for-one relationship between actual andcapacity growth, so that any increase in the latter is matched by anequivalent increase in the actual growth rate.

22


The combined package of demand- and supply-orientedpolicies (Simulation b) would still result in a decline inthe growth rate in the first period, but now the negativeimpact of the demand-management policies is partiallyoffset by the positive effect of the increase in the growthof capacity. Since the supply-side policies are assumedto raise the growth rate permanently by increasing capac-ity growth, the overall package succeeds in putting theeconomy on a higher secular growth path.

Despite being only illustrative, these simulation exper-iments yield three particular insights into the effects ofFund programs, as well as into reconciling the conflict-ing evidence produced by the time-series and cross-country studies. First, the combined effect of severalpolicies implemented simultaneously turns out to be dif-ferent from the effect of such policies enacted individu-ally. Second, it is clear that suitable supply-side policies

Simulations of the Growth Effects of Programs

can help to offset, at least partially, any adverse short-term effects on growth that may result from demandrestraint. Since the time-series studies look mainly at thedemand-side policies, it is understandable why theyreach the conclusion that certain policies included inFund programs reduce the growth rate in the short run.Cross-country studies are arguably more ambiguous onthis issue because they implicitly incorporate supply-sidepolicies into the analysis. Third, it is clear that in judgingthe effects of programs, care has to be exercised withregard to the time period in question. In the presentexample, if one makes only one-year comparisons, theresults in Figure 1 would imply that the stabilizationprogram had significant costs, irrespective of the t^ ) s ofpolicies it contained. In contrast, if the period of com-parison were extended to, say, three years, the conclu-sion would be quite different.

23


VI Conclusions

The view that Fund-supported adjustment programsimpose significant economic costs, particularly by reduc-ing growth and employment, has been strongly voiced insome quarters. Consequently, critics of the Fund haveargued that more attention should be directed towarddeveloping alternative, less costly, approaches to stabili-zation. This paper has surveyed the available empiricalliterature in order to determine what light can be thrownon the direction and magnitude of the short-run effects ofFund programs on the level and growth of output.

In reviewing the literature, one is immediately struckby the paucity of empirical studies that directly examinethe relationship between Fund programs and economicgrowth. A lack of such studies is surprising in light of thecontroversy that has surrounded this issue. This paperput together the evidence from the few studies that havedirectly examined the effects of Fund programs and com-bined it with indirect evidence from empirical studiesthat consider the relationship between policies of the typeusually associated with Fund programs and short-runvariations in the rate of economic growth. This ap-proach, while obviously not ideal, nonetheless allowssome useful inferences to be drawn from the availablematerial.

The main patterns to emerge from the present surveycan be briefly summarized. First, the studies reviewedgenerally indicated that, while the size of the effectvaried, tighter monetary and credit policies would resultin a fall in the growth rate in the first year after they wereimplemented. Furthermore, if monetary and credit re-straint took the form of a reduction in the flow of credit tothe private sector, the empirical evidence showed thatprivate capital formation and possibly the long-run rateof growth would be adversely affected. Second, nostudies showed any clear empirical relation betweengrowth and fiscal policy. There are close institutionallinks between monetary and fiscal policies in developingcountries and thus, once monetary policy variables aretaken into account, the various studies have found itdifficult to measure the independent role of fiscal policy.Third, there is some evidence that supply-side policies,particularly policies to increase producer prices and thedomestic interest rates, have favorable effects on produc-tion and savings. For example, price elasticities of sup-

ply of agricultural commodities tend to be higher thannormally assumed, so that increases in prices encouragethe production of primary goods. The effect of variationsin real interest rates on savings is, however, quite small,implying that it would take fairly sizable increases innominal interest rates to change the savings rate. Fourth,a number of studies find a close relationship between thegrowth rate and capital formation. Therefore, policiesdirected at increasing investment and improving itsefficiency will tend to have a beneficial effect on long-run development. Finally, such empirical evidence as iscurrently available is consistent with the view that de-valuation would, on balance, exert an expansionaryrather than contractionary effect on domestic output,even in the short run. This result clearly has an importantbearing on the use of exchange rate policy in developingcountries.

One explanation of the view that Fund programs sys-tematically reduce growth is perhaps the misconceptionthat programs are designed solely to reduce aggregatedemand through the use of contractionary monetary andfiscal policies. Since some empirical evidence indicatesthat such policies slow growth temporarily, it is con-cluded that Fund programs must therefore be deflation-ary. As discussed in this survey, this interpretation of thepolicy content of Fund programs is far too narrow, andaccount has to be taken of the other growth-inducingmeasures contained in Fund programs. This aspect isbrought out clearly in the results of cross-country studiesmeasuring the effects of Fund packages that combine thewhole range of demand-management and supply-sidepolicies. These studies found that the rate of growthdeclined in a number of countries during the course of aprogram, but this result was matched by a number ofcases where the growth rate in fact rose. Once the in-fluence of all relevant policies on the growth rate isrecognized, there is no clear presumption that Fund-supported adjustment programs adversely affect growth.

In conclusion, this paper has shown the serious limita-tions of existing empirical analysis of Fund-supportedadjustment programs and economic growth. To evaluatethe criticism that Fund programs are unnecessarilydeflationary would require more systematic empirical

24


studies. Such studies would have to be in the nature of acase-by-case approach, taking the whole range of Fundpolicies into consideration rather than focusing on indi-vidual elements of programs. They would also have to besupplemented by some type of modeling and simulationanalysis so as to handle the issues that arise in comparingthe set of policies included in a Fund program with ahypothetical alternative package of measures or in com-paring the effects of a Fund program with the outcomethat would occur in the absence of a program. In thecourse of conducting such an exercise, a number ofquestions would have to be addressed. In particular, itwould be necessary to ask what was the nature and extentof the disequilibrium that led to the adoption of theprogram. Was the growth rate already starting to decline,or was it being maintained at an unsustainable level?What would the growth rate have been in the absence of

Conclusions

the financial resources directly provided and indirectlygenerated by the Fund? Is the fall in the growth rate ashort-run phenomenon, and what are the likely medium-term growth effects of a Fund-supported adjustment pro-gram? In this context it has to be recognized that a Fundprogram may lead to lower growth in the first year, butcan pave the way to a recovery in succeeding years.Would an alternative feasible set of policies that differedfrom Fund programs in either the emphasis placed oncertain instruments or in the choice of instruments, orboth, have achieved the same objectives at lesser cost?Have the judgments exercised in the setting of objectivesbeen, on average, unnecessarily severe? As this paperhas stressed, such questions are difficult and, if theultimate issue of the impact of Fund-supported adjust-ment programs on economic growth is to be resolved,further efforts to deal with these questions are needed.

25


Appendix

Description of the Simulation Model

The model used for the simulation experiments re-ported in Section V is essentially a variant of theeconometric model developed by Khan and Knight(1981, 1982). As the present version involves importantextensions, however, it is worthwhile to summarize itsbasic features. This Appendix discusses in turn thespecification of the model, the values of the parameters,and how the various policies under consideration areexpected to affect the rate of growth of output.

Structure of the Model

The complete model, reported in Table 6, contains sixbehavioral equations and five identities. While highlyaggregated and simple in structure, this type of modelhas been found to provide a fairly useful framework foranalyzing the dynamic effects of macroeconomicpolicies. Unlike its predecessor (Khan and Knight(1981)), this model explicitly considers the compositionof the balance of payments, and more important, allowscapacity output to be endogenously determined. Thislatter change is deemed essential if supply-side policiesare to be incorporated into the analysis.

The first equation in Table 6 is a standard demand formoney function, relating the desired stock of real moneybalances (md) to real income (y), the rate of interest ondeposits (r), and the expected rate of inflation (IT).5 4

This is a standard formulation with one important excep-tion. The equation reflects the assumption that in devel-oping countries there is a lack of financial assets that canbe held as an alternative to money and goods. Thevariable r is an "own" rate of interest, and the onlyopportunity cost to holding money is the rate of return onholding goods or real assets. The absence of a developedfinancial or capital market thus excludes the interestrate on alternative financial assets (bonds) fromconsideration.

The next two equations (2 and 3) describe the behaviorof imports and exports. The demand for importspecification is identical to that formulated by Khan(1974) and is the one most often used in the literature.The demand for real imports is written as a positivefunction of real income and the ratio of the price of

imports (in domestic currency terms) to the domesticprice level:

log(IM/Pm.e)dt = a4logyt - a5log(Pm.e/P)t (2a)

The actual quantity of imports is assumed to adjustproportionally to the difference between the demand forimports and actual imports in the previous period. Thispartial-adjustment model is specified as:

(2b)

4All parameters are written so as to be positive.

where B is the coefficient of adjustment, 0 < B < 1. Asis well known, this type of adjustment model introducesa distributed lag process (with geometrically decliningweights) into the behavior of real imports. Substitutingequation (2a) into (2b) and solving for the level ofnominal imports yields equation (2) in Table 6.

In contrast to the case of import demand, the equationfor exports is more problematic. Although the theory ofexport supply is still very much a contested and unre-solved subject in empirical trade work, the idea behindequation (3) in Table 6 is quite simple. The volume ofexports will increase with the productive capacity of theeconomy (represented by y*) and with the profitability ofproducing and selling exports (captured by the ratio ofexport prices to domestic prices—Px.e/P). Basically thedomestic price index (P) serves a dual role in the supplyfunction. First, for a given level of the export price (indomestic currency terms), the profitability of producingexports falls when the factor costs in the export industriesincrease. As these factor costs are likely to move with thegeneral price index, the variable P is assumed to serve asa suitable proxy for them. Second, to the extent thatresources involved in the production of exportables canbe transferred to other uses, or that the export price of agiven good can be kept different from the domestic price,the relative profitability of selling exports falls with anincrease in domestic prices. Lags are introduced into thespecification in a different way from the procedure usedin the case of imports to avoid the fact that in the partial-adjustment model the largest effect of any change in theexplanatory variables occurs in the current period. Forthe supply of exports, particularly from developing coun-tries producing primary goods, this would be an unreal-

26

Alog(IM/Pm.e)t = B[log(IM/P/m.e)dt

- log(IM/Pm.e)t-1]


Structure of the Model

Table 6. Specification of the Model1

1. Demand for moneylogmf = a j logy, + a2\ogrt - a3log7rr

2. Importslog/M, = log(/>m.e), + p [a4logyt - a5\og(Pm.dP)t]

+ (l-p,)[logMf,_, - log(/>m.€),_,]3. Exports

logX, = log(/>jc.e), + a6logy*t + a7log(Px.e/P)t + a%\og(Px.dP)t-X

+ a9\og(Px.dP)t_2

4. InflationAlogP, = axo[\og(MIP)t-x - \ogmd

t] + au(MogPmt + Aloger)+ (l-anXAlogPm,-!) + Alog€,_,)

5. Real outputAlogy, = a]2(MogDCPt - AlogP,) + ^ ( A l o g D C P , . ! - A log /V , )

- fl14log(yr-,/y*r) + tf,5(AlogG, - AlogP,)+ «16[AlogXr - Alog(/>x.€)r]

6. Capacity outputAlogy* = al7(IR/y)t + 018AlogL,

7. Money supplyAM, = ARt + ADC,

8. Balance of paymentsARt = Xt - lMt + S, + AF/P, + AF/G,

9. Domestic creditADC, = ADCP, + ADCG,

10. Domestic credit to the public sectorADCG, = Gt - Tt - AFIG,

11. Expected inflationAIT, = 7(AlogP,_, - Tr,^)

'The definitions of variables are:Endogenous

md = demand for real money balancesIM = value of imports (in domestic currency)X = value of exports (in domestic currency)AlogP = rate of inflationAlogv = rate of growth of actual outputAlogv* = rate of growth of capacity outputM = nominal stock of (broad) moneyR = net foreign assets of the consolidated

banking system (in domestic currency)DC = net domestic assets of the consolidated

banking system (domestic credit)DCG = domestic credit to the public sector7T = expected rate of inflation

Exogenous

DCP€

FIGFIPGIRLPmPxrST

— domestic credit to the private sector= exchange rate (index of units of domestic currency

per unit of foreign currency)= net external indebtedness of the public sector= net external indebtedness of the private sector= government expenditures= gross capital formation (in real terms)= labor force= price of imports= price of exports= domestic nominal interest rate= net services account= government revenues

istic assumption. Consequently, a more general three-period lag structure was introduced. This allows the lagpattern to be determined from the data.

The inflation formulation (equation 4) is taken almostdirectly from Khan and Knight (1981, 1982). The do-mestic rate of inflation (AlogP) is assumed to be pos-itively related to the excess supply of real money bal-ances and the rate of foreign inflation. The latter variableis taken to be the rate of growth of import prices(AlogfVw) adjusted by the percentage change in the ex-change rate (Aloge). It is assumed that a given change inimport prices or the exchange rate would be fully

reflected in domestic prices within two periods, and thusa restriction is imposed on the parameters of the currentand lagged values to ensure this outcome.

The rate of growth of output (Alogj)55 is specified torespond to both monetary and fiscal variables, the devia-tions of output from capacity output (the output gap), andthe rate of growth of real exports. An increase in realcredit to the private sector is assumed to have a tempo-rary positive effect on output which works through its

55Real output and real income are one and the same variable here,and the terms are used interchangeably.

27


APPENDIX • DESCRIPTION OF THE SIMULATION MODEL

effects on private expenditures. The degree to which thisoccurs is measured by the sum of the parameters on thecurrent and lagged values of the flow of real credit to theprivate sector, that is, (ai2 + an).56 While there are nofirm theoretical views on the size of (aX2 + al3), moststudies relating monetary Variables to output have foundthe effect to be small (see Table 6). The equation alsohypothesizes that, when the actual level of output isbelow its capacity level, current output wi l l tend toexpand. If the parameter <214 is equal to unity, then thedependent variable would itself become the output gap.57

The fiscal variable is represented by the growth in realgovernment expenditures, although, as was pointed outin Section III of the paper, the relationship between therate of growth of output and real government expendi-tures is complex and has not been amenable to simpleeconometric modeling. Finally, the rate of growth of realexports is introduced as a variable to be able to incorpo-rate the possible advantages of policies designed to pro-mote exports.

Unlike the earlier versions of the model, in this partic-ular variant the rate of growth of capacity output is madeendogenous. Since the rate of growth of capacity output(Alogv*) is central to analyzing supply-side policies, it isuseful to indicate how it was derived. Equation (6) inTable 6 is basically a simple growth model that startswith an aggregate production function (f) relating output(v) to the capital stock (K) and the labor force (L):

v =f(K x L) (6a)

Converting this equation into rates of growth yields:

y bK y \bL y I L(6b)

where the variable dK is defined as equal to the rate ofgross real investment {IK). A log-linear approximation toequation (4a) would be:

Alogy, = axl (IR/y)t + alsA\ogLt (6c)

with a17 = bK

and aXs tty_,L\\*L y)

The fitted values from equation (6c) then can be usedas values for the rate of growth of capacity output, as inequation (6) in Table 6. It should be noted, however, thatwhile this equation shows how capacity output is deter-mined, it does not make this variable fully endogenous.For purposes of the exercise here, it was assumed thatboth the explanatory variables, that is, (IR/y) and(AlogL), were effectively exogenous. In any realistic

56The lag pattern was determined from the data.57 With al4 = 1 the dependent variable becomes \og(y/y*)t.

setting, of course, it would be expected that the invest-ment-income ratio would itself be influenced by bothmonetary and fiscal policies. At this stage this additionalstep was not undertaken.

The remaining equations in the model are identities.The money supply (identity 7) is the balance sheet rela-tionship for the banking system, in which the changes inliabilities of both the central bank and commercial banks(broad money) are equal to the change in assets (foreignand domestic). The variable A/? is the change in netforeign assets (equal to the balance of payments) andA D C is domestic credit expansion. The balance of pay-ments (equation 8) is equal to the trade balance (X — IM),plus the services account (S), and the change in netexternal indebtedness of the private sector (AF/P) and ofthe public sector (AF/G) . Changes in domestic credit(ADC) can take place through changes in the bankingsystem's claims on the private sector (ADCP) and on thegovernment (ADCG)—equation (9). Equation (10) sim-ply links the fiscal accounts to the monetary accounts byassuming that any government deficit (G — T) can befinanced only by borrowing from the banking system(ADCG) or borrowing abroad (AF/G) , that is:

G, - Tt = ADCG, + AF/G, (10a)

where G and T are government expenditures and reve-nues, respectively. Solving (10a) for A D C G yields theequation (10) in Table 6.

Finally, expectations of inflation (TT) are assumed to begenerated by an adaptive process in which these expecta-tions are revised proportionally to the difference be-tween the actual rate of inflation in the previous period(AlogP,- / ) and the rate that was expected to prevail( I T , - / ) :

Air , = 7(AlogP,_, - i r , _ i ) (Ha)

where 7 is the coefficient of expectations, 0 ^ 7 ^ 1.Solving for TT yields equation (11). Note that in thisformulation a value of 7 equal to unity would mean theexpected rate of inflation is equal to the actual rate ofinflation in the previous period (TT, = A log /^ - i ) -

Values of Parameters Used in Simulations

The model contains 18 structural parameters and 2adjustment parameters. While in principle the completemodel could be empirically estimated, as was done forexample in Khan and Knight (1981), in the presentexercise the values of the parameters were imposed onthe system. The specific choice of values was dictated bytwo conditions. First, the parameters should be broadlyconsistent with the estimates obtained by empiricalstudies on various aspects of stabilization policies indeveloping countries.58 To allow for this, some flexibil-

58 In all cases the estimates chosen were based on cross-section datafor groups of developing countries.Thus the results are not conditionalon the parameters for any single developing country.

28


Simulation Experiments

ity had to be maintained, and thus the model was notestimated with a given set of data. Second, the combina-tion of parameters had to ensure the dynamic stability ofthe model. In other words, after a shock the model had tosettle down to a steady state, which may or may notnecessarily be the same as the original steady-state equi-librium. The values of the parameters used for the modelare shown in Table 7.

The coefficients for the money demand function arebasically taken from Khan and Knight (1981), who es-timated this function using pooled data for a group of 29developing countries. Since the Khan and Knight modeldid not contain an interest rate variable, the elasticity (a2)was arbitrarily chosen. The specific value, however, isconsistent with the estimates for savings functions ob-tained by Fry (1980) for 14 Asian countries andMcDonald (1983) for 12 Latin American countries.

The parameters in the trade equations (import demandand export supply) were estimated on a pooled sample of34 developing countries covering the period 1971-80.This estimation was necessary as there are relatively fewcross-country estimates available for the import demandequation, and practically none at all for export supply.Since the data are annual, the lags correspond to years,and it can be seen that the initial response of imports to achange in the explanatory variables is faster than in thecase of export supply. The effect of the explanatory

Table 7. Values of Parameters

Equation Parameter Value

1. Demand for money:IncomeInterest rateExpected inflation

2. Imports:AdjustmentIncomeRelative price

3. Exports:Capacity outputRelative prices

4. Inflation:Excess money demandForeign inflation

5. Real output:Real private credit (current)

(lagged)Output gapGovernment expendituresExports

6. Capacity output:CapitalLabor

7. Expected inflation:Adjustment

Ba4

a5

a9

a10a11

a12

a15a16

a17

1.200.201.20

0.400.410.20

0.130.100.300.50

0.330.27

0.060.030.900.040.05

0.180.59

1.0

variables on the latter does, however, build up over time,and by the third year the total effect turns out to be quitelarge.

The estimates for the inflation equation parameters aretaken directly from Khan and Knight (1981). The coeffi-cient a11 can be interpreted to represent the share ofimports in final expenditure and thus measures the first-round effects on domestic prices of a change in importprices or the exchange rate. The parameters in the outputequation are also mainly obtained from Khan and Knight(1981). The effect of real domestic credit to the privatesector, as given by (a12 + a13), is consistent with theestimates reported in Table 1. Both the elasticities forreal government expenditures and real exports were notfound to be statistically significant in the earlier Khanand Knight (1981) estimates, but in this exercise theywere re-incorporated into the model. The elasticities ofgrowth with respect to the investment-income ratio andthe growth in the labor force are from Blejer and Khan(1984). It can be noted that these values are broadlyconsistent with most other available estimates for suchmodels (Table 4). Finally, the coefficient of expectationswas set equal to unity, using the results reported by Khanand Knight (1981, Appendix V).

Simulation Experiments

The structural model, with the given values of theparameters, was used to perform the various simulationexperiments reported in Section V. These experimentsillustrate the effects of certain policies of the type gener-ally contained in Fund-supported adjustment programs.It may be useful to describe the channels through whichthese policies influence the growth rate or, in otherwords, the transmission process.59

Consider first the case of a reduction in the rate ofgrowth of domestic credit to the private sector that resultsin decline in total domestic credit growth (by 10 percent-age points in the specific example in Section V). Thiswill directly reduce the growth of output, although theeffect will be dampened by the fall in domestic inflationthat keeps the rate of growth of real credit from decliningas much as it would otherwise. A similar output responsewould occur if the rate of growth of nominal governmentexpenditures was lowered, except that, because of thelink between the fiscal and monetary sectors, the growthof the money supply would tend to fall. Both fiscal andmonetary policies would, other things being equal, workin concert toward lowering the rate of growth of output.

A devaluation in the context of this model has twodistinct effects. First, it creates a wealth effect throughthe increase in domestic prices. Both the growth rates ofreal credit and real government expenditures would de-

59Since the model is dynamic and involves several important feed-backs, the verbal discussion will obviously be only heuristic.

29

a1

a2

a3

a6

a7

a8

a13

a14

a18

y


APPENDIX • DESCRIPTION OF THE SIMULATION MODEL

cline as a result, and thus real output growth would fall.Second, as real exports begin to rise in response to thechange in relative prices, output is stimulated. The waythe model is set up, it would be expected that devaluationwould be contractionary in the short run as the wealtheffect initially dominates the relative-price effect. Lateron, the process is reversed and devaluation becomesexpansionary.

Finally, the effects of a supply-side policy are illus-trated by raising the investment-income ratio by 2.5 per-centage points over a period of four years. The growth ofthe labor force is assumed to be constant at 2 percent ayear. Given the relevant parameters, the increase in theinvestment-income ratio, brought about by an un-specified set of policies, raises the growth of capacityoutput by 0.5 percentage point a year over the four years.

30


References

Aghevli, Bijan B., and Mohsin S. Khan, "Credit Policy andthe Balance of Payments in Developing Countries," in W.L. Coats and D. R. Khatkhate (eds.), Money and Mone-tary Policy in Less Developed Countries (Oxford: Perga-mon, 1980), pp. 685-711.

Balassa, Bela, "Exports and Economic Growth: Further Evi-dence," Journal of Development Economics (Amster-dam), Vol. 5 (June 1978), pp. 181-89., The Process of Industrial Development and AlternativeDevelopment Strategies, Essays in International FinanceNo. 141 (Princeton, New Jersey: Princeton University,1980).

Barro, Robert J., "Money and Output in Mexico, Colombia,and Brazil," in J. Behrman and J. A. Hanson (eds.),Short-Term Macroeconomic Policy in Latin America(Cambridge, Massachusetts: National Bureau of Eco-nomic Research, 1979), pp. 177-200.

Bhagwati, Jagdish N., and T. N. Srinivasan, "Trade Policyand Development," in Rudiger Dornbusch and Jacob A.Frenkel (eds.), International Economic Policy: Theoryand Evidence (Baltimore: Johns Hopkins University,1979), pp. 1-35.

Blejer, Mario I., and Roque B. Fernandez, "The Effects ofUnanticipated Money Growth on Prices and on Outputand Its Composition in a Fixed-Exchange Rate OpenEconomy," Canadian Journal of Economics (Toronto),Vol. 13 (February 1980), pp. 82-95.

Blejer, Mario I., and Mohsin S. Khan, "Government Policyand Private Investment in Developing Countries," StaffPapers, International Monetary Fund (Washington),Vol. 31 (June 1984), pp. 379-403.

Bond, Marian E., "Agricultural Responses to Prices in Sub-Saharan African Countries," Staff Papers, InternationalMonetary Fund (Washington), Vol. 30 (December 1983),pp. 703-26.

Branson, William H., Stabilization, Stagflation, and Invest-ment Incentives: The Case of Kenya, 1979—80, WoodrowWilson School of Public and International Affairs Discus-sion Papers in Economics No. 89 (Princeton, New Jersey:Princeton University, 1985).

Cline, William R., and Sidney Weintraub, eds., EconomicStabilization in Developing Countries (Washington: TheBrookings Institution, 1981).

Connors, Thomas A., The Apparent Effects of Recent IMFStabilization Programs, International Finance DiscussionPapers, No. 135 (Washington: U. S. Board of Governorsof the Federal Reserve System, International FinanceDivision, April 1979), pp. 1-15.

Cooper, Richard N., Currency Devaluation in DevelopingCountries, Essays in International Finance No. 86(Princeton, New Jersey: Princeton University, 1971).

Crockett, Andrew D., "Stabilization Policies in DevelopingCountries: Some Policy Considerations," Staff Papers,International Monetary Fund (Washington), Vol. 28(March 1981), pp. 54-79.

Diaz-Alejandro, Carlos, Exchange Rate Devaluation in aSemi-Industrialized Economy: The Experience of Argen-tina 1955-1961 (Cambridge, Massachusetts : Massachu-setts Institute of Technology, 1965).

, "IMF Conditionality: What Kind?" PIDE Tidings(Islamabad), No. 4 (January-February 1984), pp. 7-9.

Donovan, Donal J., "Real Responses Associated with Ex-change Rate Action in Selected Upper Credit TrancheStabilization Programs," Staff Papers, International Mon-etary Fund (Washington), Vol. 28 (December 1981),pp. 698-727., "Macroeconomic Performance and Adjustment UnderFund-Supported Programs: The Experience of the Sev-enties," Staff Papers, International Monetary Fund(Washington), Vol. 29 (June 1982), pp. 171-203.

Dornbusch, Rudiger, Open Economy Macroeconomics (NewYork: Basic Books, 1981).

Ebrill, Liam P., "The Effects of Taxation on Labor Supply,Savings, and Investment in Developing Countries: A Sur-vey of the Empirical Literature" (unpublished, Washing-ton: International Monetary Fund, April 1984).

Edwards, Sebastian (1983a), "The Short-Run Relation Be-tween Growth and Inflation in Latin America: Com-ment," American Economic Review (Nashville, Tennes-see), Vol. 73 (June 1983), pp. 477-82.(1983b), "The Relation Between Money and GrowthUnder Alternative Exchange Rate Arrangements: SomeEvidence from Latin-American Countries," paper pre-sented at the Econometric Society Meetings, San Fran-cisco, December 1983.

Fry, Maxwell J., "Saving, Investment, Growth and the Cost ofFinancial Repression," World Development (Oxford),Vol. 8 (April 1980), pp. 317-27.

, "Saving, Investment, Growth and the Terms of Trade inAsia" (February 1984), unpublished, Irvine, California.

Giovannini, Alberto, "The Interest Elasticity of Savings inDeveloping Countries: The Existing Evidence," WorldDevelopment (Oxford), Vol. 11 (July 1983), pp. 601-07.

31


REFERENCES

Guitian, Manuel, "The Effects of Changes in the ExchangeRate on Output, Prices, and the Balance of Payments,"Journal of International Economics (Amsterdam), Vol. 6(February 1976), pp. 65-74.

, Fund Conditionality: Evolution of Principles and Prac-tices, IMF Pamphlet Series No. 38 (Washington: Interna-tional Monetary Fund, 1981).

Gylfason, T., "Credit Policy and Economic Activity in Devel-oping Countries: An Evaluation of Stand-By Programs,1977-79," Institute for International Economic Studies,Seminar Paper No. 268 (Stockholm: University of Stock-holm, December 1983).

, and Michael Schmid, "Do Devaluations Cause Stagfla-tion?" Institute for International Economic Studies, Semi-nar Paper No. 201 (Stockholm: University of Stockholm,April 1982).

Hanson, James A., "The Short-Run Relation Between Growthand Inflation in Latin America," American EconomicReview (Nashville, Tennessee), Vol. 70 (December1980), pp. 972-89.

, "Contractionary Devaluation, Substitution in Productionand Consumption, and the Role of the Labor Market,"Journal of International Economics (Amsterdam),Vol. 14 (February 1983), pp. 179-89.

Harberger, Arnold C , and Sebastian Edwards, "Causes ofInflation in Developing Countries: Some New Evidence,"paper presented at the American Economics AssociationMeetings, December 1982.

Kelly, Margaret R., "Fiscal Adjustment and Fund-SupportedPrograms, 1971-80," Staff Papers, International Mone-tary Fund (Washington), Vol. 29 (December 1982),pp. 561-602.

Khan, Mohsin S., "Import and Export Demand in DevelopingCountries," Staff Papers, International Monetary Fund(Washington), Vol. 21 (November 1974), pp. 678-93.

, and Malcolm D. Knight, "Stabilization Programs inDeveloping Countries: A Formal Framework," Staff Pa-pers, International Monetary Fund (Washington), Vol. 28(March 1981), pp. 1-53.

, "Some Theoretical and Empirical Issues Relating toEconomic Stabilization in Developing Countries," WorldDevelopment (Oxford), Vol. 10 (September 1982),pp. 709-30.

, "Determinants of the Current Account Balances of Non-Oil Developing Countries in the 1970s: An EmpiricalAnalysis," Staff Papers, International Monetary Fund(Washington), Vol. 30 (December 1983), pp. 819-42.

Killick, Tony, "The Impact of Fund Stabilization Programs,"Chapter 6 in T. Killick (ed.), The Quest for EconomicStabilization: The IMF and the Third World (London:Heinemann, 1984), pp. 227-69.

Killick, Tony, and others, "Towards a Real Economy Ap-proach," Chapter 7 in T. Killick (ed.), The Quest forEconomic Stabilization: The IMF and the Third World(London: Heinemann, 1984), pp. 270-320.

Krueger, Anne O., and others (eds.), Trade and Employmentin Developing Countries: Vol. I—Individual Studies(Chicago: University of Chicago, 1981).

32

Krugman, Paul, and Lance Taylor, "Contractionary Effects ofDevaluation," Journal of International Economics (Am-sterdam), Vol. 8 (August 1978), pp. 445-56.

Leiderman, Leonardo, "On the Monetary-Macro Dynamics ofColombia and Mexico," Journal of Development Eco-nomics (Amsterdam), Vol. 14 (May 1984), pp. 183-201.

Lipschitz, Leslie, "Domestic Credit and Exchange Rates inDeveloping Countries: Some Policy Experiments withKorean Data," Staff Papers, International Monetary Fund(Washington), Vol. 31 (December 1984), pp. 595-635.

McDonald, Donogh, "The Determinants of Saving Behavior inLatin America" (unpublished, Washington: InternationalMonetary Fund, April 1983).

Michalopoulos, Constantine, and Keith Jay, "Growth of Ex-ports and Income in the Developing World: A Neo-classical View," U.S. AID Discussion Paper No. 28(Washington: U.S. Agency for International Devel-opment, November 1973).

Nashashibi, Karim, "A Supply Framework for Exchange Re-form in Developing Countries: The Experience of Su-dan," Staff Papers, International Monetary Fund (Wash-ington), Vol. 27 (March 1980), pp. 24-79.

Nowzad, Bahram, The IMF and its Critics, Essays in Interna-tional Finance No. 146 (Princeton, New Jersey: PrincetonUniversity, 1981).

Nugent, Jeffrey B., and Constantine Glezakos, "PhillipsCurves in Developing Countries: The Latin AmericanCase," Economic Development and Cultural Change(Chicago), Vol. 30 (January 1982), pp. 321-34.

Reichmann, Thomas M., and Richard T. Stillson, "Experiencewith Programs of Balance of Payments Adjustment:Stand-By Arrangements in the Higher Tranches 1963-72," Staff Papers, International Monetary Fund (Wash-ington), Vol. 25 (June 1978), pp. 293-309.

Robinson, Sherman, "Sources of Growth in Less DevelopedCountries: A Cross-Section Study," Quarterly Journal ofEconomics (Cambridge, Massachusetts), Vol. 85 (August1971), pp. 391-408.

Sundararajan, V., and Subhash Thakur, "Public Investment,Crowding Out, and Growth: A Dynamic Model Appliedto India and Korea," Staff Papers, International Mone-tary Fund (Washington), Vol. 27 (December 1980),pp. 814-55.

Taylor, Lance, "IS/LM in the Tropics: Diagrammatics of theNew Structuralist Macro Critique," in W. R. Cline and S.Weintraub (eds.), Economic Stabilization in DevelopingCountries (Washington: The Brookings Institution,1981), pp. 465-502.

Taylor, Lance, and Jeffrey A. Rosensweig, "Devaluation,Capital Flows and Crowding-Out: A Computable GeneralEquilibrium Model with Portfolio Choice for Thailand"(unpublished, Washington: International Bank for Recon-struction and Development, November 1984).

Tun Wai, U., and C. Wong, "Determinants of Private Invest-ment in Developing Countries," Journal of DevelopmentStudies (London), Vol. 19 (October 1982), pp. 19-36.


References

Tyler, William G., "Growth and Export Expansion in Devel- van Wijnbergen, Sweder, "Stagflationary Effects of Monetaryoping Countries: Some Empirical Evidence," Journal of Stabilization Policies: A Quantitative Analysis of SouthDevelopment Economics (Amsterdam), Vol. 9 (August Korea," Journal of Development Economics (Amster-1981), pp. 121-30. dam), Vol. 10 (April 1982), pp. 133-69.

United Nations Conference on Trade and Development, "Sur- Williamson, John, ed., IMF Conditionality (Washington: In-vey of Commodity Demand and Supply Elasticities," stitute for International Economics, 1983).Research Memorandum No. 48 (Geneva: UNCTAD,March 1974).

33


This page intentionally left blank


Occasional Papers of the International Monetary Fund(Continued from inside front cover)

22. Interest Rate Policies in Developing Countries: A Study by the Research Department of theInternational Monetary Fund. 1983.

23. International Capital Markets: Developments and Prospects, 1983, by Richard Williams, PeterKeller, John Lipsky, and Donald Mathieson. 1983.

24. Government Employment and Pay: Some International Comparisons, by Peter S. Heller andAlan A. Tait. 1983. Revised 1984.

25. Recent Multilateral Debt Restructurings with Official and Bank Creditors, by a Staff Team

Headed by E. Brau and R.C. Williams, with P.M. Keller and M. Nowak. 1983.

26. The Fund, Commercial Banks, and Member Countries, by Paul Mentre. 1984.


28. Exchange Rate Volatility and World Trade: A Study by the Research Department of theInternational Monetary Fund. 1984.

29. Issues in the Assessment of the Exchange Rates of Industrial Countries: A Study by the ResearchDepartment of the International Monetary Fund. 1984

30. The Exchange Rate System—Lessons of the Past and Options for the Future: A Study by theResearch Department of the International Monetary Fund. 1984

31. International Capital Markets: Developments and Prospects, 1984, by Maxwell Watson, PeterKeller, and Donald Mathieson. 1984.

32. World Economic Outlook, September 1984: Revised Projections by the Staff of the InternationalMonetary Fund. 1984.

33. Foreign Private Investment in Developing Countries: A Study by the Research Department of theInternational Monetary Fund. 1985.

34. Adjustment Programs in Africa: The Recent Experience, by Justin B. Zulu and Saleh M. Nsouli.

1985.

35. The West African Monetary Union: An Analytical Review, by Rattan J. Bhatia. 1985.

36. Formulation of Exchange Rate Policies in Adjustment Programs, by a Staff Team Headed byG.G. Johnson. 1985.

37. Export Credit Cover Policies and Payments Difficulties, by Eduard H. Brau and ChanpenPuckahtikom. 1985.

38. Trade Policy Issues and Developments, by Shailendra J. Anjaria, Naheed Kirmani, and Arne B.Petersen. 1985.

39. A Case of Successful Adjustment: Korea's Experience During 1980-84, by Bijan B. Aghevli andJorge Marquez-Ruarte. 1985.

40. Recent Developments in External Debt Restructuring, by K. Burke Dillon, C. Maxwell Watson,G. Russell Kincaid, and Chanpen Puckahtikom. 1985.

41. Fund-Supported Adjustment Programs and Economic Growth, by Mohsin S. Khan and MalcolmD. Knight. 1985.

International Monetary Fund, Washington, D.C. 20431, U.S.A.Telephone number 202 623-7430

Cable address: Interfund


occasional papers of the international monetary fund · 2019-12-11 · occasional papers of the...

Documents