consolidation and market structure in emerging

8/12/2019 Consolidation and Market Structure in Emerging

1/21

Emerging Markets Review 5(

2004)

3959

1566-0141/04/$ - see front matter 2003 Elsevier B.V. All rights reserved.doi:10.1016/j.ememar.2003.12.002

Consolidation and market structure in emergingmarket banking systems

R.G. Gelos*, Jorge Roldos1

International Monetary Fund, 700 19th St., NW, Washington, DC 20431, USA

Received 1 June 2003; received in revised form 1 December 2003; accepted 1 December 2003

Abstract

This paper examines the evolution of market structure in emerging markets bankingsystems during the 1990s. While a significant process of bank consolidation has been taking

place in these countries, reflected in a sharp decline in the number of banks, this process hasnot systematically been associated with increased concentration as measured by standardindices. Moreover, econometric estimates based on the Panzar and Rosse(1987)methodologysuggest that overall, markets have not become less competitive in a sample of eight Europeanand Latin American countries. Lowering barriers to entry, such as allowing increasedparticipation of foreign banks, appears to have prevented a decline in competitive pressuresassociated with consolidation. 2003 Elsevier B.V. All rights reserved.

JEL classifications: G21; L13; L51

Keywords: Banking; Market structure; Competition; Emerging markets; Panzar and Rosse methodolo-gy; Contestability

1. Introduction

The financial services industry has been subject to dramatic changes over thepast decades as a result of advances in information technology, deregulation and

*Corresponding author..E-mail addresses:[email protected] (R.G. Gelos), [email protected] (J. Roldos).

The views expressed in this paper are those of the authors and do not necessarily represent those

1

of the IMF. The authors wish to thank Silvia Iorgova for research assistance, Susan Collins, GiovanniDellAriccia, Linda Goldberg, and participants at the May 2002 conference Financial Globalization, ABlessing or a Curse? for comments, and Luc Laeven for providing data.


2/21

40 R.G. Gelos, J. Roldos / Emerging Markets Review 5 (2004) 3959

globalization. This has reduced margins in traditional banking activities, leadingbanks to merge with other banks as well as with non-bank financial institutions,both at home and abroad. The ongoing process of consolidation has raised a number

of positive and normative issues for both mature and emerging banking systems(see, for instance, Group of Ten, 2001; International Monetary Fund, 2001); oneimportant question is whether consolidation has reduced competition. In this paper,we review the main characteristics of the consolidation process in the majoremerging market banking systems and we study its impact in the market structureof the industry.

The forces driving the consolidation process are similar in both mature andemerging markets, but the latter show some distinguishing features. First, whilecross-border mergers and acquisitions are the exception in the mature markets, they

account for a large share of the consolidation activity in emerging markets. Second,while consolidation in the mature markets has served to eliminate excess capacitymore efficiently than bankruptcy or other means of exit, in emerging marketsconsolidation has often been a way of dealing with problems stemming fromfinancial crises. Third, the authorities played a major role in the consolidationprocess in emerging markets, whereas the role of markets forces was more dominantin the mature markets.

While a number of studies have examined the effects of bank consolidation oncompetitive conditions in mature markets, hardly any systematic research has beencarried out for emerging economies. We attempt to fill this gap. First, we discuss

the main forces shaping bank consolidation in major emerging markets and describethe patterns of consolidation and concentration using traditional indicators of marketstructure. Then, we employ the method developed by Panzar and Rosse (1987) toassess changes in the competitive structure in these markets following the consoli-dation process of the second half of the 1990s. This approach, which is based onthe relationship between revenue and marginal costs, has typically been applied tocross-sectional data from developed countries. By contrast, the panel data approach2

followed here allows for assessing changes in market structure over time, in additionto providing more reliable estimates.

We find that while the number of banks has fallen in all the emerging markets

covered in this study during the period 19942000, this decline has not systemati-cally resulted in an increase in concentration. In central Europe, for instance, areduction of the number of banks has been associated with lower concentrationasmeasured by the share in total deposits of the largest banks and by HirshmanHerfindahl (HH) indicessince the large, former state-owned savings banks havebeen losing market share to the more dynamic medium-sized banks. In most of theAsian crises countries, government-led restructuring processes have led to a drop inthe number of banks, but the degree of concentration also fellwith the exceptionof Malaysia. The process of bank consolidation is more advanced in Latin America,as a result of the earlier occurrence of crises and foreign bank entry; in this region,

Two exceptions are Bikker and Groeneveld (2000) and De Bandt and Davis (2000), who examine2

the competitive conditions in European markets using data from various countries.


3/21

41R.G. Gelos, J. Roldos / Emerging Markets Review 5 (2004) 3959

the reduction in the number of banks has been accompanied with a clear increasein the level of concentration in the industry.

There is no indication of a broad decline in competition intensity. In accordance

with the results found in the literature for most of the mature markets, ourestimations indicate that emerging banking markets are characterized by monopolisticcompetition. Moreover, our results suggest that lowering barriers to entry andallowing for increased participation of foreigners in domestic banking markets hasto some extent offset any adverse effects on competition intensity brought about byconsolidation in the banking sector. For example, there is a positive correlation3

between the index of competition intensity obtained through the Panzar-Rossemethodology and indices of foreign bank participation. However, the process is stillevolving, and particularly in central Europe, it may still be too early for a definitiveassessment.

Our findings are in line with some, albeit not all, studies for mature markets. Partof the available evidence for Europe and the US supports the notion that consoli-dation and the resulting drop in the number of banks allows banks to exercise moremarket power, but the results are generally mixed. Moreover, the relationship4

between market concentration and market power seems to have weakened recentlyin advanced economies; banking markets appear to have become more competitivedue to deregulation, globalization and technological progress (see Berger et al.,1999; Canoy et al., 2001; Hannan, 1997; Radecki, 1998). For Germany, Hempell(2002) reports a similar result as presented here: despite a drop in the number of

banks during 19931998, no significant change in the competitive behavior ofbanks can be detected.

2. Patterns of banking system consolidation in emerging markets

The main forces encouraging consolidation in mature market banking systemsnamely globalization, advances in information technology, and deregulationaswell as those discouraging itlack of information and transparency, cross-countrydifferences in regulatory frameworks, ownership structures, and culturesare alsoat work in emerging markets. However, the relative importance of these factors5

The extent to which intense competition in the banking sector is desirable, is, of course, subject of3

considerable debate. See, for example, Bonaccorsi di Prati and DellAriccia (forthcoming), for asummary of the literature on the effects of competition on lending behavior and Matutes and Vives(1996), or Cordella and Yeyati (2002) for examples of studies of the link between bank competitionand financial fragility. We do not address this normative question in this paper.

De Bonis and Ferrando (1997) and Egli and Rime (2000), for example, find a positive relationship4

between bank concentration and loan interest rates in Italy and Switzerland, respectively. However,Fuentes and Sastre (1998), using the dispersion of interest rates as a proxy for competition, show thatconsolidation did not negatively affect competition among banks. Using U.S. data, Berger(1995)arguesthat once efficiency and other effects are controlled for, concentration is negatively related to profitability.

Berger et al. (1999), Canoy et al. (2001) and Group of Ten (2001) provide comprehensive surveys onthis issue.See Group of Ten (2001) for a survey of the main causes of consolidation in industrialized5

countries.


4/21


varies across countries, yielding patterns of consolidation that differ from those inthe mature markets. The main features of the consolidation process in emergingmarkets, including the role of government-led restructuring and foreign bank entry,

are discussed in this section.The most notable difference between the consolidation process in mature vs.

emerging markets is the overwhelming cross-border nature of mergers and acquisi-tions (M&As) in the latter. In particular, as noted in the Group of Ten (2001)6

study, cross-border merger activity in continental Europe and also between US andEuropean institutions has been more the exception rather than the rule. In contrast,the staggering increase in foreign ownership of emerging market banks has continuedunabated with foreign institutions controlling more than half the banking systemassets in Argentina, Chile, Czech Republic, Hungary, Poland, and Mexico. In severalLatin American and Central European countries, foreign banks are in the process ofintegrating previous acquisitions with some of the larger banks bought in the late1990s, but it is expected that some foreign banks, which do not reach market sharesabove 23% will exit the market in the near future. In addition, the merger ofparent banks in the mature markets is spilling over to the local banking environmentsin both regions, accelerating the consolidation process and contributing to thecreation of large dominant institutions.

Another difference is the more important role played by the authoritiesand thesmaller role played by market forcesin the financial sector consolidation processof emerging markets. In the mature markets, consolidation has been seen as a way

of eliminating excess capacity more efficiently than bankruptcy or other means ofexit, as it allows preservation of some of the preexisting franchise value of themerging firms. In emerging markets, consolidation has been predominantly a way7

of resolving problems of financial distress, with the authorities playing a major rolein that process. As a result of implicit or explicit deposit guarantees, the bankingauthorities have usually intervened in troubled institutions and then sold them backto the private sectoras whole institutions or in purchase and assumptiontransactions.

Even when consolidation was seen as a desirable outcomeduring normal timesor as a second stage of the crisis resolution processmarket forces appear to have

often failed to deliver the desired results. Ownership structures, in particular familyownership, regulatory shortcomings, and concerns about job losses remain the mainobstacle to a faster, market-driven consolidation process, except for the transitioneconomies. In most emerging markets, local banks started as family-owned institu-tions that often became parts of industrial conglomerates. This ownership structure8

The extent and consequences of foreign ownership in emerging market banking systems has been6

studied extensively; see, for instance, Clarke et al. (2001) and Mathieson and Roldos (2001), and the

references therein to regional and individual country studies.SeeBerger et al. (1999).7

Yoshitomi and Shirai (2001) note that commercial banks in Indonesia, the Republic of Korea, and8

Thailand are often owned by family businesses under family controlled conglomerates. Claessens et al.(1999) document that smaller, as well as older, corporations in Asia are family-controlled. Familycontrol is generally enhanced through pyramid structures, cross-shareholdings, and deviations from one-share-one-vote rules.


5/21


has at times combined with economic and prudential regulations to provide franchisevalue to institutions that would otherwise be taken over or liquidated. Only a fewbanks are publicly listed in emerging market-banking systems, and this makes

takeoversboth friendly and hostiledifficult to carry out.The need to restructure banking systems affected by crises in the second half of

the 1990s created difficult tradeoffs for the parties involved. In several cases, thegovernment asked acquiring institutions to minimize any negative employmentimplications, and in some instances agreements on employment freezes were reached.This may have hindered market-driven M&As initially, but it does not appear tohave been a major impediment to acquisitions in countries where a low initial levelof bank penetration ensured rapid credit growth potential in the near term.

While the process of bank consolidation in emerging markets is still ongoing, itsconsequencesas in the case of the advanced economiesare still subject to muchcontroversy. One of the main questions concerns the effects of consolidation oncompetition intensity.

2.1. What do standard indicators tell us?

A first step to examining market evolution is to assess the changes in keyindicators such as the number of banks in each country, the share of deposits of thelargest banks, and the Herfindahl Hirschman (HH) index. The HH index is astandard measure of consolidation in any industry and it is defined as the sum of

the squared deposit market shares of all the banks in the market. By construction,the HH index has an upper value of 10 000 in the case of a monopolist firm with a100% share of the market; the index tends to zero in the case of a large number offirms with very small market shares. A market with 10 firms with equal shareswould have an HH index of 1000, but an uneven distribution of market shares mayaffect the index substantially.Table 1presents these indicators for 13 major emergingmarket banking systems at end-1994 and end-2000.

The first fact to notice is that in all but one countries, the number of banks fellbetween 1994 and 2000. This fall is particularly strong in the cases of Korea,Malaysia, and Mexico and less pronounced in Central Europe. Turkey is the only

country that saw an increase in the number of banks over this period. This broaddecline in the number of institutions is consistent with the notion that bankconsolidation is proceeding at fast pace not only in advanced, but also in emergingmarkets. However, as discussed below, we cannot automatically conclude thatconcentration increased correspondingly.9

The fact that a reduction in the number of banks does not directly translate intoan increase in concentration is illustrated by the case of Asia. While the number ofbanks fell substantially in the Asian crises countries included in Table 1, the levelof concentration in their banking industries fell as well. This is visible in the decline

of the share in total of deposits held by the three largest banks in each market, asSeeCetorelli (1999) for examples of how the HH index varies with different patterns of large and9

small banks.


6/21

Table 1Number of banks and market concentration in selected emerging market banking systems1

Country Number 1994 HH index Number 20

of banks2 Share in total deposits 1994 of bank s2

Sh1994

(%) 2000

(%

Largest Largest La3 banks 10 banks 3 b

AsiaRepublic of Korea3 30 52.8 86.9 1263.6 13 43Malaysia 25 44.7 78.3 918.9 10 43Philippines 41 39.0 80.3 819.7 27 39Thailand 15 47.5 83.5 1031.7 13 41

Latin America

Argentina 206 39.1 73.1 756.9 113 39Brazil 245 49.9 78.8 1220.9 193 55Chile 37 39.5 79.1 830.4 29 39Mexico 36 48.3 80.8 1005.4 23 56Venezuela 43 43.9 78.6 979.2 42 46

Central EuropeCzech Republic 55 72.0 97.0 2101.5 42 69Hungary 40 57.9 84.7 1578.8 39 51Poland 82 52.8 86.9 1263.6 77 43Turkey 72 40.7 79.1 957.2 79 35

Source: Staff estimates based on data from Fitch IBCAs Bankscope and official data.

Analysis is based on data available as of end-2000 for the largest thirty banks in a specific country, including 1

1999 or most recent available in Fitch IBCAs Bankscope.The number of banks is based on official data provided by country authorities, the OECD, or Fitch IBCA. I2

specific country includes only domestic commercial banks.Includes the merger between Kookmin and Housing and Commercial Bank, as well as the merger between Shin3


7/21


well as by the decline in the HH indiceswith the exception of Malaysia. Theimmediate task of crisis resolution in the Republic of Korea and Malaysia led tosome consolidation, but the authorities in both countries are pushing a second stage

of reforms where consolidation plays a central role; consolidation has been slowerin Thailand and the Philippines, where the process was left more to marketforces.10

In Latin America, by contrast, consolidation has been associated with higherconcentration. The process of bank consolidation is more advanced in Latin America,as a result of the earlier occurrence of crises and foreign bank entry. The numberof banks has fallen in all of the major countries (Table 1), especially in Argentinaand Brazil, where the authorities carried through a process of guided consolidationthat dramatically reduced the number of banks. The widely praised approachesfollowed by both central banks during the 1990s involved separating troubled banksinto good and bad banks and selling the former with the aid of subsidized loans.11

Chiles banking system has undergone a gradual but steady process of consolidationthat has been accelerated recently. In Mexico, consolidation has proceeded in12

various steps, involving both large purchases by foreign banks and consolidationamong the local banks; the process has reached a very advanced stage. Moregenerally, although there was substantial government involvement in bank consoli-dation in the aftermath of crises, the latter part of the 1990s shows a relativelylarger role of market-driven transactions. The fall in the number of bankinginstitutions was associated with increased concentration throughout Latin America,

as measured by both, the share in total deposits of the largest banks and the HHIndices.Similarly as observed in Asia, in the major Central European banking systems a

reduction in the number of players has been associated with a decline in concentra-tion. However, the forces accounting for this evolution are very different for thesecountries than for Asia. First, there was the legacy from the pre-market-reform era,namely large state-owned savings banks concentrating a large share of deposits.Second, all three countries pursued liberal entry policies and a large number ofbanks entered the markets in the first half of the 1990s. Third, the state-ownedbanks suffered a sharp reduction in market share, partly as a result of clean-up

operations before their privatization to strategic (and mostly foreign) investors inthe second half of the 1990s. A consolidation trend has only recently begun to takehold in the region, beginning at approximately 2000. This trend is being driven bystronger banks being forced to absorb weaker ones to ensure continued stability, byshareholders deciding to exit the market, and by mergers of the parent companiesof foreign banks present in the region. Turkey is of course a different case. Itsbanking system was highly fragmented as of end-2000 and the HH index is thelowest in the sample of countries included in Table 1.Concentration has even fallen

For more details, see International Monetary Fund (2001) and Bank for International Settlements10

(2001).See also Peek and Rosengren (2000). A chronology of the main transactions in the countries11

included in our sample is provided in an Annex.See alsoBank for International Settlements (2001) and International Monetary Fund (2001).12


8/21


since 1994 as a result of a decline of the large state-owned banks and the rapidincrease in the number of medium- and small-sized private banks. More recently,the number of banks has again dropped substantially as a result of the resolution of

the 20002001 financial crisis.In sum, the reduction in the number of banks in the major emerging markets

does not seem to have consistently translated into an increase in market concentra-tion, as measured by traditional indicators. Nevertheless, the more important questionconcerns the extent to which competitive conditions have been affected by consoli-dation. Has bank market power increased? The view that market concentration isdirectly linked to competitive conduct is referred to as the structure-conduct-performance paradigm. In principle, however, there is no one-to-one relationship13

between market concentration and the degree of competition. For example, in theextreme case of a contestable market with no barriers to entry, even in highlyconcentrated markets, banks would not be able to exploit market power due to thethreat of potential competition.14

Some of the same forces promoting consolidation in emerging markets, such asglobalization and increased foreign bank entry, are also likely to have fosteredcompetition. Moreover, as mentioned earlier, there is evidence from mature markets15

that the relationship between market concentration and market power seems to haveweakened recently in advanced economies, and that technological progress hasreduced barriers to entry. Therefore, the next section moves beyond the largelydescriptive approach followed so far and uses an econometric method to assess

changes in competitive conditions.

3. Econometric methodology

In order to carry out a quantitative assessment of changes in market structure inthe banking sectors of these countries, we conduct a test based on reduced formrevenue functions proposed by Panzar and Rosse (1987). Panzar and Rosse showthat the sum of the elasticities of a firms revenue with respect to the firms inputprices (the so-called H statistic) can be used to identify the nature of the marketstructure in which the firm operates. In long-run competitive equilibrium, the H

statistic should be equal to one, since any increase in input prices should lead to aone-to-one increase in total revenues. This is true since those firms that cannotcover their increase in input prices will be forced to exit the market. The sameargument applies if the firm operates as a monopolist in a perfectly contestablemarket. By contrast, H will be negative if the firm operates as a monopolyanupward shift in the marginal cost curve will be associated with a reduction inrevenue as a result of the optimality condition for the monopolist. If the marketstructure is characterized by monopolistic competition, the H statistic will lie

See, for instance,Cetorelli (1999).13

SeeBaumol et al. (1982) and Tirole (1988), p.309. For a different argument on why regulation-

14

induced higher concentration could yield more intense competition, see Schargrodsky and Sturzenegger(2000).

SeeClaessens et al. (2001).15


9/21


between 0 and 1. If the elasticity of demand is constant, then there is a monotonerelationship between the mark-up over marginal costs and the Hindex.

More formally, letting R denote a revenue function of input prices w and

exogenous variables z that shift the firms revenue function:

*R wi

RsR(w,z) Hs . (1)8w Ri ii

Various assumptions need to be made to apply this framework in our context.Firstly, one needs to assume that banks can be treated as single product firms (DeBandt and Davis, 2000); consistent with the intermediation approach to banking,banks are viewed as producing intermediation services using labor, physical capital,

and financial capital (

i.e. deposits and funds borrowed from financial markets)

asinputs. Secondly, one needs to assume that higher input prices are not associated16 17

with higher quality services that generate higher revenues, since such a correlationmay bias the computed H statistic. This means, however, that if one rejects thehypothesis of a contestableycompetitive market, this bias cannot be too large(Molyneux et al., 1996). Thirdly, and possibly less innocuous assumption, given thevolatile economic environment in the economies we are studying, is that one needsto be observing banks in long-run equilibrium. As discussed below, we try toovercome this problem by using a panel data specification. Moreover, the problemmight be less severe if we are mainly interested in changes in the Hmeasure over

time. In other words, the hope is that, even if we cannot assess with certaintywhether at any point in time the market structure in the countries studies falls intoone of the three categories, we will still be able to infer the direction of change inmarket structure by testing for changes in the H values over time. Nevertheless,18

we will conduct equilibrium tests to assess the validity of this assumption.The Panzar and Rosse (1987) approach has been applied widely to the analysis

of mature banking systems. Early studies examine competitive conditions in the USand Canada (seeShaffer (1989) and Nathan and Neave (1989),respectively). Laterwork has focused mainly on European economies, where the market structure hasmostly been found to be characterized by monopolistic competition. For the US,19

De Bandt and Davis (2000) find a higher H statistics than for France, Germany,and Italy, but they still reject the null hypothesis of perfect competition for the USmarket. The authors do not find a clear trend in the competitive conditions in the

Note however, that product differentiation is allowed for in the monopolistic competition model.16

SeeFreixas and Rochet (1997).17

Further assumptions include profit maximization and normally shaped revenue and cost functions.18

Note that with constant elasticity of demand, there is a one-to-one relationship between H and theLerner index that measures the mark up above marginal cost. See Angelini and Cetorelli (1999) for anexamination of the Italian banking industryusing Lerner indices.

For studies of European countries, see Molyneux et al. (1994), Bikker and Groeneveld, (2000)19

and De Bandt and Davis (2000). Individual country studies have examined banking markets in Austria(Mooslechner and Schnitzer, 1995), Brazil (Belaisch, 2003), Bulgaria (Feyzioglu and Gelos, 2000),Colombia (Barajas et al. 2000), Italy (Coccorese, 1998), Switzerland (Rime, 1999), Germany (Lang,1997; Hempell, 2002), Japan (Molyneux et al., 1996), and Finland (Vesala, 1995).


10/21


economies studied. For Japan, Molyneux et al. (1996) cannot reject the hypothesisthat banks behaved as monopolists or short-run oligopolists in 1986 but in 1998,market structure was characterized by monopolistic competition.To derive the H

statistic, we estimate the following reduced form revenue equation:

IRln scqaln w qbln w qcln w qd oth (2)L F K

cap

where IR, interest revenue (or interest revenue divided by total assets); C constant;w unit price of labor; w unit price of funds;w unit price of capital; cap capacityL F Kindicators, such as total fixed assets; oth other factors potentially affecting interestrevenues, such as the business mix of the bank and the size of non-performing loans

This specification is similar to the ones used in other studies. Following DeBandt and Davis, we include the ratio of loans to total assets as an explanatoryvariable in order to control to some extent for differences in the banks productionfunction. Alternatively, we also estimate an equation for unscaled interest revenue,where we control for size effects on the right-hand side.

In contrast to most of the literature, we do not rely on a simple cross-sectionalestimation, but carry out a panel estimation with fixed effects, allowing thecoefficients on the unit input prices to change over time. This approach has various20

advantages. First, by including bank fixed effects, we can control for unobservedheterogeneitythis is important since the regressions are otherwise likely to suffer

from omitted variable problems. All bank-specific, non time-varying determinantsof revenues not explicitly addressed in the regression specification are captured bythe fixed effects. Second, as noted above, panel estimation allows us to obtain morereliable estimates by observing the behavior of banks over time and testing forchanges in the coefficients. This test is implemented by dividing the period 19941999 into two sub-periods and interacting the input price variables (lnw, lnw, andl flnw ) with a dummy variable that takes the value of one in the second sub-period.kIf the interaction term yields significant estimates, they indicate a structural breakin the statistical relationship between input prices and revenues, and we can ascertainif and in which direction the sum of the elasticities changed.

Depending on the country, we chose either 1997 or 1998 as the year marking thestructural break. In the case of Argentina, Brazil, Chile, and Hungary we chose1997 as the year of the structural break since these countries began the consolidationprocess somewhat earlier than the other countries, for which we let 1998 mark thebreak year. The main conclusions drawn from the estimations are not sensitive toalternative choices of break years, as discussed below.

4. Data

Data were obtained from the Fitch-IBCA Ltd Bankscope CD Rom. The samplecovers 126 banks in the case of Argentina, 189 banks for Brazil, 37 for Chile, 33

An exception isDe Bandt and Davis (2000).20


11/21


Table 2Descriptive bank statistics

Total Interest Personnel Interest Other Equity Fixed Loansy

Depositsassets revenues expensesy costs operating assets Assets(Deposits costsqLoans)

Argentina 1,233 94 0.032 45 38 1,469 38 0.53 788(3,492) (185) (0.046) (96) (58) (401) (81) (0.18) (1,842)

Brazil 4,488 725 0.096 558 208 351 87 0.37 2,059(1,670) (3,482) (0.58) (3,189) (2,494) (1,068) (355) (0.20) (8,426)

Chile 2,663 253 0.018 183 26 141 45 0.56 1,406(5,677) (365) (0.014) (328) (32) (299) (64) (0.23) (1,959)

Czech 2,758 240 0.010 172 11 167 78 0.40 2,150Rep. (4,752) (405) (0.016) (293) (47) (308) (154) (0.19) (3,587)Hungary 1,926 215 0.020 180 48 77 31 0.39 920

(5,687) (510) (0.041) (506) (111) (103) (57) (0.16) (1,556)Mexico 6,029 1,020 0.034 802 38 408 117 0.57 4,312

(1,010) (1,634) (0.12) (1,258) (198) (761) (243) (0.24) (7,039)Poland 2,000 179 0.092 119 7 220 40 0.41 1,192

(4,985) (326) (1.05) (228) (28) (893) (65) (0.17) (2,180)Turkey 2,288 514 0.031 384 62 139 55 0.37 1,666

(5,180) (1,040) (0.043) (902) (159) (267) (127) (0.20) (3,515)

Note: Figures are means (in millions of US$) for 199499. Standard errors are given in parentheses.

for the Czech Republic, 72 institutions for Mexico, 55 for Hungary, 55 for Polandand 69 in the Turkish case. However, coverage of individual variables varies over21

time. The data include both public and private banks. Table 2 provides somesummary statistics.

A feature of the Fitch IBCA database is that it does not provide a completehistorical panel of banks over time. For example, if a merger occurs, only thelargest of the merged banks is typically kept retroactively in the database. Similarly,exiting banks are deleted fully from the database. However, we believe that the biasthis feature of the database introduces is not problematic for our case. In fact, as

mentioned earlier, applying the methodology proposed by Panzar and Rossepresupposes that banks have reached their steady state, and the inclusion of newentrants or banks that went bankrupt would therefore be problematic from thatperspective.

Another possible source of bias, as De Bandt and Davis (2000) point out, stemsfrom the fact that only the more prominent banks are included in the database,excluding smaller ones which potentially might have more market power in localmarkets. This is a problem that we cannot address. It should also be noted that theFitch-IBCA database contains frequent missing observations for some variables,such as the number of bank employees.

The severe balance sheet impairment and negative revenues of banks in the Asian crisis countries21

prevented their inclusion in the regression analysis of this section.


12/21


As we do not have exact data on unit factor prices, we approximate them in thefollowing way:22

w sPersonnel expensesy(depositsqloans)Lor personnel expensesy(total assets)23

w sInterest expensesy(depositsqinterbank time and demand deposits),24Fw sOther expensesyfixed assetsKcapsLogarithm of total assetsbmsTotal loansytotal assets

5. Results

The econometric results suggest that for most countries, market structure can be

characterized by monopolistic competition, a result also observed in many maturemarkets. Table 3 shows the results from fixed-effects estimations of Eq. (2) for25

each country, the derived H statistics and corresponding tests. In most cases, theestimated coefficients have the expected signs and are statistically significant at the5% level. The H-statistics are always between zero and one, indicating eithermonopolistic competition or inconclusive results. For Argentina and Hungary, the Hstatistic is compatible with either monopolistic or perfect competition.

Over the period 19941999, market structure changed in only one of the eightemerging markets examined, with constancy of the H statistic in the other cases.For Turkey, we cannot reject a decline in the Hstatistic, suggesting that competition

has become less intense since 1998. Interestingly, while we cannot reject constancyof the H statistic for Argentina, competition seems to have increased in the lateryears of the sample. This is consistent with the results of Burdisso et al. (2001)who find that competitive conditions in Argentina during 1997 1999 were veryclose to perfect competition.

These results are largely in line with the simple statistics presented earlier. Theshare in total deposits of the largest banks and the HH indices decreased in CentralEurope, and consistent with these numbers, we do not find a decrease in the Hstatistic for these countries. For Latin America, while Brazil and Chile showedmoderate increase in the large banks shares and the HH indices, the econometricresults do not reject constancy of the Hstatistic.26

For similar definition of variables, seeDe Bandt and Davis (2000), Molyneux et al. (1994, 1996),22

Nathan and Neave (1989), Rime (1999) and Shaffer (1982).It would have been preferable to use the number of employees in the denominator. However, the23

Fitch IBCA data for this variable is very scant, and we were not able to collect comprehensive timeseries from other sources, except for the case of Poland, and, to a more limited extent, Argentina.

There was not enough information available on other liabilities such as subordinated debt or long-24

term borrowing.See, for example,De Bandt and Davis (2000).25

The results for Brazil are in line with the findings by Nakane (2001), who, using a different26

methodology, concludes that Brazilian banks do have some market power and that neither perfectcompetition nor monopoly accurately describes the competitive conditions of the Brazilian bankingmarket. Belaisch (2003) recently also finds evidence for noncompetitive market structure in Brazilianbanking.


13/21

Table 3Fixed-effects estimation of revenueEq. (2)

Argentina Brazil Chile Czech Hungary MexiRepublic

Lnwl 0.22 0.23 0.29 0.37 0.37 0.12(2.58) (4.20) (4.91) (4.81) (2.52) (2.17

Lnwf 0.42 0.36 0.46 0.26 0.37 0.33(5.25) (7.55) (10.21) (5.07) (2.73) (2.21

lnwk 0.19 0.08 0.02 y0.04 0.09 0.05(2.29) (3.52) (0.53) (y1.89) (1.45) (1.25

lnw*l 0.13 0.04 0.03 y0.06 y0.04 y0.0D(after) (2.43) (2.89) (1.50) (y1.60) (y0.93) (y2.lnw*f y0.20 y0.06 y0.03 0.23 0.12 0.13D(after) (y2.93) (y1.46) (1.06) (3.01) (1.28) (1.17lnw *k 0.20 0.03 y0.01 y0.16 y0.13 y0.0

D(after) (1.88) (1.41) (y

0.55) (y

2.73) (y

1.53) (y

0.Bm 1.32 0.60 0.25 y0.16 0.49 0.79(3.92) (3.89) (2.11) (y0.55) (1.10) (4.03

Hearly 0.84 0.66 0.76 0.59 0.83 0.50

(7.99) (10.64) (11.96) (7.06) (4.31) (4.19

Market Inc. MC MC MC Inc. (MC MCstructure (MC or or Perfectearly Perfect Comp.) Comp.)H late 0.97 0.69 0.75 0.60 0.77 0.51

(8.43) (12.38) (12.46) (7.52) (4.22) (5.24

Market Inc. MC MC MC Inc. (MC MCstructure (MC or or Perfect

late Perfect Comp.) Comp.)Test for Cannot Cannot Cannot Cannot Cannot Cannchange reject reject reject reject reject rejectin H constancy constancy constancy constancy constancy const

(Ps0.17) (Ps0.57) (Ps0.52) (Ps0.88) (Ps0.57) (Ps0 of 211 752 197 103 77 186obs.


14/21

Table 3 (Continued)

Argentina Brazil Chile Czech Hungary MexiRepublic

R (within)2 0.85 0.60 0.94 0.76 0.77 0.69Year of 1997 1997 1997 1998 1997 1998structuralbreak

Note: Dependent variable: interest incomeytotal assets. Hstatistic is the sum of the elasticities of interest rate revepanel data regressions using yearly data on individual banks for the period 19941999. Test for change in H rechanged in the period starting with (and including) the year of the structural break, at the 5% confidence level. MCthat the hypotheses H)0 and H-1 could both not be rejected at the 2.5% confidence level. Inc. (Inconclusive) iwith various types of market structure. w is defined as total personnel expenses divided by total assets. The reportedlerrors (HubertyWhite). The tstatistics reported below the Hvalues refer to tests of Hs0. Early refers to the perirefers to the period after the structural break(including the year of the structural break.


15/21


The results are robust to alternative choices of the year dividing the sample intoearly and late periods. For example, when choosing 1 year earlier as the structuralbreak date for all countries, all results on the classification of market structure in

both subperiods remain unchanged. Moreover, the results concerning tests for achange in Hremain unaltered except for the case of Mexico, where the estimationssuggest a decline in competitive pressures in the second period when using 1997 asthe dividing year.

The only countries for which the econometric results do not appear to beconsistent with the statistics presented earlier are Mexico and Turkey. Mexicoexperienced both a rise in the share of the largest banks and an increase in the HHindex, but the H statistic remains constant. Turkey saw a decline in the share oflarge banks and in the HH index, while displaying a decrease in the Hstatistic. The

financial turbulence in Turkey at the beginning and end of our sample might accountat least to some degree for this result.To assess whether we are indeed observing equilibrium behavior, we follow

Shaffer (1982) and other authors in verifying that input prices are not correlatedwith returns. Specifically, we estimate Eq. (2) with the ratio of net income to totalassets as the independent variable; if Hs0, the data reflect an equilibrium. Theresults indicate that we are indeed observing an equilibrium outcome except for thecase of the Czech Republic, Mexico, and Turkey. Therefore, the results for thesecountries need to be interpreted with caution, and the inconsistency between theHH index results and the Hstatistic comparison for Mexico and Turkey may partly

be explained by this fact.

5.1. Robustness

While individual results vary somewhat with the exact econometric specification,the main resultthe rejection of the notion of a widespread worsening in competitiveconditions across countriesalways holds.

We first follow De Bandt and Davis (2000) in estimating an unscaled revenueequation and defining the unit price of labor as personnel expenses divided by total

deposits and total loans (not shown). While this yields negative coefficients on27

the unit price of labor, the Hstatistic remains between zero and one in all but onecase (Poland) for both sub-periods. Overall, according to the estimation results ofthis specification, the confidence bands for the Hstatistics are more often compatiblewith two different forms of market structure than in the specification reported inTable 3.As previously, we are unable to reject the null hypothesis of no change inthe Hstatistic in most cases. The decline in competition intensity reported earlier isconfirmed for the case of Turkey. In contrast to the results inTable 3,but consistentwith the developments of market concentration discussed in the previous section,we also find a decline in competition for Mexico.

De Bandt and Davis (2000) argue that deposits and loans represent the most labor-intensive bank27

activities.


16/21


In order to control for time-varying, aggregate factors such as demand shocks,we also carry out estimations including time dummies (not shown). Therefore, the28

results suggest no change in competitive conditions for Argentina, Brazil, Mexico,

and Poland, in line with the results presented earlier. However, the estimations withtime effects indicate a worsening in competitive conditions in Chile and Hungary,while the result for Turkey disappears.

Finally, for two of the countries in our sample, Argentina and Poland, we areable to estimate Eq. (2) using personnel expenses divided by the number ofemployees as the proxy for unit labor costs (not shown). For Poland, while thecoefficient on the unit price of labor becomes positive, the results continue tosuggest a structure of monopolistic competition, without a change in the later years.For Argentina, data on the number of employees is available only starting 1997.

For this period, consistent with the results in Table 3,the confidence interval aroundthe H statistic does not allow us to distinguish between the market structuresmonopolistic competition or perfect competition. For Poland, theHstatistics obtainedthis way are somewhat higher than those presented in Table 3, and we cannotconclusively distinguish between monopolistic competition and perfect competitionin neither sub period. However, consistent with the baseline results, we cannot rejectconstancy of the Hstatistic over the entire sample period.

5.2. Competition and foreign bank participation

The positive correlation between the H statistics and measures of foreign bankparticipation supports the notion that foreign competition (or the threat thereof)hashelped to maintain competitive pressures. If the opening to foreign competition hasindeed attenuated any reductions in competition intensity resulting from consolida-tion, one should see a correlation between our measure of competition intensitytheHstatisticsand the degree of foreign bank participation. Here, we use measuresof foreign control for 1994 and 1999 reported in the International Monetary Fund(2000) International Capital Market Report (Table 4). The level of foreign bankcontrol increased substantially between 1994 and 1999 in all countries exceptTurkeythe only country for which we could not reject a decline in competitivepressures inTable 3. Moreover, there is a clear positive correlation (0.43) betweenthe H statistics reported in Table 3 and the measures of foreign control, as shownin Fig. 1. While one has to be careful in interpreting this correlation since thenumber of observations is low and the changes in the H statistics are mostly notsignificant, the picture is in line with findings reported by Claessens et al. (2001)that a higher number of foreign banks reduce profits and overhead expenses ofdomestic banks.29

It is not obvious, however, that including time dummies is preferable to excluding them since

28

factor costs are also likely to vary over time in the aggregate. Including time dummies may thereforemake it more difficult to isolate the effects we are interested in.

See alsoBarajas et al. (2000) for similar findings for the case of Colombia.29


17/21


Table 4Foreign bank participation

Country Foreign control Foreign control1994 1999

Argentina 17.9 48.6Brazil 8.4 16.8Chile 16.3 53.6Czech Republic 5.8 49.3Hungary 19.8 56.6Mexico 1.0 18.8Poland 2.1 52.8Turkey 2.7 1.7

Source: International Monetary Fund (2000). Foreign control denotes the ratio of assets of banks

where foreigners own more than 50% of total equity to total bank assets.

Fig. 1. Foreign control and Hstatistic. Note: H statistics are those reported for the two sub-periodsreported in Table 4. The H statistics for the early sub-periods are plotted against the foreign controlshares for 1994; the Hstatistics for the later sub-periods are plotted against the foreign control sharesfor 1999.

6. Conclusions

The process of consolidation in emerging market banking systems has, to a largeextent, not yet translated into a decline in competitive pressures. To some degree,this may be due to the fact that the process is yet in its infancy in some of thecountries analyzed here, particularly in Central Europe and Turkey. Nevertheless,the results for two countries in which the consolidation process is more advanced,namely Argentina and Mexico, shows that the ultimate effect on competition


18/21


intensity is by no means obvious. We do find support for the view that foreign bankcompetition has attenuated any decline in competition intensity stemming fromconsolidation.

However, going forward one should not conclude that the potential for increasesin market power is absent even if consolidation proceeds further in the emergingmarkets. In some instances, local authorities have already acted on preventing thecreation of institutions that were perceived as too dominant, probably with goodreasons. For example, the potential merger between the two largest Mexican bankswhich would have created an institution with control over 40% of the systemsdepositsraised concerns among the regulatory authorities and finally did not comethrough. Similarly, the control of the two largest Chilean banks (commanding 27%of deposits)by Spains BSCH, prompted the authorities to modify the Banking Lawand require authorization when a merger leads to the creation of an institution thatcontrols more than 20% of deposits.30

An important, related issue, is whether consolidation and increased foreigncompetition could restrict access to credit for smaller enterpriseseven if it doesnot lead to less overall competition. Large foreign banks are often seen as shyingaway from lending to smaller and non-transparent firms. Hence, emerging markets31

may have to strengthen efforts to adopt international accounting standards andincrease transparency in the corporate sector, to ensure that the process of consoli-dation with increased foreign bank competition trickles down the credit spectrum tosmaller enterprises as well.

Further examination of how consolidation and foreign bank participation ischanging the financial services landscape in emerging markets, including theimplications for prudential regulation, competition regulation, and macroeconomicpolicies, remains an important area of research.

Appendix A:

A.1. Chronology of major transactions in our sample of banking systems

This chronology is based on reports from Fitch IBCA and Standard and Poors,as well as on Guillen and Tschoegl (1999) and several issues of The Banker and

Euromoney.

A.1.1. Argentina

1996: BBVA acquires Banco Frances; Banco Central Hispano acquires BancoTornquist; Bank of Nova Scotia acquires Banco Quilmes; Credit Agricole acquiresBanco Bisel and HSBC buys Banco Roberts.

1997: BBVA acquires Banco de Credito Argentino and Banco Santander buysBanco Rio de la Plata;

1998: Banco Central Hispano acquires a 10% stake in Banco Galicia y BuenosAires; Citibank buys Banco Mayo.

See IMF (2001) for a more thorough discussion of policy issues associated to the consolidation30

process in emerging markets financial systems.SeeBerger et al. (2001) and Clarke et al. (2001).31


19/21


1999: Banco Santander acquires Banco de Rio Tercero and BBVA buysCorpBanca.

A.1.2. Brazil1995: Unibanco acquires Banco Nacional1996: Banco Interatlantico is merged with Banco Boavista; HSBC acquires

Bamerindus.1997: Banco Santander acquires Banco Noroeste and Banco Geral do Comercio;1998: BBVA acquires Banco Excel Economico; ABN Amro acquires Banco Real

and Banco do Estado de Pernambuco; Caixa General buys Banco Bandeirantes;Wachovia Bank acquires Banco Portugues do Atlantico-Brasil.

1999: Banco Bradesco acquires Baneb (Banco do Estado de Bahia).

A.1.3. Chile

1996: Banco Santander acquires Banco Osorno y La Union; HSBC buys a 7%stake in Banco Santiago; Banco Santander-Chile merged with Banco Osorno;

1997: Banco Santiago merged with Banco OHiggins.1998: BBVA acquires Banco Hipotecario de Fomento.

A.1.4. Mexico

1996: Citibank acquires Banco Confia; BBVA acquires Banco Oriente and BancaCremi; HSBC buys a 20% stake in Banco Serfin; Bank of Montreal buys a 16%

stake in GFBancomer;1997: Banco Santander acquires Grupo Financiero InverMexico;1999: FOBAPROA intervenes Banco Serfin.2000: Santander Central Hispano buys Grupo Financiero Serfin; BBVA acquires

Bancomer.

A.1.5. Czech Republic

1998: IPB sold to Nomura Bank; GE Capital acquires Agrobanka.1999: Belgiums KBC acquires Ceskoslovenska Obchodni Banka (CSOB);2000: Erste Bank acquires Ceska Sporitelna, the main Czech retail bank.

A.1.6. Hungary

1995: GE Capital and EBRD acquire Budapest Bank;1996: ABN Amro acquires Hungarian Credit Bank;1997: DG Bank purchases Savings Bank; Kredietbank and Irish Life acquire

K&H Bank; Erste Bank buys Mezobank; OTP, the largest reatil bank was soldthrough private placement to Hungarian and foreign institutional investors.

A.1.7. Poland

1996: ING acquires Bank Slaski; three regioal banks are merged with Bank

Pekao; Bank Gdanski and BIG merge;1997: Kredit Bank and PBI merge; Commerzbank acquires BRE; Hypovereins-

bank buys BPH (Bank of Industry and Commerce); Allied Irish Bank (AIB)


20/21


acquires WBK; PBK sold to foreign and domestic institutional investors; BankHandlowy sold to employees, individual, institutional and strategic investors;

1998: Unicredito Italiano and Allianz buy Bank Pekao; AIB acquires Bank

Zachodni; Bank Austria-Creditanstalt acquires PBK (General Credit Bank);2000: Deutsche Bank acquires BIG Bank Gdanski; Citibank acquires Bank

Handlowy.

References

Angelini, P., Cetorelli, N., 1999. Bank competition and regulatory reform: the case of the Italian bankingindustry, Federal Reserve Bank of Chicago Working Paper 9932.

Bank for International Settlements, 2001. The banking industry in the emerging market economies:competition, consolidation, and systemic stability, BIS Papers No. 4.

Barajas, A., Steiner, R., Salazar, N., 2000. The impact of liberalization and foreign investment inColombias financial sector. J. Dev. Econ. 157196.

Baumol, W.J., Panzar, J.C., Willig, R.D., 1982. Contestable Markets and the Theory of IndustryStructure. Harcourt Brace Jovanovich, New York.

Belaisch, A.J., 2003. Do Brazilian Banks Compete?, IMF Working Paper 03y113.

Berger, A.N., 1995. The profitstructure relationship in banking-tests of market power and efficientstructure hypothesis, J. Money Credit Banking 27(2). pp. 404431.

Berger, A.N., Demsetz, R.S., Strahan, P.E., 1999. The consolidation of the financial services industry:causes, consequences, and implications for the future. J. Bank. Financ. 23, 135194.

Berger, A.N., Klapper, L., Udell, G.F., 2001. The ability of banks to lend to informationally opaquesmall businesses. J. Bank. Financ. 25, 21272167.

Bikker, J.A., Groeneveld, M.J., 2000. Competition and concentration in the EU banking industry. Kreditund Kapital 1 (1), 6298.

Bonaccorsi di Prati, E., DellAriccia, G., forthcoming. Bank Competition and Firm Creation, Journal ofMoney, Credit and Banking.

Burdisso, T., Catena, M., DAmato, L., 2001. Bank Competition in Argentina, mimeo.

Canoy, M.M., Ruud de Mooij, J., Weigand, J., 2001. Competition and Stability in Banking, CPBDocument No. 015.

Cetorelli, N., 1999. Competitive analysis in banking: appraisal of the methodologies. EconomicPerspectives. Federal Reserve Bank of Chicago, Chicago.

Claessens, S., Djankov, S., Lang, L.H.P., 1999. Who Controls East Asian Corporations? Policy ResearchWorking Paper No. 2054, Washington: World Bank.

Claessens, S., Demirguc-Kunt, A., Huizinga, H., 2001. J. Bank. Financ. 25 (5), 891911.

Clarke, G., Cull, R., Martnez-Peria, M.S., Sanchez, S., 2001, Bank Lending to Small Businesses in Latin America: Does Bank Origin Matter?, World Bank Policy Research Working Paper.

Coccorese, P., 1998. Assessing the competitive conditions in the Italian banking system: some empiricalevidence. BNL Q. Rev. 205, 171191.

Cordella, T., Yeyati, E.L., 2002. Financial opening, deposit insurance, and risk in a model of bankingcompetition. Eur. Econ. Rev. 46, 471485.

De Bandt, O., Davis, E.P., 2000. Competition, contestability and market structure in European bankingsectors on the eve of EMU. J. Bank. Financ. 24, 10451066.

De Bonis, R., Ferrando, A., 1997. Da Che Cosa Dipendono i Tassi dInteresse nelle Province, BancadItalia, Temi di Discussione n. 319.

Egli, D., Rime, B., 2000. The UBS-SBC Merger and Competition in the Swiss Retail banking Sector,Studienzentrum Gerzensee Working Paper 00.02.

Feyzioglu, T., Gelos, R.G., 2000. Why is Bank Lending so Low in Bulgaria?in International MonetaryFund, BulgariaSelected Issues and Statistical Appendix, Country Report No. 00y54.

Freixas, X., Rochet, J.-C., 1997. Microeconomics of Banking. Cambridge, Massachusetts, MIT Press.


21/21


Fuentes, I., Sastre, T., 1998. Implications of restructuring in the banking industry: the case of Spain.BIS Conf. Papers 7, 98120.

Group of Ten, 2001. Report on Consolidation in the Financial Sector (Basel: Group of Ten).

Guillen, M.F., Tschoegl, A.E., 1999. At Last the Internationalization of Retail Banking? The Case ofthe Spanish Banks in Latin America. The Wharton School, University of Pennsylvania, mimeo.

Hannan, T.H., 1997. Market share inequality, the number of competitors, and the HHI: an examinationof bank pricing. Rev. Ind. Organ. 12, 2335.

Hempell, H.S., 2002. Testing for Competition Among German Banks, Deutsche Bundesbank DiscussionPaper 04y02.

International Monetary Fund, 2000. International Capital MarketsDevelopments, Prospects, and KeyPolicy Issues, Washington, DC.

International Monetary Fund, 2001. International Capital MarketsDevelopments, Prospects, and KeyPolicy Issues, Washington, DC.

Lang, G., 1997. Wettbewerbsverhalten Deutscher Banken: Eine Panelanalyse auf Basis der Rosse-PanzarStatistik. Jahrbuch der WirtschaftswissenschaftenRev. Econ. 48, 2138.

Mathieson, D., Roldos, J., 2001. Foreign banks in emerging markets. In: Robert, E.L., Masson, P.,Pomerleano, M. (Eds.), Open Doors: Foreign Participation in Financial Systems in DevelopingCountries. Brookings Institution Press, Washington DC.

Matutes, C., Vives, X., 1996. Competition for deposits, fragility, and insurance. J. Financ. Intermed. 5,184216.

Molyneux, P., Lloyd-Williams, D.M., Thornton, J., 1994. Competitive conditions in European banking.J. Bank. Financ. 18, 445459.

Molyneux, P., Thornton, J., Lloyd-Williams, D.M., 1996. Competition and market contestability inJapanese commercial banking. J. Econ.Bus. 48, 3345.

Mooslechner, P., Schnitzer, Y., 1995. Structure-performance in banking: an application of a typicaluniversal banking system. In: Aiginger, K., Finsinger, J. (Eds.), Applied Industrial Organization.

Kluwer, Dordrecht, pp. 167186.Nakane, M., 2001. A test of competition in Brazilian banking, Banco Central do Brasil Working PaperNo. 12.

Nathan, A., Neave, H., 1989. Competition and contestability in Canadas financial system: empiricalresults. Can. J. Econ. 22, 576594.

Panzar, J.C., Rosse, J.N., 1987. Testing for monopoly equilibrium. J. Ind. Econ. 35 (4), 443456.

Peek, J., Rosengren, E.S., 2000. Implications of the globalization of the banking sector: the latinAmerican experience. New England Econ. Rev. 1, 4562.

Radecki, L.J., 1998. Small Expanding Geographic Reach of Retail Banking Markes, Federal ReserveBank of New York Economic Policy Review 4, 1534.

Rime, B., 1999. Mesure du degre de concurrence dans le systeme bancaire suisse a laide du modele

de Panzar et Rosse. Revue Suisse DEconomie Politique et de Statistique 135 (1), 2140.

Schargrodsky, E., Sturzenegger, F., 2000. Banking regulation and competition with product differentia-tion. J. Dev. Econ. 63 (1), 85111.

Shaffer, S., 1982. A non-structural test for competition in financial markets, In: Bank Structure andCompetition, Conference Proceedings, Chicago: Federal Reserve Bank of Chicago, 225243.

Shaffer, S., 1989. Competition in the US Banking Industry. Econ. Lett. 29 (4), 349353.

Tirole, J., 1988. The Theory of Industrial Organization. MIT Press, Cambridge, Massachusetts.

Vesala, 1995. Testing competition in banking: behavioral evidence from Finland, Bank of FinlandStudies Working Paper No. E:1.

Yoshitomi, M., Shirai, S., 2001. Designing a financial market structure in post-crisis Asiahow todevelop corporate bond markets, Asian Development Bank Working Paper 15.

consolidation and market structure in emerging

Documents