Financial Fragilities in Latin America - The 1980s and 1990s Liliana Rojas-Suarez and Steven R. Weisbrod INTERNATIONAL MONETARY F U N D Washington D C October 1995 O 1995International Monetary Fund Library of CongressCataloging-in-PublicationData Rojas-Sukz, Liliana. Financial fragilities in Latin America : the 1980sand 1990s/ Liliana Rojas-Sukz and Steven R. Weisbrod. p. cm. -(Occasional Paper, ISSN 0251-6365 ;no. 132) Includes bibliographical references. ISBN 1-55775-502-7 I . Banks and banking-Latin America. 2. Finance-Latin America. I. Weisbrod. Steven Riess, 1948- . 11. Title. 111. Series: Occasional paper (International Monetary Fund) ;no. 132. HG2710.5.A6R64 1995 332.1'098--dc20 95-44242 CP Price: US$15.00 (US$12.00 to full-time faculty members and students at universities and colleges) Please send orders to: International Monetary Fund. Publication Services 700 19th Street, N.W., Washington, D.C. 20431, U.S.A. Tel.: (202) 623-7430 Telefax: (202) 623-7201 Internet: publications@imf.org recycled paper Contents Page Preface vii I Introduction II Role of Banksin DevelopingCountries Overview of Financial Structure in Latin America Unique Features of Banks in Developing Countries Measuring the Franchise Value of the Bank in Developing Economies Central Bank and Franchise Value of Banking System Ill Franchisevalueof Banksand Resolutionof BankingCrises: 1982-90 Evaluating Banks' Strength at the Onset of the Debt Crisis Macroeconomic Environment and the Franchise Value of Banking Systems How the Systems Responded to Immediate Crisis: 1982-85 Crisis Resolution: 1985-90 Chile and Colombia Argentina, Mexico, and Peru State of the Banking Systems at the Beginning of the 1990s IV New BankingCrisis in LatinAmerica: 1994-95 Mexico Evaluating the Banking Franchise Restructuring the Banking System Argentina Evaluating Banks Rescuing Banks Comparison of Franchise Value in Mexico and Argentina on the Eve of the Crisis V Long-Term Challengesto FranchiseValueof Banks Sterilization and Franchise Value of Banking System Strength of the Franchise and the Impact of Sterilization Capital Inflow Challenge and Franchise Value of Banks: Some Empirical Evidence Capital Market Threat to Banking Franchise Value VI Financial Soundness and MacroeconomicStability: A Bank Balance SheetApproach Bank-Intermediated System Facing Significant Dollarization Monetary Base Fully Backed by Foreign Exchange Reserves Monetary Base Partially Backed by Foreign Exchange Reserves 60 A More Diversified Financial System 61 Lack of Confidence in Exchange Rate Policy 61 Lack of Confidence in Both Exchange Rate Policy and Soundness of Banking System 62 Results 62 Appendix. Mechanics of a SpeculativeAttack on a Sound, Dollarized BankingSystem 64 References 66 Tables Section I1 1. Composition of the Stock of Net Credit to the Private Sector by Commercial Banks and Other Financial Institutions 2. Market Capitalization 3. Number of Domestic Companies Listed on Selected Stock Exchanges 4. Chile: Private Sector Finance Structure 5. Corporate Finance Structure in Germany and the United States 111 6. Indicators of Bank Franchise Value, 1982 7. Fiscal Balance and Inflation IV 8. Selected Mexican Banking Data 9. Selected Argentine Banking Data V 10. Chile: Corporate Bonds Outstanding 11. Chile: Pension Fund Assets by Type of Issuing Agency, December 1993 Charts Section 111 1. Balance of Payments: Current Account and Overall Balance 2. Real Interest Rates 3. Capital to Assets 4. Selected Financial Indicators 5. Inflation 6. Cash Assets to Deposits 7. Loans to Assets 8. Total Deposits IV 9. Indexes of Brady Bond Prices 10. Equity Price Indexes, U.S. Dollar Terms 11. Mexico: Deposit and Repurchase Agreement Rates 12. Argentina: Peso and Dollar Interest Rates on Bank Deposits and Loans, 1993 V 13. Ratio of International Reserves to Narrow Money Supply 14. Volatility of Local Equity Market Index 15. Volatility of Local Equity Market Index, U.S. Dollars 16. Price Earnings Ratios 17. Mexico: Interest Rate Spreads on Selected Financial Instruments, March 1991-March 1994 AppendixTables Al. Argentina: Summary Accounts of the Financial System A2. Mexico: Summary Accounts of the Financial System A3. Peru: Summary Accounts of the Financial System A4. Chile: Summary Accounts of the Financial System A5. Colombia: Summary Accounts of the Financial System The following symbols have been used throughout this paper: . . . to indicate that data a[-enot available; - to indicate that the figure is zero or less than half the final digit shown, or that the item does not exist; - between years or months (e.g., 1991-92 or January-June) to indicate the years or months covered, including the beginning and ending years or months; I between years (e.g., 1991192) to indicate a crop or fiscal (financial) year. "Billion" means a thousand million. Minor discrepancies between constituent figures and totals are due to rounding. The term "country," as used in this paper, does not in all cases refer to a territorial entity that is a state as understood by international law and practice; the term also covers some territori- al entities that are not states. but for which statistical data are maintained and provided internationally on a separate and independent basis. Preface This study was prepared by Liliana Rojas-SuBrez, currently on leave from the IMF as Principal Advisor to the Chief Economist at the Inter-American Development Bank, and Steven R. Weisbrod, a consultant to the IMF. Norma Alvarado, Isabel Car- dona, and Cecilia Coder provided excellent assistance in typing text and tables. Ivan Sergio Guerra, Kellett W. Hannah, and Subramanian S. Sriram provided highly com- petent research assistance. The authors are also grateful to Elisa Diehl of the External Relations Department, who edited the manuscript and coordinated production of the publication. The study has benefited from the comments of staff and consultants in the Re- search Department and other departments in the IMF. Opinions expressed are those of the authors and do not necessarily reflect the views of the IMF. I Introduction n late 1994, several Latin American economies, I whether the financial reforms implemented in the particularly Mexico and Argentina, experienced early 1990s have strengthened their hand in dealing sharp reversals of international capital inflows that with financial crisis. By reviewing the experiences had characterized the previous four years. The im- of Latin American countries with the restructuring of mediate cause of the reversals was the loss of in- their financial sectors since 1982, this paper derives ternational investor confidence in these countries' lessons regarding the most effective ways to deal ability to defend their exchange rate and in the au- with banking difficulties in developing countries. It thorities' ability to service their external debt on a then discusses whether these lessons have been out timely basis. Short-term interest rates rose to extra- into practice during the latest crisis. A sample of five ordinarily high levels. Because many borrowers countries-Argentina, Chile, Colombia, Mexico, could not afford to service their debts at these high and Peru-is used for this oumose. * . interest rates, the credit quality of domestic banks' In addition, the paper analyzes policy issues asso- loan portfolios deteriorated dramatically, creating ciated with the long-run health of the financial sys- concerns about the solvency of the banking systems tem: (1) the proper design of policies to respond to in these economies. large and volatile flows of capital having the com- The current crisis has occurred during a period of plementary objectives of maintaining long-run economic and financial reform in Latin America. macroeconomic stability and a healthy financial sys- After almost a decade of macroeconomic and finan- tem; and (2) the effect on bank soundness of in- cial difficulties, since the last part of the 1980s and creased competition from recent developments in the beginning of the 1990s, many Latin American domestic capital markets. countries have undertaken maior transformations of Section I1 sets the stage by presenting a frame- their economic structures. These efforts have in- work for analyzing banks' behavior. A lack of trans- cluded not only comprehensive stabilization pro- parency in the legal and accounting infrastructures is grams aimed at correcting macroeconomic imbal- the main feature that defines the soecial role of ances but also deep structural reforms designed to banks in developing countries: because investors improve the efficiency of market mechanisms in cannot rely on the legal infrastructure to aid in eval- pricing and allocating resources among the different uating the creditworthiness of borrowers, they sectors of these economies. search for alternative methods of evaluation. One Among the reforms, the restructuring of financial method is for lenders to force borrowers to remain markets was crucial. Latin America does not stand liquid by restricting their borrowing opportunities to alone in having learned the lesson that difficulties in short-term funds and carefully monitoring their cash the financial sector may conflict seriously with pol- flow; that is, in these financial systems liquidity be- icy objectives as governments have had to abandon comes a primary proof of solvency. In developing their fiscal and monetary targets to rescue insolvent countries, banks are in a unique position to interme- or troubled financial institutions. Several industrial diate between borrowers and lenders because they countries, including the United States and three are the only nongovernment issuers of short-term Nordic countries, are among the recent cases where liabilities. banking difficulties have had a severe impact on the A sound banking system issues the appropriate fiscal stance. Within the Latin American region, the loan contracts and establishes the appropriate moni- banking difficulties experienced in a number of toring procedures to maintain borrower liquidity. countries following the eruption of the debt crisis in Conclusions regarding the behavior of banks in this the 1980s imposed constraints on policymakers that, environment are based on incentive-driven argu- in some cases. lasted for almost an entire decade. ments: where policymakers put in place mechanisms As policymakers again face challenges similar to to motivate banks to perform their role of main- those faced in the 1980s, it is important to assess taining borrower liquidity, bankers become well equipped not only to evaluate properly the risks they a sound banking system in dealing with an adverse undertake, but also to develop "workout" programs shock that leads to a speculative attack on the ex- with their borrowers if sudden adverse shocks change rate. weaken the quality of loans, thereby leading to bank- Section V focuses on two factors that affect the ing difficulties. Thus, although a strong banking sys- strength of the banking system: sterilization policies tem does not mean that banking difficulties can be and the securitization of many financial instruments prevented altogether, it implies that problems are that heretofore appeared mostly on bank balance faced promptly and bankers have incentives to re- sheets. Regarding the first factor, the impact of ster- store defaulted borrowers to performing status. ilization policies on the soundness of banks is em- Section 111 uses the framework described above phasized. Conclusions regarding both the desirabil- to examine the resolution of banking crises in the ity and the method of sterilization are linked to the sample of five Latin American countries during the strength of the central bank relative to that of com- 1980s. The discussion shows that the strength of mercial banks. The message is straightforward: as banks at the onset of the crisis and the quality of the decision to sterilize or not implies a decision to central bank leadership were important determi- concentrate resources in the central bank rather than nants in how quickly public confidence was re- in the commercial banks or other financial institu- stored to the financial systems in the sample coun- tions, it follows that resources should be channeled tries. Countries are grouped by the strength of their to the institutions that can manage those funds better. banking systems at the inception of the banking cri- This section concludes that the stability of invest- sis. In those countries where the strength of the ments made in the domestic equity markets is banking system was greater, bank supervisors and closely related to the strength of the banking sys- bankers were able to respond to the crisis with a tems: stock market volatility is lower in those coun- credible program to restore confidence in the bank- tries with relatively strong banking systems. ing system; although the programs involved a sub- The analysis also shows that the second risk to the stantial increase in credit, the soundness of the res- strength of banks, namely, the recent development of cue programs prevented the eruption of inflation. In capital markets, is still years away from becoming a sharp contrast, in those countries with relatively serious threat to bank soundness. Even in those weak banking systems, banking regulators further countries where fixed-income markets-such as aggravated the problem by attempting to take over commercial paper or the corporate bond market- the role of banks as direct lenders; that is, supervi- have developed, open market interest rates are still sors removed authority from hankers, substituting high relative to bank interest expenses and the in- the credit judgment of the central bank or the gov- struments are still held by only a few investors. The ernment directly. In those cases, credit expansion discussion shows that a strong banking system com- was associated with big inflations as the public real- plements, rather than competes with, the develop- ized that the credit extended would not be repaid in ment of healthy markets for equity issues. The de- real terms. velopment of equity markets, however, may pose By the early 1990s, many of the perceived mis- some dangers for policymakers worried about unsta- takes of the 1980s were corrected, and comprehen- ble capital inflows. In this connection, the analysis sive efforts were made to liberalize financial markets shows that these dangers can be minimized by and strengthen bank supervision. However, by the strengthening the domestic banking sector. end of 1994, the reformed systems of two of the five Finally, Section VI deals with a key macroeco- sample countries-Mexico and Argentina-were se- nomic issue: the ability of central banks to withstand verely challenged when a reversal of international speculative attacks on the exchange rate, with an capital flows led to sharp increases in domestic in- emphasis on the role of banks' performance in facili- terest rates and a general weakening of the quality of tating the policy objective. The main argument is financial assets. Section IV discusses the strengths that the degree to which a Latin American central and weaknesses of the banking systems in these two bank may be able to withstand a speculative attack economies on the eve of the 1994crisis and assesses on its domestic currency is influenced by two as- the options available to policymakers in restructur- pects of the financial markets: the first one, which is ing their financial systems. a natural extension of the well-known literature on The last two sections of the paper deal with the speculative attacks, is the extent of the commitment long-run challenges policymakers face in dealing of the central bank to the stabilization of prices in with the performance of the banking sector. Section V the financial sector. The second one, and the one em- discusses the impact of stabilization policies and phasized in this paper, is the strength of the banking structural developments in capital markets an the sector. In this connection, the paper also deals with strength of the banking system, and Section VI turns the role of dollarization. A conclusion from the the question around and analyzes the contribution of analysis is that, when the banking system is sound, dollarization may be an ally for governments pursu- straint on policymakers facing a speculative attack. ing exchange-rate-based stabilization programs. In In dealing with these topics, this section addresses an insolvent, bank-dominated financial system, the issue of the appropriate holdings of foreign ex- however. dollarization imposes an additional con- change reserves by central banks. I I Role of Banks in DevelopingCountries Financial systems in developing countries are typ- cia1 sector reforms) and external (the decline in in- ically dominated by banks: bank deposits consti- terest rates in the United States and other industrial tute the most important form of household savings, countries, which increased foreign investors' de- and bank loans are the most important source of ex- mand for Latin American securities), contributed to ternal finance for firms. As will be shown below, the rapid expansion of alternative sources of finance Latin American countries by and large have this type for firms, such as bonds and equity. However, bank of financial structure. This section focuses on the loans remained the most important source of finance particular features that distinguish banking systems for the private sector. in developing countries from those in industrial The composition of credit commercial banks and countries and on the special role that the banking other financial institutions (excluding the stock ex- sector plays as a source of economic growth in de- changes) provided to the private sector in a number veloping economies. of Latin American countries indicates a clear bank Banks-in both industrial and developing dominance through the 1980s and early 1990s al- economies--can be distinguished from other finan- though the importance of banks varied across coun- cial institutions by a unique characteristic that will tries and across time (Table 1). In this regard, two be termed here "the franchise value of banks," that clarifications of the data are needed. is, the special power conferred by the banking char- First, owing to a lack of consistent data across ter to issue liabilities that are accepted as a means of Latin American countries. develovment banks-in- payment. In developing countries, the state of the stitutions, typically government owned, established legal and accounting systems makes it difficult for to extend credit to specific sectors of the economy- institutions that are not connected to the payments are included under-"other financial institutions."2 system to issue short-term liabilities such as com- Hence, in several countries, a large component of mercial paper. Hence, banks are the only nongovern- credit extended through other financial institutions is ment issuers of these liabilities. Because investors also bank credit.3For example, assets held by devel- require borrowers' liquidity as proof of their sol- opment banks in Bolivia, Guatemala, and Peru ac- vency, borrowers are restricted to the short-term counted for 21, 14, and 47 percent of total assets of market, which is dominated by banks.' financial institutions, respectively, by the end of 1987 (Morris and others, 1990). The importance of development banks, however, declined significantly Overview of Financial Structure in during the late 1980s and early 1990s, reflecting the LatinAmerica privaiization programs in many Latin American countries. Mexico is a clear example of the privati- Commercial banks played a central role in Latin zation efforts; there, the share of commercial bank America in the 1980s.As reported in Morris and oth- credit in total credit to the private sector increased ers (1990). these institutions provided short-term fi- from 76 percent in 1987 to 9 1 percent in 1992. The nancing that, owing to the severe economic difficul- same trend is apparent in Bolivia, Brazil, Ecuador, ties these countries faced during the decade, was the only kind of resource that financial institutions could 2The exception is Argentina, where the category "coo~mercial mobilize. banks" includes the Caja Nacional de Ahorro y Seguros. the During the late 1980s and early 1990s, a number Banco Hipotecario Nacional, and the Banco Nacional de Desar- of variables, both domestic (macroeconomic stabi- rollo (BANADE). BANADE was closed in May 1993. As re- lization programs, financial liberalization, and finan- ported in Morris and others (1990). assets of developing financial institutions in Argentina accounted for about 20 percent of total assets of the financial sector by 1987. 'For a discussion of the role of banks in transition economies, Vypically, development banks are recipients of public funds see Blommestein and Spencer (1994). and enjoy special privileges from the government. Table I. Cornpodtionofthe Stockof NetCreditto the PrivateSector by CommercialBanla and Other Financiallnrtitufions 1980 1987 1992 Other Other Other Commercial financial Commercial financial Commercial financial Country banks institutions banks institutions banks institutions Argentina' 85.42 14.58 97.99 2.0 1 98.75 1.25 Bolivia 73.90 26.09 76.37 23.63 93.982 6.022 Brazil' - - 48.884 51.124 62.35 37.65 Chile - - 95.44 4.575 78.70 21 .305 Colombia6 46.39 53.6 1 42.12 57.887 49.0 1 50.997 Ecuador 52.32 47.68 57.98 42.02 66.862 33.143 Guatemala 79.158 20.85 85.1 28 14.88 85.718.9 14.299 Honduras 60.05 39.95'0 74.79 25.2 1'O 78.09 21.91'0 Mexico 80.62 19.38 75.84 24.16 9 1.27 8.73 Peru 53.37 46.63 'I 51.01 48.99l 75.93 24.07l' Venezuela 55.25 44.75'2 67.2 1 32.79'2 85.26 14.74'2 Source: IMF staff estimates. Note: Commercial banks include private and g~vernment~ownedcommercial banks. Other financial institutions include development banks and all other financial institutions whose activitiesare reported by the central banks.The informal financial markets and the stock exchanges are not included. lCommercial banks includenationaland provincialbanks also. =June1992 data. 'Aker 1985, commercial banks include the Bank of Brazil.The rest of the banking system includes the National Development Bank (BNDES), state developmentbanks, investment banks, the FederalSavings Bank (CEF), state savings banks,savings and loan associations, housing credit companies, the National Housing Bank (BNH), and the National Bank of Cooperative Credit (BNCC). 4 1988 data. Slncluding nonbank financial intermediariesand pension funds. 6A new reporting system for financial system accounts was introduced at end-1990. Data for 1990are therefore not strictly comparable with earlier dam. Data for 1990 exlude PROEXPO. 'Comprising development finance corporations, trade finance companies, savings and loan companies, cooperative institutions, and development banks (BANCOLDEX, FINAGRO, FINDETER). 81ncludingdevelopmentbanks. 9August 1992 data. 'Olncludes National Bank for Agricultural Development (BANADESA), the Municipal Bank (BMA), the specializedsavings institutions, the National In- vestment corporation (CONADI), and private nonbankfinancial intermediaries.The private nonbank financial intermediaries consist of the Honduran Federationof Savings and Loan Cooperative (FACACH), insurancecompanies, and the Honduran Federationof Housing Cooperatives (FEHCOVIL). 'Includes the National Bank and developmentbanks only. lllncludes mortgagebanks, the Agricultural Development Bank (BANDAGRO), savings and loans, and theworkers' Bank. and Peru.4 The reduction in government participa- Second, the data in Table 1 do not include the in- tion in the financial sector has, therefore, resulted in formal financial market and the stock exchanges. a significant increase in the importance of commer- The informal market became an important source of cial banks in financing the activities of the private financial intermediation in many Latin American sector. Indeed, the share of credit from commercial countries during the years of severe financial diffi- banks in total credit to the private sector clearly ex- culties in the banking sectors of several Latin Amer- hibited an upward trend during 1980-92 in all but ican countries; that pattern, however, reversed in the one of the countries shown in Table 1. The excep- late 1980s and early 1990s in many countries as sta- tion was Colombia, where savings and loan institu- bilization policies and policies directed at restructur- tions maintained about one-third of the total finan- ing and recapitalizing banks resulted in a reinterme- cial credit to the private sector throughout the period diation of financial activities back into the formal under consideration. institutions (see Section 111). In contrast, the stock exchanges were not important sources of finance for the private sector as a whole during the early and 4Although Venezuela showed a similar trend b y 1992, the re- cent banking crisis was followed b y large injections o f govern- mid-1980s when many Latin American stock mar- ment funds into the commercial banks in early 1994. kets even contracted, as reflected by their market Country 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 - - - Argentina Brazil (Sio Paulo) Chile Colombia Costa Rica Mexico Peru Venezuela Sources:lnternational Finance Corporation. Emerging Stock Markets: Factbook, various issues;and International Monetary Fund, World EconomicOutlook capitalization measured both in U.S. dollars and as a concentration in several developing countries in percentage of GDP (Table 2).5 Asia (including China, Indonesia, Korea, Malaysia, consistent with better economic prospects and, and haila and),-which averaged about 30 percent by more recently, with a large inflow of foreign capital, the end of 1993.7 market capitalization in several of these countries re- Have banks remained the single most important covered in 1989-90 and expanded dramatically in source of finance for the private sector? The paper 1991-93.6 In spite of the tremendous growth of eq- explores the possibility of using national flow of uity prices andin the daily turnover of private sector funds data, which provide information on financial shares, equity finance in Latin America is still con- asset and liability structures across sectors. Data on fined largely to the largest corporations and has not corporate financial structure are essential to assess yet become a significant competitor for bank finance the relative importance of bank loans in financing for the corporate sector as a whole. production. Unfortunately, flow of funds data are The number of com~anieslisted on several Latin ~racticallvnonexistent in most Latin American American stock exchanges is shown in Table 3. Al- countries. Indeed, time-series data on flow of funds though the absolute number of listed companies has were found only for Chile (for 1978-91).8 However, limited usefulness because it does not account for given the recent developments in financial markets the difference in size among corporations or the size in Chile-the rapid growth of pension funds, the sig- of the country, the important conclusion that can be nificant increase in the issuance of corporate bonds, derived from Table 3-is that, with the exception of and the rapid increase in market capitalization in the Peru, the number of listed companies did not show any significant growth in spite of the sharp increase 'Data on market concentration are taken from International Fi- in market capitalization during the most recent pe- nance Corporation, Emerging Stock Markers: Factbook (various riod. These data, therefore. indicate that the large issues). XLimited data for Mexico are also reported in Singh and Hamid capitalization observed recently in several countries (1992). The data show that during 1984-88, the top fifty manu- may be attributed to the performance of a limited facturing corporations listed on the stock exchange used equity as number of stocks and not to a generalized improve- their most important external source of finance. However, this re- ment in the performance of corporations in Latin sult is derived using financing flows rather than stocks and should therefore be viewed with caution as it may lead to serious misin- America. Moreover, by the end of 1993, market con- terpretations of the Mexican financial structure. During 1984-88, centration-measured as the share of market capital- restrictive monetary policies in Mexico had a significant impact ization held by the ten largest stocks-was above 50 on the availability of bank loans. Moreover, during that period, percent in a number of Latin American countries, in- corporations in Mexico achieved almost no real growth. Thus, as cluding Argentina, Chile, Colombia, Peru, and the authors themselves concluded: ". . . in the peculiar circum- stances of the Mexican economy in the mid-1980s. the Mexican Venezuela.This compares with a much lower market corporations achieved relatively little growth; but of the growth that did occur, a large proportion of it was financed by equity" 5Brazil is a noticeable exception to this trend. Market capital- (p. 47). The point to be learned from this analysis is that in ization is defined as the market \.slue of the equity of firms economies facing large variations in real eco~~oniicactivity and in quoted on the stock exchanges. the design of economic policies, conclusions regarding the corpo- hFor a discussion of the issues related to the rapid increase in rate financial structure cannot be based on flow data, which may market capitalization in a number of developing countries, see be temporary. The particular advantage of using stock data is that Feldman and Kumar (1994). they dilute temporary fluctuations. Country 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 Argentina 278 263 248 238 236 227 217 206 186 178 179 174 175 180 Brazil (510 Paulo) 426 477 493 505 522 541 592 590 589 592 581 570 565 550 Chile 265 242 212 214 208 228 231 209 205 213 215 221 245 266 Colombia - - 193 196 180 102 99 96 86 82 80 83 80 80 Costa Rica 13 19 24 32 41 51 61 71 76 78 - 82 93 - Mexico 259 229 206 163 160 157 155 190 203 203 199 209 195 190 Peru 103 133 144 150 157 159 177 197 236 256 294 298 287 - Venezuela - - 98 - 116 108 108 110 60 60 66 66 66 93 Source Internat~onalF~nanceCorporat~on(1993) stock exchange-it is worth exploring what this ex- increase in equity prices may be due to capital gains perience may show in terms of the importance of on projects undertaken and financed in previous bank loans in financing private sector activities. years. For example, assume a firm undertakes a mil- Table 4 shows the financing sources available (in lion peso project in year 1 that it finances half with stocks) for households and firms in Chile during equity and half with bank loans. In year 2, the mar- 1978-91. It is well known that increases in equity ket value of the project increases to 2 million pesos prices may present a problem in measuring the rela- in real terms. The equity issued in year 1 will reflect tive sources of corporate funding because part of the the entire capital gain on the project, not just that earned on the equity portion because the bank loan does not capture capital gains. Thus, in year 1, using current market prices of equity, bank financing and Table4, Chile: PrivateSectmr Finance equity financing are each 500,000 pesos. In year 2, structure equity financing increases to 1.5 million pesos and ( R ~ o f m I d o m o J t k ~ ~ ~ & @ bank loan funding remains constant, even though the proportion of the project funded by each instrument has not changed. Corbo and others (1992) attempt to Corporate correct for this effect, and it is their adjusted figures Year Bank Loans' Bonds Equity* that are used here.9 1978 0.69 0.0 1 0.22 Bank loans remained the most important source of 1979 0.73 0.0 1 0.23 domestic finance for the Chilean private sector 1980 0.76 - 0.22 (households and firms) during the period under 1981 0.80 - 0.17 1982 0.78 0.03 0.18 study, averaging 79 percent over the entire period. 1983 0.79 0.03 0.18 The share rose during the years immediately after 1984 0.8 1 0.03 0.15 the debt crisis, peaking at 86 percent in 1985-86. Fi- 1985 0.86 0.03 0.1 I nancial reform measures taken since 1986,including 1986 0.86 0.02 0.1I 1987 0.84 0.03 0.12 the new banking law, which limited permissible ac- 1988 0.82 0.05 0.12 tivities by banks, and the increased competition from 1989 0.81 0.07 0.1 I other financial institutions lowered the share of bank 1990 0.75 0.1 I 0.14 loans in total domestic financing.I0 Corporate bonds 1991 0.70 0.13 0.17 rather than equity-which by the end of 1991 had Source: Corbo and others. ElSistema BancorioChileno: Desorml- not reached its predebt crisis share in domestic fi- 10s Rec~entesy sus Perspeciivas. Institute de Economia. Pontificia nancing-were the main source of domestic compe- Universidad Catolica de Chile (1992).Table 3. tition for bank loans during the period considered. Note: The finance structure includes domestic sources of fi- As will be further discussed below, the emergence nance and government enterprises. Ratios do not need to add to one because there are other financial liabilities (such as liabilities from Sisrema Nacional de Ahorro y Prestamos (SINAP) not in- 9Corboand others (1992) ind~catetheir methodology may under- cluded here. state equity funding, but the understatement is less severe than the Ilnclude banks and savings companies. overstatement that would arise if current market prices were used. =Adjustedfor capital gains on previous stock issues. IoTherecent increased access by a number of firms to interna- tional capital markets may be viewed as an additional source of competition to bank loans. This issue is discussed in Section IV. often cited as representing the two extremes in the Tab& 5. CorporateFinanceStructurein range of financial market structures in the industrial Germany and the UnitedStates world. Germany represents the "universal banking (k~ent totor of totorkddnies l i o b i l ~ k s o u b ~ system," where banks face limited competition from other financial institutions, and the United States represents the "Anglo-Saxon" financial structure, - Other where securitized money and capital markets com- Bank Nonbank pete with wholesale banks as sources of funds for Loans Bonds Securit~esl Equlty the corporate sector. Not surprisingly, a major finding is that the ratio Germany 1980 60.7 1.5 24.5 13.3 of bank loans to domestic corporate liabilities in 1985 55.4 1.4 22.5 20.7 Germany was much higher than in the United States 1990 55.9 1.9 22.7 19.5 throughout the period under consideration.l2 The av- Un~tedStates erage share of bank loans in Germany's corporate li- 1980 12.2 15 7 22.9 49.3 ability was about 57 percent compared with an aver- 1985 11.8 16.1 23.5 48.6 age of about 12 percent in the United States. Indeed, 1990 11.9 17.1 23.8 47.1 in Germany, private securities markets remain frag- mented and relatively illiquid,'3 and, at the short end Sources: Deutsche Bundesbank (1992); and Un~tedStates, Board of Governors of the Federal Reserve System (1992) of the market, there are very few domestic substi- Note. Only domest~csources of finance are ~ncluded. tutes for bank loans. In contrast, commercial paper is 'Include cornmerc~alpaper, loans from finance cornpanles, and among the preferred instruments for short-term fi- government loans nancing by U.S. firms. Although the Latin American experience is in- dicative of the predominance of banks as a source of and rapid growth of private pension funds have finance, the discussion also shows that the impor- played a crucial role in the surge in medium-term tance of banks in domestic financial markets varies corporate bonds. across countries. Indeed, some financial systems, In contrast to the expansion of medium-term such as Argentina's, seem to be more bank oriented bonds, the market for commercial paper in Chile re- than, say, Chile's. Because financial maikets are still mains small and illiquid. This pattern diverges from being reformed, there is a significant probability that that observed in a number of industrial countries, a wide range of financial structures may emerge where corporations have been able to satisfy their from the transformation. An important lesson demand for liquidity and their short-term financing emerges from the experience of the industrial coun- needs directly in liquid securities markets." tries, namely that, even in a highly integrated and Notwithstanding the decline in the ratio of bank global financial environment, alternative financial loans in total financing in recent years, the ratio re- structures can coexist. There is no reason to believe mained high at 70 percent by the end of 1991. This that they cannot also coexist in Latin America. finding is important in the 1.atin American context. Among Latin American countries, Chile's banks probably face the greatest competition from other Unique Features of Banks in domestic financial institutions; that competition DevelopingCountries largely comes from private pension funds, which are only at an early stage of development in some coun- Gerald Corrigan, former president of the Federal tries (like Argentina and Peru) and nonexistent in Reserve Bank of New York, argues that banks are most others. Therefore, the predominance of bank special because the bank charter gives them the loans as a source of private sector finance in Chile unique power to provide means of payment in non- reinforces the view that this feature is a Latin Ameri- cash transactions; this special power is called the can phenomenon. franchise value of banks.14 When a bank customer How does the role of banks in Latin America com- pare with that in industrial countries? Table 5 uses 12For a more detailed comparison between the financial sys- flow of funds data for Germany and the United tems in Germany and the United States. see International Mone- States to show the ratio of bank loans to corporate tary Fund (1992). domestic liabilities during the 1980s and 1990s. "As reported in International Monetary Fund (1992). commer- These two countries were chosen because they are cial paper programs in Germany started only in 1991. It is inter- esting to note, however, that notwithstanding the limitations of private securities markets in Germany, the government bond mar- "See International Monetary Fund (1992) for a discussion of ket is large and liquid. the trend toward securitization observed in industrial countries. '4See Corrigan (1991). withdraws funds from his or her bank deposit or experience high levels of inflation, and the value of writes a draft against that account, the bank delivers the bank franchise would be destroyed; that is, the an "outside" asset, or "good fundsw-namely, re- special power conferred by the banking charter serves on deposit at the central bank, or cash-to the would be worthless because the bank liabilities customer or to the bank of the payee named on the would have lost their real value. Therefore, if the draft.I5 central bank wants to preserve the franchise value of In fact. when other liability issuers promise to de- the banking system, it must avoid lending large liver payment, they promise to deliver bank de- amounts of funds to banks on a sustained basis. To posits. Consider, for example, money market mutual this end, the central bank must ensure that banks funds in the United States, which invest in money have procedures in place to monitor the ability of market assets, such as commercial paper and trea- their loan customers to deliver cash. In a market with sury bills, and issue short-term claims on this portfo- undeveloped accounting standards, this implies re- lio to investors. When an investor wants to use stricting borrowers to short-term loans and frequent money market mutual fund shares to purchase goods calls for payment of principal, which effectively and services, the money market mutual fund must means that most investment projects will be short deliver funds from its bank deposit to the bank de- term. In addition, banks must use the tools at hand- posit of the payee designated by the investor. namely, the threat to seize a firm's bank deposits and Even though money market mutual funds must freeze its ability to make payments-in order to en- rely on banks to make payments for them, the mar- force loan contracts. When a borrower gets into trou- kets for their assets have become so liquid that they ble, a bank must have established procedures to re- can be sold for bank deposits immediately and at solve problems quickly if it is to maintain its very low cost. Hence, the public is willing to hold commitment to deliver cash against deposits. money market mutual funds as perfect substitutes In sum, a sound banking system is taken to mean for bank deposits for some transactions. one that is able to preserve its franchise value-that In contrast, in developing countries, there are no is- is, banks' commitment to deliver good funds against suers of perfect substitutes for bank deposits, such as their deposit liabilities. In a developing country, the money market mutual funds, because markets for banking system must effectively monitor the liquid- such nonbank liabilities as commercial paper are ity of its borrowers and, in the face of difficulties, illiquid. Commercial paper markets are illiquid be- quickly establish loan workout programs to restore cause the accounting and legal frameworks are insuf- defaulted borrowers to performing status. ficiently developed to permit investors to evaluate corporate cash flow and their legal standing in the event of default for all but the most well-known Measuringthe Franchisevalue of the firms; that is, investors cannot rely on the legal infra- Bank in Developing Economies structure to aid in evaluating the creditworthiness of most potential borrowers. Hence, the potential pool When investors evaluate the quality of banks in of issuers is not large enough to create a liquid market developing countries, they confront the same obsta- for nonbank short-term paper. Therefore, in develop- cles that they face in evaluating nonfinancial firms: ing countries, banks are not only the unique issuers of accounting data are often undependable guides to the means of payment, they are also the unique non- quality. For example, usual indicators of bank sound- governmental issuers of all liquid instruments. ness in industrial countries, such as ratios of capital In developing country markets, the only institu- to assets and loan loss provisions to nonperforming tions that can credibly promise to deliver bank de- loans, are often uninformative because banks are not posits are banks. Banks' claim to deliver means of subject to standard procedures for placing loans on payments is more credible than claims of other nonaccrual status or deducting defaulted credits from liability issuers because banks maintain deposits at capital and loan loss accounts. Hence, investors in the central bank and have access to a central bank developing countries must look for other ways to as- credit facility, usually referred to as discount win- sess the quality of bank balance sheets. dow privileges. A bank could convince investors that it is sound The public in developing economies is willing to and, therefore, able to deliver good funds by holding accept bank deposits as liquid liabilities because a large amount of cash assets--cash and deposits at banks hold deposits at the central bank. If, however, the central bank (reserves)-relative to its deposit lia- banks borrowed excessively from the central bank to bilities. In other words, a bank could convert itself keep their deposits liquid, the economy would soon into a vault. If, however, banks were to act as vaults, they would have less incentive to press borrowers to '5The unique role of banks as providers of "good funds"is ana- remain liquid, and they would reduce the amount of lyzed in Garber and Weisbrod (1992). credit supplied to borrowers for a given amount of deposits issued. In other words, the liquidity demands Central Bank and Franchisevalue of of investors would be met by holding cash assets in BankingSystem the central bank rather than by supplying credit to do- mestic borrowers in a form that forces borrowers to If a bank maintains low cash ratios and high loan remain liquid. The market discipline imposed by and deposit ratios relative to assets, it must, as indi- banks on borrowers would be adversely affected. cated above, ensure that its borrowing customers re- As the evidence presented in the next section main liquid. Even the best banks, however, cannot demonstrates, when banks do not discipline borrow- depend solely on their own loan customers for liq- ers, the credit risk in the financial system increases. uidity; they must have access to good funds through Some other institution, usually government related, the banking system to satisfy temporary shortages of ends up supplying credit without imposing disci- liquidity. Good funds take two forms: the interbank pline on borrowers. When borrowers default, bank market for short-term funds and loans from the cen- depositors are often forced to absorb the losses tral bank. through outright confiscation or through inflation. In many developing countries, the interbank mar- Hence, one measure of the quality of the bank fran- ket is uncompetitive; hence, practically speaking, chise is the ratio of cash assets to deposit liabili- the central bank must play a pivotal role in maintain- ties-a relatively high ratio represents a weak fran- ing bank liquidity. To fulfill this role, the central bank chise.16 That is, the market discipline exerted by must take the same attitude toward the banks with banks on borrowers-by requiring frequent delivery which it has a lending relationship that the banks of good funds as a way to prove borrowers' credit- must maintain with their borrowing customers. That worthiness-is reduced. In this connection, exceed- is, the central bank must ensure that the credit it ex- ingly high reserve requirements may jeopardiae the tends to a bank for liquidity purposes is not used to franchise value of banks. provide credit to borrowers that are in default. The A second and related measure of franchise value is central bank must therefore play a major role in su- the ratio of loans to assets. Banks that hold a high pervising banks or at least have access to the super- ratio of nonloan assets to assets-usually govern- visory data collected by other agencies.18In fact, a ment bonds, development bonds, and central bank major benefit of having a banking system with a low bonds-are not fulfilling their role of policing the ratio of cash to assets and a high ratio of loans to as- liquidity of borrowers. When reserve requirements sets is that the central bank is forced to maintain a are high, issuers of bonds often use the proceeds to close supervisory relationship with its banks because provide long-term credit to borrowers who are not it may be called upon to provide liquidity assistance. able to use these funds efficiently. In addition, when Moreover, when banks maintain high cash ratios, banks hold a high portion of their assets in govern- the central bank ends up with a large balance sheet ment-related bonds, bankers do not obtain the expe- relative to deposits outstanding because the cash, in rience of helping private borrowers work their way the form of either vault cash or reserves, is a liability out of credit problems. When credit crises occur, of the central bank. The evidence presented in Sec- bankers tend to try to resolve them by expanding tion 111 for a number of Latin American countries credit without establishing loan workout programs suggests that when cash assets held by banks at the to ensure that the new credit is used to correct the central bank were exceedingly large, the central deficiencies in the borrower's business plan that led bank took over much of the role of extending do- to credit problems in the first place." mestic credit from the banks.I9 When the central bank operates as a bank that provides direct credit to Ihln industrial economies, low liquidity ratios are often taken the market, one of the most important checks on loan as an indicator of problems in a bank; this is because banks in in- decisions is removed. As lender of last resort to dustrial countries operate at much lower cash ratios than those in banks, the central bank maintains an am's-length re- developing economies. It is necessary to use caution when apply- lationship with ultimate borrowers. Hence, it is in a ing, for developing countries, the same ratios used to assess banks7performance in industrial countries, ~h~ ratio of cash as. position to criticize the lending decisions of banks. sets to deposits has a completely different meaning when it When the central bank lends directly to the market, it reaches the high levels found in some developing countries. no longer has a supervisory role to play. "Once again, an important difference needs to be taken into account when analyzing ratios in developing countries relative to those in industrial countries. In industrial economies, a high loan- IREvenin countries that supervise banks through other agen- to-asset ratio can imply an unsound bank. However, the ex- cies, such as an independent deposit insurance system, a banking tremely low ratios found in some developing countries cannot be commission, or the ministry of finance, the central bank must interpreted the same way as the moderately low loan-to-asset ra- have access to bank supervisors if it is to fulfill its role as lender tios of sound banks in industrial economies. The main reason is of last resort. the quality of nonloan assets held by banks. In industrial coun- I9The evidence indicates that, with the exception of Peru, tries, banks frequently hold bonds from highly rated companies banks held a high ratio of cash to assets only when they were suh- or other high-quality assets. ject to high reserve requirements. The experience of the central bank in acting as su- credit to a program that provides bankers with incen- pervisor rather than as direct lender is a crucial de- tives to work with borrowers so that nonperforming terminant of how a banking system survives a sys- or restructured loans return to performing status. temwide banking crisis, as occurred in many Latin Central banks have devised several strategies to deal American countries in the 1980s. In a major crisis, with this problem, and several of these will be dis- the central bank has an important role in reviving cussed and evaluated in Section 111. confidence in the system, and, to perform this role, it If a central bank lacks credibility in its lending usually finds itself in a position in which it must lend policies, it may find it desirable to signal to the mar- funds to banks with severe credit problems. To pre- ket that it is willing to subject itself to constraints that vent this credit expansion from being viewed as in- encourage prudence. One such constraint is to permit flationary, the central bank must establish a lending the banking system to freely offer loans and deposits program to banks that creates incentives for bankers denominated in a hard foreign currency, such as the to work with their borrowers to improve their busi- U.S. dollar, a policy known as dollarization. Because nesses rather than to provide new funds for projects the central bank cannot extend credit in the foreign with no economic value. In other words, investors currency without borrowing that currency in the in- must be convinced that the credit created by the cen- ternational marketplace, it is more likely to lend tral bank will lead ultimately to real revenue gains. funds less carelessly than it would in the domestic The credit created will lead to real revenue gains currency, which it can create. If banks can maintain if bank stockholders have incentives to help borrow- high loan and low cash ratios in their foreign cur- ers with nonperforming loans regain solvency. If rency portfolios, they will have the proper incentives central bank credit is provided to banks at below- to monitor their borrowers. For example, dollariza- market interest rates and banks view this credit as an tion of a banking system may not be detrimental to an unlimited source of funds, they might use the low in- economy if it is the only means by which loans can terest rate credit to cover the unpaid interest pay- be extended in a disciplined manner. ments on nonperforming loans-which might even The next section describes how the strength of the restore the spread they enjoyed when they funded banking payments franchise and the quality of cen- through the market. This policy, however, would do tral bank leadership were important determinants in nothing to create future real revenue gains. In fact, it quickly restoring public confidence in the financial would only lead to continuous central bank losses. systems in the five Latin American economies after Instead, the central bank must tie its subsidized the onset of the debt crisis in the early 1980s. Ill FranchiseValue of Banksand Resolution of Banking Crises: 1982-90 t is well documented that the challenges faced by I In contrast, banking regulators in Argentina, Mex- policymakers in Latin America following the out- ico, and Peru, in their attempts to solve the crisis, re- break of the debt crisis included not only the correc- moved authority from bankers and substituted the tion of the countries' macroeconomic imbalances but credit judgment of the central bank or the govern- also the restoration of their domestic banking sys- ment directly. These policies also eventually led to tems.20 It is now fully recognized that the serious credit expansion, which proved to be highly infla- problems faced by the banking systems in a number tionary because investors did not believe the credit of countries in the region placed additional con- created would be repaid in real terms. straints on the effectiveness and sustainability of the stabilization programs implemented in the years im- mediately after the onset of the debt crisis." EvaluatingBanks' Strength at the This section draws on the characterization of the Onset of the Debt Crisis role of banks in developing countries presented in Following a period that started in the late 1970s Section I1 to analyze how the authorities and the characterized by large inflows of foreign capital,23 banking sectors in Argentina, Chile, Colombia, by the end of 1982 and continuing during 1983-84, Mexico, and Peru responded to the financial difficul- all five countries under consideration experienced ties during the early and mid-1980s.22 It shows that large outflows of capital. These outflows reflected the state of the franchise value of banks-as defined residents' and foreigners' deteriorated perceptions in Section 11-at the inception of the debt crisis was regarding the creditworthiness of borrowers in these a central element in determining whether the author- e c o n ~ m i e sThe~turnaround of voluntary capital in- . ~ ities in each of the countries followed a disciplined flows experienced by each country since 1982 was set of policies in managing the banking problems. accompanied by a negative overall balance of pay- The main conclusions derived from this section ments (Chart I). The loss of international creditwor- are obtained by grouping the sample countries ac- thiness experienced by these economies called for a cording to the strength of their banking franchise at domestic policy response-each country had to ei- the inception of the debt crisis and by making rela- ther put into place policies that restored the confi- tive comparisons among those countries. In sum- dence of the international financial community or mary, the analysis will show that, because the bank- generate sufficient cash flow on its trade account to ing systems in Chile and Colombia had relatively compensate for the outflow of capital.25 strong franchises, bank supervisors and bankers were able to respond to the crisis with a credible pro- 23For a discussion of the factors that expiain the evolution of gram to restore confidence in the banking system. foreign capital flows before and during the debt crisis in several Incentives were put into place to enable bankers and Latin American countries, see Rojas-Suircz (1991). A compari- their shareholders to gain from salvaging value from son hetween the capital inflows problem in the 1970s and early 1990s is presented in Calvo, Leiderman, and Reinhart (1992). bad credits. The programs in both these countries re- Z4Adetailed analysis of the capital flight problem experienced quired a substantial increase in credit, but because by Latin American countries during this period is contained in credit creation was associated with a program to re- Rojas-Suarez (1991). store stability, it was noninflationary. ?5Confidence of the international community in the financial performance of most Latin American countries was not restored during most of the 1980s. Indeed, the large capital flight that fol- - - ~ 20Fora review of the issues and experiences in a number of de- lowed the onset of the debt crisis, as well as the deceleration of veloping countries that faced banking crises in the early 1980s, external loans, was accompanied by a sharp increase in the re- see Sundararajan and BaliRo (1991). source balancedefined as net exports of goods and nonfactor 21Thisissue is discussed furthcr in Baliiio (1991). servicesduring 1983-86; that is, for the Latin American coun- '2As the discussion below will show. the diversity of experi- tries, net transfers of resources abroad were a direct cost associ- ences across these countries is enough to guarantee an appropri- ated with their severely reduced access to external credit. Further ate represcntation of the banking difficulties in Latin America. discussion of these issues appears in Rojas-Suirez (1991). - - Overall Bolonce Current Account Chile 10 Colombia Source: IMF. World Economic Outlook The lack of confidence in borrowers' ability to ratio is relatively low-that is, banking systems repay their loans had an adverse impact on the bank- where the franchise value is high-have the greatest ing systems in these five countries-albeit in differ- experience in establishing credible loan workout ent proportions. In accordance with the discussion in programs in periods of systemic bank crises. In Section 11, banking systems in which the loan-to- banking systems where the franchise is low, the ten- asset ratio is relatively high and the cash-to-deposit dency in a crisis is to provide new credit to borrowers A well-known example of how the behavior of a lender of last resort can contribute to the undermin- ing of a bank franchise is the U.S. savings and loan crisis. U.S. savings and loans offer government- insured deposits, mostly to consumers, similar to those offered by banks. In contrast to commercial CashAssets Loans to banks, which hold a portfolio of short-term loans, Country to Deposits Assets savings and loans hold mostly long-term home mortgages, which, until the late 1980s, were illiquid Argentina securities. Chile Colombia In the 1970s, the deposits offered by savings and Mexico loans were short term and subject to interest rate Peru ceilings. When market interest rates rose substan- tially above the ceilings in an inflationary environ- Sources: Central Bank of Argentina; Chile, Superintendentla ment, consumers began to withdraw their deposits. de Bancos e lnstituciones Financieras, lnfarmacion F~nanciero; Colombia. Banco de la Republics; Mexico. Comision Nacional Savings and loans could not sell their mortgage as- Bancaria: Peru, Superintendencia de Banca y Seguros; and IMF sets, which were fixed-interest-rate securities, be- staff estimates. cause the market was limited. Even if they had been IData correspond to 1981. able to sell, however, the losses sustained from the increase in interest rates would have wiped out the capital of many institutions. During this phase of the savings and loan crisis, policymakers tried to nurse failing savings and in arrears on their payments, without establishing loans along with marginal extensions of credit adequate controls, in the expectation that these bor- through the Federal Home Loan Bank Board, a kind rowers will be able to salvage a bad business with of central bank for savings and loans, and through the new funds. Such credit expansion has often led regulatory forbearance on accounting issues. De- to greater credit problems in the near future. posit interest rates were deregulated, which stopped At the end of 1982, the banking systems of the the outflow of deposits but did not improve the five countries fell into two distinct groups. The first profit situation as interest expenses exceeded inter- group, consisting of Argentina, Mexico, and Peru, est income. The asset powers of savings and loans had comparatively weak franchises, and the second were also liberalized. For example, federally char- group, consisting of Chile and Colombia, had com- tered savings and loans were permitted to make di- paratively strong franchises as measured by the rect investments in real estate as well as to hold ratios described above. The first group had cash-to- mortgages. Giving the savings and loans the right to deposit ratios ranging from 55 percent in Peru and bid for insured deposits paying market interest rates 65 percent in Mexico to over 75 percent in Ar- and liberalized powers resulted in increased risk gentina. In contrast, in Chile and Colombia, the ratio taking, and, with the decline in the property market was about 21 percent. Although confirming the clas- in the late 1980s, the savings and loan industry was sification of countries described above, the differ- in critical condition. ences in the loan-to-asset ratios between the two At the end of the 1980s, marginal aid policies groups were not quite as extreme. In the first group were abandoned, and large sums of public money of countries, the ratio was about 45 percent. In Chile, were committed. The commitment of public money it was 63 percent and in Colombia about 59 percent came with a new policy that made shareholders (Table 6). partners with a government agency to rescue the in- Incentives to monitor the liquidity of bank bor- dustry. The government agency, the Resolution Trust rowers-as measured by the above ratios-deter- Company, took over the assets of failing savings and mine the quality of bank supervision as well. In loans. It then sold the assets through a bidding banking systems where borrower liquidity deter- process. Thus, prospective buyers bid on the assets mines the liquidity of the banking system, the cen- based on their calculations about future movements tral bank must monitor bank lending decisions be- in the market as well as on their ability to manage cause it may often be called upon for liquidity defaulted loans back to profitability. The previous assistance. To provide this assistance without under- policy of providing distressed savings and loans ac- mining the franchise value of the banking system, it cess to lender-of-last-resort funding on a continuous must be familiar with how its member banks make basis often committed regulators to lend money to their lending decisions and how they monitor their institutions that had no capital. Thus, owners had no borrowers. incentive to use the new money wisely because they Table 7. FSsd Balanceand Inflation Argentina Chile Colombia Mexico Peru -- Fiscal Fiscal Fiscal Fiscal Fiscal Year balance' lnflation balance lnflation balance2 lnflation balance1 lnflation balance lnflation 1980 -8.0 100.8 - 35.1 -2.4 26.5 -7.6 26.4 4 . 4 59.1 1981 -16.7 104.5 0.8 19.7 -5.9 27.5 -14.6 27.9 4 . 4 75.4 1982 -18.0 164.8 -3.4 9.9 -7.6 24.5 -17.8 58.9 -9.1 64.4 1983 -17.9 343.8 -3.0 27.3 -7.6 19.8 4 . 5 101.8 -11.8 I1 1.2 1984 -7.6 626.7 -4.3 19.9 4 . 3 16.1 -6.4 65.5 4 . 2 1 10.2 1985 -5.1 672.1 -2.6 30.7 -3.6 24.0 4 . 7 57.7 4 . 3 163.4 1986 -2.5 90.1 -1.9 19.5 -0.3 18.9 -14.8 86.2 -5.9 77.9 1987 -5.6 131.3 -0.4 19.9 -2.0 23.3 -I5.0 131.8 -7.9 85.8 1988 -3.0 343.0 2.5 14.7 -1.8 28.1 -1 1.6 1 14.2 -9.2 667.0 1989 -7.3 3,079.8 5.3 17.0 -2.0 25.8 -5.2 20.0 -7.9 3,398.7 1990 -2.4 2,3 14.0 3.8 26.0 -0.7 29.1 -3.6 26.7 -5.4 7.48 1.7 1991 -1.3 171.7 2.3 21.8 -1.0 30.4 -0.4 22.7 -2.6 409.5 1992 0.5 24.9 3.2 15.4 -0.9 27.0 1.5 15.5 4 . 3 73.5 Sources:Argentina, Ministry of Economy. and Central Bank of Argentina; Central Bank of Chile; Colomb~a,Banco de la Republics; Bank of Mexico; Central Reserve Bank of Peru: and International Monetary Fund, InternationalFinonciolStatistics, various issues. Note: Fiscal balance(overall surplus (+) or deficit (-) of the nonfinancialpublic sector) data are expressedas apercentageof GDP: inflationfigures are shown in percenrages. lnflation is calculatedusing the following formula: ((P[t] - P[t * - I])lP[t - I]) 100, where P is the consumer price index (CPI) (IFS line item 64). The CPI is calculatedbased on the average price prevailing in the economy during the year. Ilncludes quasi-fiscal operations of the central bank. ?Refersto the overall surplus (+) or deficit (-) of the consolidatedpublic sector. 'From 1984 onward, the data represent the economic balance;that is, financial intermediation is not included. had nothing at risk. In contrast, the new policy com- five countries according to the strength of their mitted government money immediately and then banking franchise matched the relative soundness of turned the assets over to new owners, who were re- their macroeconomic systems in the period immedi- quired to supply new capital to the project. The new ate before the debt crisis. capital commitment was large enough to give the Based on the fiscal policy stance and the level of new owners an incentive to manage the assets back inflation, Chile had the strongest macroeconomic to profitability. environment among the countries under considera- Successful bailout programs of banking systems tion in the early 1980s (Table 7). Consistent with in Latin America paralleled the second phase of the having the strongest fiscal stance, Chile's average U.S. savings and loans crisis, while unsuccessful rate of inflation was the lowest among the countries bailouts paralleled events in the first stage. That is, in the sample. Colombia followed Chile in the bailout programs of banking systems in Latin Amer- ordering of countries. With this criterion in mind, ica were successful only when the supply of central Argentina had the weakest macroeconomic condi- bank credit to distressed institutions was made con- tion among the countries in the group: even two ditional on a realistic appraisal of the loan portfolio. years before the outbreak of the debt crisis. the Under such programs, only borrowers with a reason- average inflation rate was running at more than able possibility of returning to solvency could be eli- 100 percent. Although the rate of inflation in gible for additional loans. Mexico did not accelerate until 1982-and was similar to that experienced in Colombia in 1980-81-the fiscal deficits (as ratios of GDP) in Macroeconomic Environment and the Mexico during the early 1980s resembled those in Franchisevalue of Banking Systems Argentina. Finally, Peru's fiscal deficit and infla- tion, although less severe than in Argentina. deteri- Closely related to the ability of the central bank to orated during the early 1980s. Indeed, by 1981, the promote discipline in financial markets is the state of Peruvian inflation rate was closer to ~rgkntina'sin- the overall macroeconomic environment. It is there- flation rate than to that of any of the other countries fore natural to question whether the ordering of the in the sample. Chart 2. Real InterestRates (An& ovemge. inpUrW 50 Argentina ---+- -250 - 3 0 0 " " " " 1 " " ' 1980 1982 1984 1986 1988 1990 1992 40 Chile Colombia 35 30 - - - c11111-11111 O;98; ;98;*8 ;99; I980 1982 1984 I986 I988 1990 1999) 30 50 Mexico Peru 20 0 10- -50 -100 v -150 ------- Sources: Argentma. Ministry of Economy,and Central Bank ofArgentina; Central Bank of Chile; Colombia, Bnco de la Republics; Bank of Mexico;Central Reserve Bank of Peru;and IMF sraff estimates. Thus, the ordering of countries in terms of their 1980s, reflecting interest rate ceilings imposed by fiscal stances and inflation performances during the the authorities26 (Chart 2). To a large extent, these early 1980s matched closely the ordering of the fran- - chise value in these countries. Moreover, in Ar- 26Although interest rates were liberalized in Argentina in 1978, gentin% Mexico, and Peru, ex post real interest rates controls on bank deposits were reimposed in 1981. See the dis- on bank deposits were mostly negative in the early cussion in the next subsection. controls, which weakened the franchise value of The central bank response can be said to have banks, were not present in Chile and Colombia, and, solved the crisis by placing as little strain as possible as a result, ex post real interest rates remained posi- on defaulting borrowers and the most troubled tive throughout the 1980~.~' banks. Banks with relatively clean loan portfolios It is not surprising to find out that, using the two were forced to aid in the bailout of troubled banks by alternative criteria, the ordering of countries matches. holding high reserves so that the central bank could After all, a stable macroeconomic environment and a obtain resources to lend. Depositors were forced to stable financial system are complementary. accept a depreciation of their deposits in real terms. Because responsibility was not placed on the shoul- ders of the parties responsible for the crisis, crisis How the Systems Respondedto resolution weakened the franchise value of the bank- Immediate Crisis: 1982-85 ing system. In contrast to this view, the central bank reaction to This subsection considers whether the banking reregulate suggests that the authorities believed that systems in the countries classified as having the the banking crisis was caused by deregulation in the highest franchise value among the countries in the late 1970s-that is, by the reduction of reserve re- sample coped with banking crises in a more disci- quirements and the resulting increase in bank loans. plined manner than those countries classified as hav- Although it is probably true that the rapid deregula- ing the lowest franchise value. At the beginning of tion of a banking system run by inexperienced bankers the international debt crisis in 1982,the banking sys- invited the banking crisis, the resolution of the crisis tems in Argentina and Mexico were already deeply did nothing to educate bankers in how to deal with mired in a credit crisis. In Chile and Colombia, the credit crises, which was the real problem facing the banking systems were about to face the conse- Argentine banking system. As such, reregulation was quences of lender concerns that borrowers could not a step backward that was to lead to much greater trou- make good on their debts. In Peru, the crisis was de- ble later on than that caused by deregulation. layed until the middle of the decade, but when it ar- An additional problem confronting the Argentine rived, it was a crisis of major proportions. banking system was the explosion of credit to the In Argentina. the banking system, which had been central government in the early 1980s, most of subject to interest rate ceilings, high reserve require- which was lent through the central bank and, again, ments, and prescribed lending policies, was rapidly financed by high reserve requirements on the banks. deregulated in 1978. Without appropriate supervi- This expansion of credit had two debilitating effects sion of credit and with bankers lacking the experi- on the banking system. First, it ensured that high re- ence to price risks properly, the banking system ex- serve requirements would persist beyond the imme- panded excessively and, by the early 1980s, suffered diate banking crisis and, second, it the central from severe credit problems. The central bank re- bank in the role of direct lender to the government. sponded to the crisis by reimposing stiff reserve re- As a direct lender and a supervisor, the central bank quirements and interest rate ceilings on deposits. was faced with a conflict of interest that undermined Real interest rates on deposits became negative (see sound bank supervision. Chart 2). and reserves in the central bank increased In Mexico, the banking crisis in 1982 occurred in to over 70 percent of deposits in 1982, compared an environment of sharply deteriorating economic with 12 percent in 1981. conditions (see Table 7). The crisis could not be The central bank followed two basic policies: it blamed on rapid credit expansion following a de- subjected deposit interest rates to a ceiling that per- cline in reserve requirements, because the ratio of mitted banks to reduce interest rates to borrowers to cash assets to deposits was extremely high in Mex- a manageable level, and it increased reserve require- ico before the onset of the crisis. However. because ments, which, in turn, provided the central bank with the central bank had used the high reserve require- resources to lend to the most troubled institutions. In ments to lend to the government, it had few re- 1982, central bank credit to the financial system sources to lend to troubled banks (see Table A2). equaled over 21 percent of GDP (see Table Al). The lack of resources could have led to a disci- plined approach to resolving the banking crisis. In- >'In Chile, during the late 1970s and early 1980s, ex post real stead, the central bank and the government engaged interest rates increased drastically and were accompanied by a in a policy of reducing the real burden of borrower significant widening of the spread between domestic interest rates debt and forcing depositors to absorb some of these (adjusted for exchange rate changes) and comparable foreign in- losses through a reduction in the real value of de- terest rates. The review of the Chilean experience suggests that posits. The latter action was achieved through a these high domeslic rates emerged when the domestic financial system was liberalized in conjunction with the opening of the combination of policies: forced conversion of for- capital account (see Mathieson and Rojas-Suirez, 1993). eign-currency-denominated deposits at unfavorable exchange rates and negative real interest rates on deficits,they were unable to respond in a disciplined peso-denominated deposits. The real value of loans manner. This generated a crisis of major proportions was reduced through the conversion of foreign cur- that will be discussed in the next subsection. rency loans to pesos at an exchange rate that over- In 1982, the banking authorities eliminated mar- stated the value of the peso relative to the dollar. ginal reserve requirements on domestic currency de- Thus, as in Argentina, one of the parties responsible posits, although average reserve requirements re- for the crisis, the borrower, was given a subsidy, and mained extraordinarily high-at more than 50 the party with less responsibility, the depositor, was percent of total deposits. They also simplified the in- forced to accept a loss. terest ceilings imposed on domestic currency de- In addition, the banks were nationalized, which, if posits. One of the primary motivations of the author- properly managed, could have had a beneficial effect ities in reducing reserve requirements was to expand on franchise value because stockholders were forced domestic bank lending activity. A central bank pol- to absorb losses. Such a policy, however, would icy of paying attractive interest rates on reserves, have been beneficial to the franchise only if the new however, encouraged banks to continue to hold high owners, whether the government or private entities. ratios of cash to deposits despite the reduction in re- had purchased the banks after the assets had been serve requirements. Moreover, banks were further properly evaluated. In addition, new owners should encouraged to hold high reserves at the central bank have been required to place substantial new equity because the uncertain macroeconomic environment into the system with a clear statement of policy that increased the risk of lending. the equity would be lost if the new managers could The reforms, however, did not lead to a substantial not successfully manage the assets after reevalua- increase in domestic currency deposits. In fact, as tion. That this was not done is suggested by the fact real interest rates on domestic currency deposits that the capital-to-asset ratio of Mexican banks re- were substantially negative in the early 1980s (see mained very low until the late 1980s (see Chart 3). Chart 2), foreign currency deposits increased from The methods used by the Mexican banking au- 46 percent of bank deposits in 1982 to 61 percent by thorities to resolve the banking crisis required fewer 1984. Because these deposits had high reserve re- uses of public funds than in Argentina. These meth- quirements, an increasing share of resources ended ods, however, were philosophically similar: they up in the hands of the central bank. The central bank were designed to spread the losses in an acceptable lent some of these funds to the government-owned manner rather than to improve the franchise. Thus, development banks but, in 1984, invested them pri- the responses of the authorities in these two coun- marily in foreign reserve assets (see Table A3). The tries parallel those of the U.S. authorities in the first large amount of liquid assets in the banking system phase of the savings and loan crisis. and the central bank precluded a major banking cri- In contrast to Argentina and Mexico, Peru did not sis in the early 1980s, but crisis loomed on the hori- experience a major banking crisis in the early zon as a government came to power that was willing 1980s.28 Peru had an extremely high ratio of cash to to spend these resources. In fact, the unwillingness deposits in 1982 and, in contrast to Argentina, had of the authorities to encourage banks to create do- not experimented with reserve requirement reduc- mestic credit may have contributed to the support of tions in the 1970s. As a result, bank loans to the pri- a new government willing to spend resources do- vate sector were small relative to GDP (Table A3), mestically. The problem was that this spending was and, in 1982, that ratio was the lowest among the funneled through a financial system without a fran- five countries under consideration. chise value, generating a crisis. This crisis is dis- Not extending bank loans may appear to be a very cussed in the next subsection. attractive way to prevent a banking crisis; if bank Chile and Colombia also experienced banking lending is limited, few loans can become nonper- crises in 1982, but, in contrast to Argentina and forming. Banks that were extending only a few Mexico, the central bank authorities attempted to loans, however, lost their credit evaluation skills, focus responsibility for problems on the sharehold- and, in the late 1980s, when the central bank ex- ers of the most distressed banks rather than on the panded credit to finance compounding fiscal system as a whole. In Chile, the rescue effort got off to a somewhat unpromising start. The central bank made extensive 2RThe crisis was confined to the collapse of two banks in early 1983.This made remaining banks very cautious in extending new credit available to the banking system and to de- loans. The impact of bank failures and bank loan loss problems faulted borrowers without putting proper controls in on bank balance sheet quality is difficult to discern from capital- place. Net domestic credit of the financial system in- to-asset ratios because the accounting data appear unreliable. creased from 54 percent of GDPin 1982to more than Thus, the large fluctuations in the capital-to-asset ratios during 73 percent of GDP in 1983 (Table A4). Much of this 198C90 shown in Chart 3 largely reflect accounting procedures rather than true changes in real capital. increase was due to central bank lending to financial Chart 3 (kmnt) 30 8 Argentina Chile 8 Mexico Sources: Central Bank ofArgentina; Ch~le,Superintendencia de Bancos e lnstituciones Financieras. Informa- cion Financiers: Colombia. Banco de la Republics; Mexico. Comision Nacional Bancaria;Peru. Super~ntendencia de Banco y Seguros;and IMF staff estimates. intermediaries, which rose from less than 7 percent In rescheduling nonperforming foreign currency of GDP in 1982 to more than 16 percent of GDP in loans, the central bank offered borrowers favorable 1983. In addition, the central bank purchased nonper- exchange rates, which created severe losses at the forming loans outright from the banks.29 central bank. Borrowers presented pesos to the cen- tral bank to repay their loans, and the central bank lyFor a thorough description of the bailout procedures in Chile, exchanged these at a dollar exchange rate that was see Morris and others (1990) and Velasco (1991). lower than that prevailing in the open market. In fact, the losses became so severe that the govern- posits with indexed principal. During the first few ment had to recapitalize the central bank. years of the program, private liability holders were The authorities quickly put in place policies that reluctant to supply such a contract. Hence, the au- attempted to provide incentives for banks to work thorities provided banks with a line of credit with in- with defaulting borrowers to improve loan quality. dexed principal. In 1984, the central bank bought the banks' nonper- Not all banks could be rescued by the above forming foreign currency assets with cash. The mechanisms, however. The portfolio condition of banks had to use this cash to repay the central bank some banks had deteriorated so much that they had for credit drawn on the lines made available in 1982. to be recapitalized before they could sell loans to the As a result of this transaction, central bank loans to central bank. The government took over the dis- the banking system declined (Table A4). tressed banks. writing down the value of stockholder - Under the initial agreement of 1984, the banks equity by marking assets to market, as manifested in were forced to buy back the bad loan portfolio over a the sharp drop in equity-to-asset ratios in 1984 (see ten-year period at the original face value of the loan Chart 3).3' But. unlike in Mexico. the central bank plus accumulated unpaid interest, rather than at a offered'the distressed banks for sale to the private value determined by how the loan was performing. sector.32 Deals were structured so that the new own- The repayment would be made through a provision ers contributed sufficient capital to give them an in- item in the income statement that flowed into an ac- centive to manage the assets prudently. With new count on the liability side of the balance sheet, re- ownership in place, the resolved banks were permit- ferred to as "a subordinated obligation to the central ted to participate in the central bank's distressed loan bank." The banks, however, were not paying market program. The new stockholders together with the interest rates on their obligations to repay the central central bank became de facto preferred shareholders: bank.30 dividends had to be paid to them before dividends As an important component of the original agree- could accrue to the original stockholders. This pol- ment, the banks were placed in charge of administer- icy had two beneficial effects: it demonstrated to an ing the loan portfolio they had sold to the central identified group of shareholders that if banks fail, bank, which meant that they had responsibility for equity holders lose money. It also put a new manage- collecting loan payments and encouraging borrow- ment in place with capital to lose, giving them the ers to remain current on their payments. Banks had appropriate incentives to manage loan portfolios. an incentive to perform this task well, because, Despite the merits of the initial rescue plan, a under the original terms of their agreement with the number of banks were unable to maintain their central bank, they had to buy loans back at the origi- scheduled repayments to the central bank. Therefore, nal face value. With this policy, the central bank at- in 1989, the ten-year payback period was extended tempted to play the role of supervisor of banks rather indefinitely. Under the new agreement, banks are re- than of manager of nonperforming loans to the pri- quired to repurchase the subordinated debt with their vate sector. flow of profits (if any) minus the dividends paid out The authorities encouraged banks to index the to prefirred private stockholders until their debt principal of domestic currency loans to inflation to obligations are fulfilled.33 For some banks, which reduce the cash-flow burden on borrowers during have not been able to cover even accrued interest, the relatively high inflation experienced in the mid- this has resulted in negative amortization of their 1980s. During high inflation, the real principal of debts to the central bank. By the end of 1993, the nonindexed debt contracts declines. The high nomi- banking system's obligation to the central bank is nal interest rate compensates the lender for this loss, estimated to have reached about US$4 billion, or but it also forces the borrower to pay off real princi- 10percent of GDP. pal more rapidly than in a noninflationary environ- ment. To offer an indexed loan contract, banks had to find liability holders who were willing to hold de- 3IMarked to market means that assets are valued at current market prices rather than at book values. In contrast, in Argentina capital-to-asset ratios actually increased during the banking cri- 30Because reserve requirements in Chile were so low, the cen- sis, indicating either that shareholders gained from the rescue tral bank had very few funds with which to aid the banks. Most of package or that nonperforming loans were not marked at market its resources came from an expansion of borrowing from over- values. seas, which increased from less than 2 percent of GDP in 1982 to 32The central bank had to provide monetary enhancements to more than 27 percent by 1985. Much of this increase was due to these deals (tax exemptions were also granted). This is similar, the depreciation of the Chilean peso; however, in U.S. dollar however, to the experiences of bank regulators in a number of in- terms, foreign borrowing by the financial system increased from dustrial countries when dealing with resolution of banking crises. almost $6 billion in 1982 to almost $10 billion in 1985.The addi- ))These are the new shareholders, the so-called capitalistas tional funds were made available through rescheduling agree- populares, who bought shares when banks were recapitalized in ments with foreign lenders. the mid-1980s. During the 1970s and early 1980s, economic con- In contrast, the banks in Argentina and Peru faced ditions in Colombia remained heavily dependent on severe competition from the central bank as direct coffee exports. To a large extent, the banking crisis lender and from government-related development of the early 1980s was triggered by the sudden end banks. In Mexico, by the middle of the decade, of the coffee boom of the late 1970s and the related banks were lending most of their funds to the gov- deterioration of the fiscal position. As the coffee ernment and government-related enterprises. In all boom ended, Colombia experienced capital flight as three countries, the banks were not able to exercise economic agents adjusted their portfolios to the ad- their monitoring function over borrower liquidity, verse external shock. The capital flight resulted in and the central bank could not function as an arm's- only a mild banking crisis because much of the for- length regulator. This lack of independence of cen- eign borrowing was concentrated directly on the bal- tral banks from expansionary fiscal policies led to ance sheet of the government rather than on the bal- high inflation by the end of the decade. ance sheets of financial institutions. The government was able to cover the flight by borrowing foreign re- Chile and Colombia serves from the central bank, as net central bank credit to public authorities increased from 4.2 per- In Chile,by 1989,the central bank's medium- and cent of GDP in 1982 to over 9 percent of GDP in long-term foreign liabilities had declined from al- 1984, causing central bank foreign reserves to de- most 28 percent of GDP in 1985 to 10.5 percent of cline to 3.4 percent of GDP (see Table A5). GDP, and central bank net domestic credit had fallen Because of the accumulation of foreign exchange from 32 percent of GDP to 3.5 percent. Net interna- reserves during the coffee boom, the Colombian tional reserves were growing strongly relative to central bank had a larger amount of funds available GDP (see Table A4). to aid troubled banks in the early 1980s than was After 1985, bank capital ratios also recovered- available in Chile. In 1982, its net international re- rising from 4 percent of assets to about 6 percent of serves equaled more than 10 percent of GDP com- assets by 1989, implying that bank earnings were pared with about 5.5 percent at the Chilean central growing (see Chart 3). Further indication that the bank (see Tables A4 and A5). The impact of capital franchise value of banks was improving was the in- flight on the banking system is evident from the in- crease in the loan-to-asset ratio, which had fallen creased central bank lending to banks, which in- when the central bank removed bad loans from the creased from -3.3 percent of GDP in 1982to -1 per- banks' balance sheets. In assessing the effectiveness cent of GDP in 1984." A few small banks failed, and of the Chilean solution to the banking crisis of the banks' capital-to-asset ratios dropped sharply, indi- early 1980s, it is necessary to take into account both cating that shareholders were forced to sustain losses the costs and the benefits of the ~rocedureschosen. from nonperforming credits. The full severity of the On the negative side, because the interest rate on crisis was not felt until the mid-1980s,and its resolu- banks' debt has remained below that paid on central tion will be discussed in the next subsection. bank paper, the central bank has sustained losses Nonetheless, unlike in Argentina, Mexico, and Peru, throughout the mid- 1980s and early 1990s. On the real interest rates in Colombia remained positive in positive side, however, the problem was managed in the early 1980s, indicating that losses were focused an orderly way and did not result in a sustained dete- on the stockholders of failed institutions. rioration of the fiscal stance or a rapid and sustained acceleration of inflati0n.~5In addition, only deposi- tors in the failed institutions were forced to absorb Crisis Resolution: 1985-90 some of the losses of those institutions. In contrast. depositors of both good and bad banks in Argentina By the middle of the decade, Chilean banks were and Mexico were forced to absorb losses under poli- recovering, whereas Colombian banks were experi- cies encouraging negative real interest rates. In encing a more severe crisis in 1985 and 1986 than in Chile, real interest rates remained positive, and there the early 1980s. Nevertheless, the second Colom- was a relatively clear policy that only those con- bian banking crisis was quickly resolved. By the end nected with a failed institution would bear the bur- of the decade, capital-to-asset ratios at the banks in den of the failure (see Chart 2). both countries had recovered to their precrisis level, Indeed, to a large extent, the management of the and the central banks' role as lender to the banking banking crisis in Chile resembles that in several in- system was declining. dustrial countries. Even in countries where a banking 351tneeds to be recognized, however, that the sustained losses 34Thenegative asset position indicates that the banks were net of the central bank have acted as a constraint to reduce the creditors of the central bank. Chilean inflation rate to industrial country levels. crisis came to an end relatively quickly, such as in cia1 banks and finance companies fell sharply (see Sweden in the early 1990s, the resolution of the fi- Chart 3). Rather, the reduction in central bank lend- nancial difficulties involved large costs to the tax- ing to banks signaled a new method for dealing with payers.36 In the United States, with the resolution of distressed banks. the savings and loan crisis, the accumulated loss suf- The Colombian authorities established a deposit fered by U.S. taxpayers has, to date, amounted to insurance fund to lend to banks with distressed loan about 3 percent of GDP. portfolios and to resolve banks that had failed. The In 1992. banks with debt outstanding to the central ., insurance system received its funding largely from the bank had net income before dividends and payments government and from extraordinary revenues made to the central bank of about CH$110 billion, which, available by a sharp rise in coffee prices in the mid- discounted at the cost of banks' liabilities, equaled 1980s. The insurance fund ultimately purchased the about CH$800 billion.37With total subordinated debt largest banks in the banking system but originally had outstanding amounting to about CH$1.5 trillion, this problems selling them, because it could not find buy- simple approximation suggests that, using 1992data, ers that met the legal criteria established to prevent about 45 percent of the subordinated debt is uncol- industrial or financial conglomerates from develop- lectible. This portion represented about 5 percent of ing40 The purpose of the legislation was to prevent 1992GDP.38This figure suggests that the cost of the concentration of credit to single borrowers and to en- Chilean banking crisis relative to GDP is higher than sure arm's-length credit decisions. Many of the bank the cost to GDP of the U.S. savings and loan crisis; failures in the 1980s were blamed on such abuses. however, it is lower than the cost incurred in resolv- In early 1994,the insurance fund successfully sold ing the banking crisis in Sweden. the largest bank-Banco de Colombia-to an indus- Although the cost of resolving the banking crisis in trial conglomerate. This sale raises the question of Chile is a substantial burden, it must be noted that, rel- the abuses that the legislation was designed to curb. ative to the original magnitude of the banking crisis, Strong supervision (such as restrictions on loans to during which central bank net domestic credit swelled related companies and limits on credit exposure to to 30 percent of GDP, the debt hangover is moderate. single borrowers), however, can alleviate many of Thus, despite the costs currently borne by the central these potential problems. If the Colombian authori- bank, theresolution of the banking crisis in Chile ties successfully complete the sale of the remaining minimized economic dislocations relative to those resolved major bank-the Banco Cafetero-and ef- suffered by other countries in the sample considered fectively strengthen the supervisory process to avoid in this paper. In addition, as a result of the experi- abuses, the Colombian rescue efforts will have rein- ences during the crisis, the Chilean bank supervisory forced the franchise value of the banking system.41 framework is now among the best in Latin America. In Colombia, in 1985, the central bank began re- Argentina, Mexico, and Peru ducing net domestic credit and accumulating net in- ternational reserves. In the same year, its net credit The banking systems in the three countries with position toward banks declined, indicating that it had relatively weak franchises all experienced an expan- reduced lending to the banks. This event did not. " sion of credit to government-related institutions rela- however, signal the resolution of the banking crisis tive to credit provided to the private sector through as it did in Chile; in fact, nonperforming loans at private financial institutions in the mid- to late commercial banks equaled 500 percent of capital 1980s. Although the details of how this credit was and 268 percent of capital at all financial institu- directed to the public sector differed by country, the tions in 1985.39 Capital-to-asset ratios at commer- outcome was the same in all countries. The projects financed through money creation did not generate a corresponding expansion in economic activity. As a 3% Sweden. by the end of 1992, capital injections and govern- result, severe inflation followed, and holders of as- ment guarantees to support troubled banks amounted to 6.4 per- cent of GDP. sets denominated in domestic currency experienced 37The assumed discount factor equals 13.5 percent, which is the a loss in their real wealth. In all three countries, the average nominal cost of liabilities of the national banks in 1992. central bank failed to act as a bank supervisor be- 38Actual payments to the central bank were CH$78 billion in cause it had a conflict of interest as a direct lender, 1992 because only a proportion of banks' net income was re- quired to be used as payments to the central bank. However, in and the banks lent funds to government institutions calculating the amount of debt that is potentially serviceable, it is appropriate to consider the entire cash flow available for debt ser- vice. Using the lower figure would imply that the unserviceable J T h e sale of the Banco de Bogota was criticized as providing a debt equals about 6 percent of GDP. large subsidy to purchasers. 39The Colombian banking system is made up of commercial 41The Banco Central Hipotecario was sold to the Social Secu- banks, savings banks, development finance companies, and trade rity Institute. The state will retain the ownership of the Banco finance companies. Popular. that were too powerful to be monitored by normal net domestic bank credit flowed to the government, bank credit policies. Thus, the checks and balances compared with 35 percent in 1982. that restrain risk taking in well-supervised banking Net domestic credit of the financial system ex- systems did not exist. panded rapidly beginning in 1986, when it increased In Argentina, starting in 1986, central bank credit to over 70 percent of GDP from 52 percent the previ- to the central government expanded faster than com- ous year (see Chart 4). As in Argentina, much of the mercial bank credit to the private sector (see Chart 4). increase in financial system credit relative to GDP The credit expansion was caused primarily by a lack during 1986-87 was accompanied by a deterioration of confidence in the government's ability to repay in the exchange rate and high inflation rates, which the central bank, which forced up interest rates, re- peaked in 1987 and 1988 (Chart 5). Real interest quiring further borrowing to service the debt. At the rates on bank deposits became negative in 1987, re- same time, central bank funding to the banks also ducing the government's real burden in paying off expanded more rapidly than commercial bank credit its bank loans, which declined substantially relative to the private sector. In 1989, the net domestic assets to GDP in 1988 (see Table A2). of the financial system relative to GDP exploded, in- In Peru, net domestic credit of commercial banks creasing from less than 40 percent of GDP in 1988to relative to GDP remained low in the early part of the more than 50 percent. decade because of very high reserve requirements. The credit explosion resulted in hyperinflation This policy, however, was abandoned in the mid- and a deterioration in the exchange rate (Chart 5). 1980s when the central bank expanded credit The rapid depreciation of the currency reflected in- sharply-partly financed with bank reserves held at vestors' perceptions that the credit created by the the central bank. Net credit of the central bank ex- central bank, both directly and through the banks, panded from -4.7 percent of GDP in 1985to 1.4per- could not be paid off in real terms. cent of GDP in 1986, and a large amount of foreign Extreme negative real interest rates on domestic exchange reserves was lost. The central bank contin- currency deposits accompanied hyperinflation (see ued to expand credit in 1987 and 1988, primarily fi- Chart 2). In addition, the central bank effectively nanced from arrears on foreign debt and increased li- confiscated domestic currency time deposits by forc- abilities to the private sector (Table A3). In 1988, ing depositors to exchange their deposits for dollar central bank credit equaled over 40 percent of net bonds worth about half the market exchange rate domestic credit, compared with less than 10 percent value of the original deposits. The reduction in inter- in 1986. Most of the credit went to the public sector, est expenses on bank liabilities permitted banks to especially the agricultural banks. renegotiate credit terms to lenders and provided The impact of the credit expansion followed a fa- funds to pay off loans from the central bank. As indi- miliar pattern. In 1988, inflation turned into hyperin- cated in Chart 3, the central bank's policy caused flation, which was accompanied by extremely nega- banks' capital-to-asset ratios to increase because, by tive real interest rates on domestic currency bank confiscating deposits, the central bank effectively deposits (see Chart 2). In addition. there was large wrote down the value of liabilities to solve banks' disintermediation in domestic currency deposits.42 credit problems. In well-functioning accounting sys- Unlike Argentina and Mexico, Peru experienced tems, the improvement in capital-to-asset ratios sig- only one major crisis because the high reserves at nals an increase in the market value of assets relative commercial banks precluded an earlier one. When to liabilities. Because this was not the case in Ar- the crisis erupted, however, it was the sole result of gentina, investors could not use the capital-to-asset central bank mismanagement. The high reserves ratio as a reliable guide to bank soundness. The Argentine experience indicates that the finan- 42Although a large number of transactions were performed using cial system lost its ability to enforce discipline on the U.S. dollar, bank deposits denominated in foreign currency-- borrowers in the mid- 1980s. This was a direct result which had been allowed since December 1977decreased signifi- cantly during August 1985-December 1990, when fully convert- of the loss of an arm's-length relationship between ible foreign currency deposits were prohibited. In particular, the central bank and the ultimate borrowers in the foreign currency deposits could be converted into intis (the domes- economy, which resulted from the inability to deal tic currency at the time) only at the official exchange rate (mea- with the banking crisis in the early 1980s in a man- sured as the number of intis per U.S.dollar), which fell sharply below the rate in the parallel (black) market. Large disintermedia- ner that would restore the credibility of the banks. tion from the banking system followed, and, rather than declining, In Me-rico, as in Argentina by the mid-1980s, the U.S.dollar transactions intensified through the development of in- government had taken an increasing share of total - - formal real and financial markets. Indeed. two forms of holding - credit, In contrast to Argentina, however, the in- U.S.dollars were clearly identified by Peruvian residents: U.S. creased flow of credit to the government occurred on currency notes that could be obtained in the well-established do- mestic black market for U.S.dollars and deposits in foreign banks the sheets the Ilationalized reflecting capital flight. Fully convertible foreign currency de- banks (see Table A2). By 1986, over 60 percent of posits were reestablished at the beginning of 1991. - Centralbonk -Centralbonk -- Commerc~olbonk -- Net domesuc cred~tto credrt to credlt to assets of the ,. publrc sector finanaol sector prrvate sector financrol system' 60 Argentina I \ 50 1 \ 120 Chile Colombia 50 I-- 100 // \ \ 0-- 1 40 80 / \ I \ \ / \--- \ A / \--- -/ 30 6 0 r - / - - - - - \\-/' 40 20 80 .Mexico 40 Peru1 Sources: Argentina, Ministry of Economy,and Central Bank ofArgentina; Central Bank of Chile; Colombia, Banco de la Repdblica;Bank of Mexico; Central Reserve Bank of Peru; and IMF staff estimates. Note: The shaded areas in Argentina, Mexico, and Peru, correspond to the high-inflation periods. 'Net domestic assets of the consolidated bankingsystem for Peru. placed considerable resources in the hands of the crisis is a classic example of a central bank abandon- central bank, which lent them without regard to ing its traditional role of examiner of bank credit de- sound lending practices. Thus, the Peruvian financial cisions for a role of undisciplined primary lender. 3500 Argentina 3000 1\ 40--- Chile Colombia 1 40 8000 Mex~co Peru 1' 120- A 7000 - 100 6000 i 5000 80 - 4000 - 60 3000 40 2000 20 1000 -- I - Source: IMF, International Financial Statistics, various issues State of the BankingSystems at the 1990scompare with their strength in the early 1980s? Beginningof the 1990s The early 1990s is an important time to evaluate the strength of the franchise for two reasons: first, both By the early 1990s, the banking crises in the coun- bankers and regulators have absorbed the lessons tries under consideration had largely come to an end. from the crisis and, second, the banking systems are How does the strength of the banking franchise in facing new challenges to their lending skills in the each of the five countries at the beginning of the form of renewed capital inflows from overseas. 80- \ 22 Argentina Chile 70 \ 21 70 ColombiaI Mexico 20 \ 60 50 19 I 40 18 \ 30 17- 20 16 10 I I" Peru 100 A Sources: Argentina. Superintendencia de Entidades Financieras y Cambiarias. Estados Contables de las Enti- dades F~nancieras,various issues, and Central Bank ofArgentina; Chile. Superintendencia de Bancos e Institu- ciones Financieras. Informocion Financiem; Colombia, Banco de la Rep~jblica;Mexico, Comision Nacional Ban- caria; Peru. Superintendencia de Banca y Seguros; and IMF staff estimates. 'Estimate for the entire banklng system (includingcommercial banks, finance corporations, and savings and loan corporations). In addition to evaluating the franchise value of keep financial funds in the domestic banking sector, banks based on the two ratios considered at the begin- as measured by the ratio of deposits to GDP.4" ning of this section, the discussion that follows exam- ines the franchise value from the perspective of finan- "As discussed in Section 11. ratios of nonperforming loans to cia1 investors-that is, the willingness of the public to total loans do not provide a good indicator of bank soundness in 65 Argentina \ Chile 60 70-- Sources: Chile. Superintendencia de Bancos e lnstituciones Financieras. Informacion Financiera: Colombia, Banco de la Republica,Revista del Bonco de lo Republico; Mexico. Comision Nacional Bancaria; and Peru. Super- intendencia de Banca y Seguros. 'Commercial banks only. lDara for total assets held by commercial banks are not available for 1982. developing countries because banks could be rolling over prob- The behavior of the franchise ratios-cash to lem loans. In this connection, recent restmcturings of the banking deposits and loans to assets--during the crisis period system that have forced banks to recognize problem loans as such in the accounting procedures may lead to the misleading conclu- and the early 1990s is displayed in Charts 6 and sion that bank difficulties have increased. 7, respectively. Overall, the ratios indicate an improvement of the franchise value of the banking the decline in the ratio of cash to deposits during systems at the beginning of the 1990s relative to that period. the early 1980s. During the crisis, however, the ra- In the three countries with weak franchises, as tios, affected by the policies used to resolve the measured by the ratios in 1982, there was consider- banking difficulties, fluctuated significantly in most able financial disintermediation during the crisis. In countries. Argentina, the deposit-to-GDP ratio fell from about In Argentina, Chile, and Mexico, the cash-to- 19 percent in the early 1980s to less than 10 percent deposit ratio declined significantly during most of during the early 1990s.The pattern in Peru was sim- the 1980s and into the early 1990s (Chart 6). In ilar. The ratio fell from about 15 percent in the early Mexico, the decline in cash assets resulted from a 1980s to 10percent in 1992.In Mexico, the ratio fell major decline in the ratio of required reserves to de- to less than 10 percent in 1988 from almost 30 per- posits, ultimately falling to zero in the early 1990s. cent in the early 1980s (Chart 8).4 In Argentina, the decline in cash assets to deposits Because of the extreme disintermediation in the also resulted from a decrease in reserve require- countries with weak franchises, means of transacting ments, which, in 1982, were 100 percent on new had to be found outside the usual deposit channels. deoosits. In Argentina and Peru, informal payments markets In Colombia, cash assets fell relative to deposits developed. Trade credit between firms increased in the mid-1980s as banks divested themselves of dramatically. In contrast, Mexico developed the li- cash assets to cover nonperforming loans. Although quidity of the short-term government bills (CETES) by 1991 this ratio had returned to its 1982 level, re- market. Banks purchased these bills from securities ductions in reserve requirements in early 1993 sug- firms under repurchase agreements, making them gest that the ratio may have declined again. In Peru, very liquid instruments that could substitute for bank cash assets increased relative to deposits throughout deposits. the 1980s. The large increase in 1988 occurred be- The deposit-to-GDP ratio in Mexico recovered in cause banks held excess reserves on which they the early 1990s, exceeding its 1982 level by 1991, earned interest. By the early 1990s, Peru had re- which is consistent with the finding that the fran- duced reserve requirements on both domestic and chise ratios improved. In the early 1990s, banks foreign currency deposits, but they remained sub- were privatized and supervision has recently been stantially higher than those of any other sample strengthened. Banks experienced some credit prob- country. Moreover, by early 1994 a portion of re- lems after privatization, which they have dealt with serve requirements on foreign currency deposits still by increasing provisions for nonperforming loans earned interest. and slowing lending growth. In addition, between The loan-to-asset ratios in Argentina and Mexico 1991 and 1992 the capital-to-assets ratio strength- improved substantially during the 1980s (Chart 7). ened after a fall during the early years of privatiza- ~ e x i c oexperienced an interruption in the im- tion (Chart 3). Investors are apparently satisfied that provement in this ratio in the late 1980s owing to the banking system is fairly well managed. Based on heavy bank investment in government securities, this confidence. Mexico, alone among the weak which was the result of a government program franchise systems in 1982, joined the relatively that allocated a significant component of banks' strong franchise group by the early 1990s. assets. In contrast, by 1992, investors in Argentina had In Chile, the loan-to-asset ratio declined during not yet recovered confidence in the banking system 1982-86 following the central bank program to despite the improvement in the ratios. This may have rescue banks, which, as discussed above, involved had several causes: the banking crisis in Argentina replacing nonperforming loans with central bank was much deeper than in Mexico, and, unlike in liabilities and removing troubled loans and liabili- Mexico, bank ownership is in the same hands as dur- ties from bank balance sheets. After 1986, as faith ing the crisis. A large segment of the banking system was restored in the banking system, the loan-to- is still in provincial hands, and these banks have, in asset ratio again began to improve. It remained the past, made loans based more on politics than on below its 1982 level, however. In Peru, as in Chile, sound banking principles. the loan-to-asset ratio did not materially improve between 1982 and 1992, and it remained the lowest of the five countries. It fell substantially in the "The sharp decline in the deposit-to-GDP ratio in Mexico in mid- to late 1980s as the cash-to-deposit ratio in- 1988 (Chart8) is somewhat overstatedbecause of a bank liability, creased. By the early 1990s, the loan-to-asset ratio called banker's acceptances, that had many of the characteristics of a bank deposit but was not classified as such. In 1988, banker's had deteriorated moderately in Colombia relative acceptances, unlike bank deposits, were not subject to interest to its position in 1982. The temporary increase in rate ceilings, and, in that year, the rules guiding their issuance the loan-to-asset ratio in the mid-1980s reflected were liberalized. Chart8. TotalDeposits (Percent of GDP) t:1-7A\7 55 Argentina Chile --+50 20 28 Colombia Mexico 20 A Peru 18 Sources: IMF, International Financial Stat~stics,various issues, and World Economic Outlook, various issues; Central Bank of Argentina; Chile. Superintendencia de Bancos e lnstituciones Financieras. Informocion Fi- nancier~;Colombia. Banco de la Republics; Mexico, Comisi6n Nacional Bancaria; Peru, Superintendenciade Banca y Seguros; and IMF staff estimates. AppendixTables TableAI. Argentina: SummaryAccounts ofthe FinancialSystem (Percent of GDPJ BalanceSheet Item 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 19921 Consolidatedfinancial system Net foreign assets (I0.16) (12.52) (12.13) (16.02) (14.72) (19.03) (18.45) (36.69) (39.93) (18.20) (9.89) Net domestic assets 31.3 1 34.09 32.64 32.61 33.05 39.56 41.85 55.78 52.58 28.03 22.59 Credit to public sector 1 1.79 19.35 15.09 13.01 14.12 19.29 25.38 44.46 33.2 1 14.61 10.44 Credit to private sector 28.49 23.65 19.34 16.02 15.93 18.47 18.18 23.61 19.63 12.59 15.54 Liabilitiesto private sector 21.15 21.57 20.5 1 16.78 18.23 20.57 23.40 19.09 12.65 9.84 12.70 Central Bank Net international reserves (3.70) (6.14) (7.08) (9.43) (9.52) (13.03) (12.15) (30.57) (33.0 1) (15.38) (7.26) Assets 2.65 2.57 2.26 5.26 4.05 2.77 3.92 4.46 10.40 5.00 4.85 Liabilities 4.68 6.70 7.21 12.25 11.60 13.85 14.75 28.70 36.22 14.85 8.36 Treasury liabilities2 1.66 2.01 2.12 2.56 2.01 2.04 1.53 6.33 7.19 5.53 4.00 Net domestic assets 23.48 24.19 20.75 26.01 24.64 27.47 29.6 1 4 1.95 38.26 19.70 12.12 Credit to public sector 7.3 1 17.38 13.49 13.20 13.02 15.69 21.78 35.95 22.93 10.59 8.00 Credit to financial system 21.84 12.73 12.02 8.67 13.12 1 6.463 20.6 1 29.62 21.06 9.9 1 9.22 Liabilities to financial system 15.17 13.36 9.50 12.44 10.92 10.46 13.50 5.77 1.97 1.44 1.47 Liabilitiesto private sector 3.94 4.16 3.93 3.96 4.61 3.99 3.87 5.60 3.27 2.88 3.39 Commercialbanks Net foreign assets (6.46) (6.38) (5.05) (6.41) (5.21) (5.53) (6.21) (5.92) (6.93) (2.80) (2.59) Net domestic assets 24.46 16.46 24.89 14.33 16.03 10.89 11.79 40.21 28.93 18.19 19.47 Credit to public sector 4.36 1.90 1.56 (0.19) 1 .OO 3.43 3.60 8.53 10.23 3.99 2.42 Credit to private sector 25.34 20.94 17.75 15.46 15.62 17.06 17.82 23.33 19.55 12.47 15.35 Reserves4 14.57 13.01 9.39 12.44 11.12 10.50 13.41 11.79 1.95 1.60 1.54 Liabilities to Central Bank 16.56 6.79 13.22 7.92 8.21 -3 - 28.45 14.63 10.11 9.23 Liabilities to private sector 16.00 16.30 16.01 12.44 13.82 15.81 18.99 13.05 9.32 6.88 9.19 Sources: Central Bank of Argentina: Ministry of Economy; IMF, International FinancialStatistics,various issues;and IMF staff estimates. 'Preliminary data. 'Foreign currency bonds (BONEX. BOTE. BOTESO, and BOCON). The item also encompasses the adjustment for BONEX included in the BCRA's reserve assets. 3The decomposition of central bank credit by financial institutions is not available. 4lncludes bonds and frozen deposits in the central bank. TableA2. Mexico:SummaryAccountsofthe FinancialSystem wantofGDP) Balance Sheet Item Financial system N e t foreign assets N e t domestic credit Credit t o public sector2 Credit t o private sector Other foreign liabilities3 Other liabilities4 Money and quasi money Bank of Mexico N e t foreign assets Assets LiabilitiesS N e t domestic credit Credit t o public sector N e t credit t o fin. N e t credit t o banks7 Credit t o private sector Currency in circulation Commercial banks8 N e t foreign assets N e t claims o n Bank of Mexico N e t domestic credit Credit t o public sector Credit t o private sector Other fore~gnliabilitie~'.~ Other liabilities10 Money and quasi money" Government development banks N e t foreign assets N e t claims o n Bank o f Mexico N e t domestic credit Credit t o public sector Credit t o private sector Other foreign liabilitie~l,~ Other liabilities4 Money and quasi money" Sources: Bank of Mexico: IMF, InternationalFinoncolStatistics, various issues; and IMF staffestimates. ISeptember 1993 data. 2Net credit to federal government plus net credit to other public sector. ]Sum of medium- and long-term liabilities plus money and quasi money in foreign currency. 4Liabilitiesto nonbank financial public sector (which excludes liabilitier to official trust funds of the Bank of Mexico); it also includescapital and sur- plus for 1982-84. 5For 1982, balanceof payments support loanfrom Bank for International Settlements (81s) was includedas a foreign reserve liability of the monetary authorities and only a portion appears as credit to the federal government. 6The sum of the following categories: net credit to official trust funds of Bank of Mexico, net claims on other government-related financial corpora- tions, net credit to commercial banks, and net credit to government development banks. 'Sum of net credit to commercial banks plus net credit to government development banks. 8Data for 1985-92 are significantlyrevised in late 1980s. and hence the 1981-83 data are not strictly consistent with the 1985-92 data. 9Med~um-and long-term liabilities item included in this category also include net disbursementsunder Commodiry Credit Corporation (CCC) loans in 1983. 1984. and 1985. "'Liabilities to nonbank financing public sector (which excludes liabilities to off~ialtrust funds of the Bank of Mexico) for 1985-92: a also includes lia- bilities to the rest of the banking system for 1983 and 1984and capital and surplus for 1982. I lExcludespublic sector deposits incorrectly classified as private sector deposits. Balance Sheet Item Consolidated bankingsystem Net international reserves Medium- and long-term foreign assets Net domestic assets Nonfinancialpublic sector Private sector Liabilitiesto private sector Central Reserve Bank Net international reserves Assets Liabilities Medium- and long-term foreign assets Net domestic assets Nonfinancialpublic sector Bankingsystem Private sector Liabilitiesto private sector National Bank Net international reserves Medium- and long-term foreign assets Net domestic assets Nonfinancialpublic sector Bankingsystem Privatesector Liabilitiesto private sector Commercial banks Net international reserves Medium- and long-term foreign assets Net domestic assets Nonfinancialpublic sector Bankingsystem Private sector Liabilitiesto private sector Development banks Net international reserves -0.34 0.22 0.04 0.02 -0.18 -0.06 -0.37 -0.06 -0.28 -0.21 -0.16 Medium- and long-term foreign assets -1.32 -1.46 -1.56 -0.39 -0.20 -0.21 -0.35 -0.14 -0.24 -0.08 -0.06 Net domestic assets 4.54 4.18 4.37 2.65 1.79 1.93 1.99 1.58 1.39 0.78 0.49 Nonfinancialpublic sector 0.99 0.37 0.21 0.45 0.34 0.20 1.1 1 0.39 0.32 0.20 0.14 Banking system -1.22 -2.36 -3.03 -1.84 -2.55 -3.13 -3.16 -3.65 -1.62 -0.21 -0.22 Private sector 7.56 8.87 8.98 6.15 5.99 6.27 5.25 3.31 3.28 1.82 1.27 Liabilitiesto private sector 2.88 2.93 2.86 2.27 1.42 1.67 1.27 1.38 0.87 0.49 0.28 Sources: Central Reserve Bank of Peru; IMF, InternotionalF~noncialStatistics,various Issues;and IMF sraff estimates 'June 1992 data. IV New BankingCrisis in LatinAmerica: 1994-95 A fter several years of tight monetary policy in the context of a managed exchange rate system, Chart 9. Indexesof BradyBondPrices in 1993 and 1994, the Central Bank of Mexico ex- (December 1, 1994 = 100) panded credit to the banking system at an accelerat- ing pace, increasing by 30 percent in 1993and by 40 percent on an annualized basis for the first nine months of 1994, compared with an 11 percent in- Venezuela crease in 1992. The rapid expansion of domestic credit to the banking system resulted in a loss of in- ternational reserves beginning in February 1994. As international investors became skeptical that Mexico could service its large stock of foreign debt, capital inflows, which had characterized the early 1990s,re- versed, accelerating the loss in international re- \,' serves. Authorities did not respond promptly to these Argentina Mexico losses by contracting domestic credit; consequently, 70 ' ' ' ' ----L--- a pressures on the exchange rate became so severe that December 1994 January 1995 February 1995 a major devaluation was unavoidable in December 1994. Source: BloombergBusiness News Despite a sharp devaluation of the peso and the es- tablishment of an economic program to restore con- fidence. investors remained uncertain that Mexico had achieved a sustainable solution to its problems; hence, interest rates remained extraordinarily high. the crisis unfolded, policies were put into place to re- The effect of the Mexican crisis was felt in a num- structure the banking systems. ber of other Latin American countries, as reflected in As discussed in the previous section, the evidence the market for Brady bonds and in the prices of equi- from the 1980s strongly suggests that the quality of ties issued bv Latin American firms (see Charts 9 the franchise going into a crisis is an important de- and 10). ~rgentina,with a strong commitment to a terminant of the success of bank restructuring pro- fixed exchange rate, was especially vulnerable to at- grams. To assess the prospects of recovery in the - - tack. Like Mexico, ~ r ~ e n t i hada growing current n a Mexican and Argentine banking systems, this sec- account deficit and a large stock of foreign debt. tion provides a detailed look at the banking fran- Concerns that these factors would lead to a devalua- chises in these two markets on the eve of the tion and a default on foreign debt resulted in a large 1994-95 crisis. Specifically, it considers whether the drop in bank deposits and Brady bond and equity following elements are present to facilitate an effec- prices. tive workout process: accurate accounting standards As a result of the sharp rise in interest rates caused that make it possible for supervisors to evaluate the by the exchange rate crisis, severe problems devel- quality of banks; the presence of a number of well- oped in the banking systems of Argentina and Mex- managed institutions that can absorb banks with ico as it became apparent that many borrowers could poor management; and evidence that bankers and in- not meet the interest and principal payments on their vestors perceive that they will suffer losses if banks debts. Consequently, investors became skeptical fail. about the solvency of banks despite the improve- It is shown that the overall quality of the banking ment in the franchise value of banks over the previ- systems in these markets improved in the early ous few years (see Section 111).As the magnitude of 1990s, as evidenced by the accounting information Evaluatingthe Banking Franchise The first step in evaluating the franchise strength of the Mexican banking system is to determine whether accounting standards permit one to distin- guish the relative riskiness of the two classes of -Argentino ....Colombio -- Venezuela banks. Perhaps the most prominent accounting --Brozil .... measure of bank soundness is the ratio of capital to -Chile -Mexico Peru risk-weighted assets. The Bank for International Set- 120 tlements has established a minimum acceptable stan- dard of 8 percent, of which half must be equity capi- tal. According to this standard, risky assets, such as loans, are given a 100 percent risk weighting, which means that each dollar of a loan must be covered by 8 cents worth of capital. The risk weighting of other as- sets varies according to perceived risk. For example, government securities denominated in home country currency are usually given a zero risk weighting. Hence, they do not require any capital at all. As a proxy for the ratio of capital to risk-weighted 20 December I994 ' January I995 februaty I995 assets, the average loan-to-capital ratio is used. In 1992, this ratio was 13.3 percent for large banks and Source: Bloomberg Business News 10.6 percent for small banks, providing an account- ing indication that the retail banks had a riskier port- folio than wholesale banks (see Table 8). The ques- tion is whether investors relied on this information in determining where and at what interest rate they placed their f&ds and whether supervisors used thk available to investors and supervisors for assessing information to constrain the growth of credit in risky the quality of individual banks. Nevertheless, in ret- institutions. rospect, pockets of weakness have persisted in each During 1992,the assets of large banks grew by 18 system, and, despite the increased reliability of percent, and small bank assets grew by 20 percent banking data, the authorities permitted weak institu- (Table 8). Thus, it appears that neither the market tions to expand their balance sheets. As interest rates nor regulators constrained the growth of small banks increased, the weak banks were the first to suffer, relative to large banks. An analysis of the funding and investor skepticism about their soundness exac- sources and the interest rates at which each group of erbated the crisis. banks was able to raise funds, however, makes it clear that the market distinguished between these two types of banks. Mexico As indicated in Chart 1I, in late 1992, time de- posit interest rates were substantially below the in- To assess the strength of the banking franchise in terest rates offered on repurchase agreements collat- Mexico prior to the 1994 financial crisis, banks are eralized by government securities.46 Usually, the divided into two categories: large banks, which pri- interest rate on repurchase agreements should be marily serve corporate customers, and small, retail- slightly below deposit interest rates because the for- oriented banks. At the end of the third quarter of mer is a collateralized loan to a bank whereas the de- 1994, large banks held about 40 percent of the assets posit is not collateralized. This was not the case in of the Mexican banking ~ystem.~5 Mexico in 1992, however, because posted deposit and repurchase agreement interest rates were af- 45The chief characteristic that distinguishes wholesale and re- fected by the riskiness of the institutions raising tail banks in Mexico is their deposit mix: in 1994. 36 percent of funds through each instrument. deposits at Mexican wholesale banks were demand deposits, compared with 19 percent at small banks. In contrast, in the United States, wholesale banks have fewer denland deposits as a 46A repurchase agreement is the sale of a security with an percentage of deposits than retail banks. Large U.S. corporations agreement to repurchase at a specified date at a specified price. hold repurchase agreements for short-term liquidity whereas in Thus, for the seller, this agreement is a way to fund a securities Mexico largecorporationshold demanddeposits. However, Mex- portfolio; that is, the agreement is a liability. Hence, banks can ican households use cash for liquidity much more frequentlythan use repurchase agreements rather than deposits as a means to U.S. households. fund their securities. Chart 11, interest rates on repurchase agreements Tgble8. SdectedMexicanB m b gData dropped below deposit interest rates, and small banks were active in raising funds in both markets, with their deposits growing by 35 percent and repur- chase agreements growing by 55 percent (Table 8). September However, large banks were very cautious about Balance Sheet Item 1992 1993 1994 expanding their balance sheets during this mood of increasing confidence. Assets of large banks grew by Capital-to-loans 11 percent during 1993, compared with almost 40 Total banking system 11.9 12.1 12.2 Large banks 13.3 13.7 13.7 percent at small banks, and deposits at large banks Small banks 10.6 10.7 1 1 . 1 grew by only 2 percent (Table 8). Large bank re- Asset growth straint was especially remarkable given the expan- Total banking system 19.4 26.2 20.5 sion in credit to the banks supplied by the central Large banks 18.3 11.1 14.5 bank in that year. Small banks 20.4 39.7 24.8 In addition, large banks strengthened their capi- Loan growth tal position between the end of 1992 and the end Total banking system 37.2 22.1 15.0 of 1993 relative to small banks: the ratio of capital Large banks 34.2 6.0 7.9 to average loans at large banks increased from Small banks 40.3 37.8 20.4 13.3 percent to 13.7 percent, whereas, for small Depositgrowth banks, the ratio increased by only '110of 1 percent, to Total banking system 17.3 18.7 13.4 Large banks 21.6 2.1 4.7 10.7percent. Small banks 13.1 35.9 20.1 During the first nine months of 1994, investors perceived increased risk in the Mexican financial en- Repurchase agreement growth Total banking system 3 1.9 58.2 27.7 vironment, and interest rates rose by 800 basis Large banks -8.0 65.5 41.3 points. Investors, however, showed continued will- Small banks 61.6 55.1 21.5 ingness to place funds with small, risky banks, as these banks' assets grew by 25 percent, compared Source: Mexico. Cornision Nacional Bancaria, Estadistico de la with 14 percent at large banks. An indicator of in- BancaM~iltiple(1994). creased risk taking by small banks is that their loan In 1992, as indicated in Table 8, small banks raised funds primarily through repurchase agree- Chart I I. Mexico: Deposit and Rcipulrc ments whereas large banks raised funds primarily by chaseAgreement Rates, Sep. 92S)ec. 94 issuing deposits. Deposits at large banks grew by 22 (Percent) percent during 1992, compared with only 13 percent at small banks. In contrast, repurchase agreements grew by over 60 percent at small banks, compared with negative growth of 8 percent at large banks. -daygovernmentrepurchase rote The high interest rates in the repurchase market re- flected the rates that risky banks had to post to raise 20 funds, whereas the low rates in the deposit market reflected the rates at which safe banks were able to raise funds. The difference in the cost of raising funds in these two markets indicates that investors 15 were well aware of quality differences across banks. Because investors were willing to supply funds to large banks at relatively low cost, the average cost of 10 funding to average assets at large banks in 1992 was I 90day deposit rate 11.76 percent. This compares with 14.51 percent at small banks. 51 I In 1993, the situation changed from that described Sep Jan Apr Jul Nov Feb May Sep Nov above. As investors gained confidence in the Mexi- Nov Feb Jun Sep Dec Apr Jul Oct 'can economy, interest rates in general declined, and Source: Mexico. Cornision Nacional Bancaria (1994). investors became more willing to offer funds to safe and risky banks on similar terms. As indicated in portfolio grew by 20 percent during this period, bear the consequences of their actions. Regulators compared with 8 percent at large banks.47 As a re- must ensure that those that pay a premium suffer sult, the ratio of loan loss reserves to average loans some losses. fell behind the ratio at large banks. At year-end At the same time, however, the authorities permit- 1993, this ratio stood at 3.7 percent for large banks ted an excessive expansion of the weakest banks, and 3.9 percent for small banks. By September 1994, signaling some deficiencies in supervisory practices. it stood at 4.5 percent for large banks and 4.1 percent In fact, in late 1994, the authorities actually funded for small banks. the expansion of weak institutions, rather than plac- The fact that relatively risky banks were able to ing institutions that could not access the market grow rapidly during 1993 and the first nine months under strict supervisory control. of 1994 demonstrates that, in periods of rising eco- nomic confidence, the role of the regulator becomes Restructuring the Banking System crucial. In 1992, investors restrained riskier banks by demanding premiums on the funds they supplied The Mexican banking authorities are currently in to these institutions. whereas in 1993 and the first the process of restructuring their banking system, nine months of 1994, this restraint disappeared. which has suffered substantial losses. To effect this Bank supervision was not strong enough to offset rescue, they have established a program to recapital- the weakened market discipline that permitted small, ize the banks and a program to restructure nonper- riskier banks to expand rapidly. While weak supervi- forming loans, similar to that used by Chile in the sion did not lead to banking problems during 1993, 1980s. when prosperity was evident, it became a crucial The recapitalization program, known as Programa factor in magnifying the banking problems associ- de Capitalizaci6n Temporal, or PROCAPTE, pro- ated with the crisis in 1994, as will be described vides for the insurance fund, Fondo Bancario de Pro- below. tecci6n a1 Ahorro (FOBAPROA), to lend funds to During the fourth quarter of 1994, as the crisis the banks in the form of subordinated debt that will unfolded, depositors began to show some prefer- count as capital. In five years, this debt will be con- ence for quality. As indicated in Chart 11, by the verted to equity. FOBAPROA can exercise conver- end of the fourth quarter, deposit rates were some sion rights before the end of the five years if bank 20 percentage points below interest rates on open capital (excluding the subordinated debt) falls to less market paper. Under these conditions, large banks than 2 percent of assets or if the regulators believe experienced deposit growth of over 17 percent, that the solvency of the bank is impaired. compared with a 10 percent growth rate at small In the loan restructuring program, the government banks. Assets at the two sets of institutions grew by will issue zero coupon bonds, paying interest in- about 18 percent during the quarter. Unlike 1992, dexed to inflation, to a trust fund established for the however, small banks could not access the repur- purpose of holding banks' nonperforming loans. To chase agreement market for funds to make up for fund the government bonds, the trust fund will issue limited access to the deposit market. Instead, asset liabilities to the government. The trust fund will then growth at small banks was financed by credit from exchange the government bonds for nonperforming the central bank, which grew by over 110percent at loans currentlv held on banks' balance sheets. As a small banks compared with less than 60 percent at result of these transactions, the trust fund's assets large banks. will be nonperforming loans, funded by liablities is- An evaluation of regulatory and supervisory ef- sued to the forts over the last few years yields mixed results. Although the loans are held in a trust fund, the The authorities have substantially improved ac- banks will remain in charge of managing these loans counting standards, thereby allowing investors to back to health. The interest on the government bond evaluate the riskiness of banks. The differences in will be paid from the portfolio of nonperforming interest rates among banks at which investors loans. The authorities recognize that this program placed deposits indicate that the market distin- may create liquidity shortages. The zero coupon guished among banks by their quality. Thus, the cri- bond has no cash flow whereas bank deposit liabili- sis presents an opportunity to demonstrate that ties are relatively liquid. If depositors withdraw their bankers and large depositors taking undue risk will funds and accumulated interest, banks will not have the cash flow from their assets to meet these de- mands. If the liquidity shortage is in U.S. dollars, 47Despite the rapid growth in the assets of small banks during FOBAPROA will provide liquidity assistance; if it is the first nine months of 1994, the ratio of capital to average loans in pesos, the central bank will provide a facility to at these banks increased from 10.7percent at the end of 1993 tu 11.1 percent at the end of September 1994, indicating some effort purchase securities from banks under short-term re- to improve their balance sheet quality (Table 8). purchase agreements. To reduce the interest burden on borrowers, the Argentina principal of the nonperforming loans will be in- dexed to inflation.48This will also aid the banks in In the years preceding the current crisis, the Ar- restructuring nonperforming loans because it will gentine banking situation differed from that of Mex- give them time to work with borrowers without ico. As mentioned in Section 111, the experiences of forcing them into bankruptcy. For this indexation depositors during the hyperinflation of the late policy to be viable, the principal of liabilities issued 1980s had a severe effect on their willingness to by the trust fund to the government must also be hold deposits in the domestic banking system. De- indexed. spite the adoption of a strong economic program in If the nonperforming loan portfolio cannot earn 1991 that brought inflation down sharply, by 1992, enough income to pay off the zero coupon bond, the Argentine investors were not as confident as their banks must cover the interest and principal on the Mexican counterparts in the quality of their banking bond from their own capital account. If this causes system because the deposit-to-GDP ratio remained a bank's capital account to fall to less than 2 per- substantially below its level of the early 1980s. In cent, FOBAPROA will have the right to close the 1993, deposits rose sharply relative to GDP, to bank and sell the assets. If a bank has enough funds 14 percent from 10 percent, indicating improved to pay off the government bonds but not enough confidence in the economy and the banking system to buy back the subordinated debt of FOBAPROA, (Chart 8). This ratio, however, was among the low- the insurance agency maintains a claim on bank in- est in Latin American countries. Reintermediation come as an equity shareholder at the end of five into the banking system continued during most of years. 1994. As in Mexico, it is important to determine To reduce the interest rate burden on solvent bor- whether this increase in confidence was based on rowers, banks will be encouraged to index the prin- tangible improvements in the quality of the banking cipal of their loans to inflation. The success of this system. program depends on depositors' willingness to ac- cept an indexed principal contract as well. Other- Evaluating Banks wise, banks will have a liability with a much greater cash-flow payment than they receive on their assets. To make this assessment, the banking system is To overcome this problem, the Banco de MCxico has divided into classes of banks to determine whether announced that the principal of its outstanding loans accounting data risk classifications correspond to in- to banks will be indexed to inflation. vestor perceptions of risk. The analysis is based on a With the central bank supplying indexed credit to comparison of March 1993 and November 1994 bal- the banking system, the size of its balance sheet will ance sheets for four categories of banks: national increase at the rate of indexation. The authorities public, provincial, private, and foreign.49National recognize that, if the central bank funds its balance public banks, dominated by Banco Nacidn, are sheet with nonindexed liabilities, it will face large owned by the central government. Provincial banks deficits for several years to come, which could cause are owned by the provincial governments. In 1994, the fiscal situation in Mexico to deteriorate. To miti- national public banks accounted for 26 percent of gate this problem, the authorities are using inflation large bank assets, municipal and private banks ac- to control a priority in the design of their economic counted for 29 percent each, and foreign banks ac- program. counted for 16 percent of large bank assets. The If restructuring actually forces the riskiest banks sample for each category of banks is limited to the out of business, future growth will be more disci- largest banks in each group, which together ac- plined throughout the banking system. The data in- counted for about 70 percent of the deposits in the dicate that, as the Mexican authorities restructure banking system at the end of 1993. their banking system in the current crisis, they have Compared with the Mexican banking system, in several well-managed banks that can take the lead 1993 all segments of the Argentine banking system in establishing loan workout programs and can also appeared to be strongly capitalized. Capital-to-loan acquire riskier banks whose loan portfolios will ratios were more than 20 percent for all categories of never be able to cover the cost of the government banks with the exception of foreign banks, where the bond. ratio stood at about 16percent (see Table 9). Hence, the relative quality of banks cannot be ascertained from capita1 ratios. As will be demonstrated below, 4*During inflation, the principal of loan contracts with fixed however, the behavior of investors indicates that nominal value depreciates in real terms. Lenders are compensated for this with high interest rates; however. borrowers must pay off real principal at a faster rate than in noninflationary times, which 490wing to differences in reporting periods across banks, data increases their cash-flow burden. for the private banks cover the first half of 1994 as well as 1993. T d e 9. SelectedArgemtine Bm&g Data (Pemnt u k othenvinindiuw!) Growth 1993 1994 Rates Public banks Capital-to-loan ratio 35 4 27.0 Gross interbank dollar borrowings (in millions of U.S. dollars) 1,579.1 2,006.5 27.1 Dollar deposits ( ~ nmillions of U.S. dollars) 1.600.2 3,046.0 90.4 Provincial banks Capital-to-loan ratlo 22.6 20.4 Gross ~nterbankborrowtngs (in m~ll~onsof U.S. dollars) 1,690.3 1,514.8 - 1 0.4 Dollar depos~ts(in millions of U.S. dollars) 1,945.7 3,041.5 56.3 Private banks Capital-to-loan ratio Gross interbank borrowings (in millions of U.S. dollars) Dollar deposits (in millions of U.S. dollars) Foreign banks Capital-to-loan ratio Gross interbank borrowings (in millions of U.S. dollars) Dollar deposits (in millions of U.S. dollars) Total banking system Capital-to-loan ratio Gross interbank borrowings (in millions of U.S. dollars) Dollar deposits (in millions of U.S. dollars) Source:Argentina. Superintendenciade Entidades Financieras. Nore: Dataare for March 1993 and November 1994. they did not accept high capital ratios as evidence posits by 55 percent (Table 9). This increase, how- that all banks were safe. ever, was substantially less than the 76 percent in- Unlike Mexican banks, Argentine banks have a crease in dollar deposits experienced by the system large portion of their assets and liabilities denomi- as a whole. Although depositors were less discrimi- nated in U.S. dollars.50 They have two primary natory in their choice of banks than interbank sources of dollar funds: interbank borrowings and lenders, they required a relative increase in the inter- deposits accepted from the public. In 1994, provin- est rate on dollar deposits to supply these funds. In cial banks began having difficulty raising dollar April 1993, dollar deposits were paying 500 basis funds. Dollar interbank funds declined by 10percent points less than peso deposits of similar maturity at provincial banks, compared with a 28 percent ex- (see Chart 12). By December, the spread was less pansion rate for all banks, a 53 percent expansion than 300 basis points. rate at private banks, and a 94 percent expansion rate In contrast to the dollar market, in the peso market at foreign banks. This suggests that banks lending there is not enough evidence to determine whether dollars in the interbank market had grown skeptical investors distinguished between the quality of banks. of the quality of provincial banks. Peso deposits at provincial banks expanded by 31 To finance dollar assets, provincial banks had to percent, faster than the 25 percent increase experi- turn to the deposit market, and, between March 1993 enced by the system as a whole but slower than the and November 1994, they increased their dollar de- 68 percent and 50 percent increases at private banks and foreign banks, respectively. Peso deposits de- clined at Banco Nacibn, and most of these deposits SoArgentina permits domestic investors to hold dollar deposits flowed into private and foreign banks rather than in the domestic banking system on the same terms that they are provincial banks. But most of these deposits were permitted to hold domestic currency deposits. In 1992,Argentine probably in the Buenos Aires metropolitan area, and banks borrowed substantial amounts of dollars from foreign banks as dollar loans exceeded dollar deposits. In addition, they it would have been difficult for outlying provincial financed a portion of their dollar assets with peso deposits. banks to capture them. The government has established a trust fund to re- Chart: 12. Argemtina: PBso3~ndDollar In- capitalize banks. The fund will be partially financed terest RatesonBank Depdis and Lorma;, by the proceeds of US$2 billion in government 1993 bonds with a three-year maturity paying a below- fkmnt) market floating interest rate sold to domestic private investors and foreign financial institution^.^' The re- mainder of the trust fund will be financed bv interna- tional multilateral agencies. The fund will purchase Peso prrme subordinated debt in banks with a maturity of three years, which will be converted to equity if a bank 10 fails to repay interest and principal. An important feature of the program is that recapitalization will be 9 .. .- available only to those banks that can be managed Dollor prime /- 8 back to solvency. A significant portion of the re- sources from the fund will be used to finance merg- ---- - 1- I -----___ ---,,,Dollardeposits ers and acquisitions. 6 To further restore investor confidence in the bank- 5 L - L L - U - u ing system, the government also established a de- Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec. posit insurance system. This system will be funded through a levy of between .03 percent and .06 per- Source: Argentina. Superintendencia de Entidades Financieras cent on average daily deposit balances each month. (BuenosAires. 1994). The fund will expand until it covers 5 percent of the deposit base or Arg$2 billion. Banks will not be able to bid aggressively for insured deposits because in- terest rates payable on these deposits will be capped to a reference rate. Rescuing Banks Although the Argentine restructuring program The spillover effects from the Mexican crisis contains crucial elements to improve the quality of caused heavy deposit withdrawals from the pro- the domestic banking system, some risks remain. An vincial banks and many smaller commercial banks. important one is that the scale of the bad loan prob- To aid these institutions, the Argentine government lem at banks might exceed the funds available in the established a "safety net" fund, supported by large trust account and through the deposit insurance banks and managed by Banco Nacibn, which was scheme. If this were the case, short-term pressures used to provide liquidity assistance to banks losing on the fiscal situation could be avoided only if the funds. The central bank also provided liquidity authorities were able to mobilize additional funds assistance to banks through swap arrangements. from the private sector. Under these programs, the central bank has lent for- eign exchange to banks, collateralized by banks' Comparison of Franchisevaluein highest- quality paper sold to the central bank under repurchase agreements. The scope of this program Mexico andArgentina on the Eve of has been limited because regulations are in place the Crisis that severely restrict the central bank's authority to act as lender of last resort. In both markets on the eve of the crisis, it was ap- In recognition that many banks face problems parent that investors attempted to distinguish banks greater than can be managed by providing liquidity by the quality of their portfolios. In Argentina, port- facilities, the government has instituted policies to folio quality was reflected in the rapid growth of restore investor confidence in banking. The philoso- U.S. dollar interbank borrowing at large private and phy underlying these policy measures is to use the foreign banks relative to provincial banks. In Mex- crisis as an opportunity to strengthen the banking ico, it was reflected in the interest rates that small system by closing badly managed banks. The au- banks had to pay for funds to maintain a growth rate thorities will encourage sound banks, whether of do- substantially higher than that of large banks. mestic or foreign ownership. to purchase closed The difference in the two markets is that in Mex- banks. The authorities have demonstrated their ico, accounting information was a more useful guide strong commitment to establishing a sound banking system by placing no restrictions on the degree of "The government has been able to raise funds at below-market foreign ownership of their banking system that may interest rates by appealing to private investors' stake in the suc- result from the restructuring. cess of economic reforms. to evaluating bank quality than in Argentina. This is An important similarity between the financial an important distinction because in Argentina, while markets in Mexico and Argentina is that the authori- investors avoided a group of highly specialized ties in both countries permitted banks deemed rela- banks-those owned by provinces-their decision tively weak and risky to expand. In Mexico, these was based not on specific information about these were small banks; in Argentina, they were provincial banks but on the financial situation of the provinces. banks. A major conclusion that emerges from this Without reliable accounting standards, it is much analysis, therefore, is that, although the reversal of more difficult to judge the performance of individual capital flows triggered banking problems, more banks engaged in lending to the private sector. For stringent controls on the activities of weak banks be- example, are low loan loss reserve ratios of many fore the crisis would have reduced the magnitude of private banks a reflection of good loan quality or un- the bad loan problem and the cost of bailing out derprovisioning? Are high capital ratios derived banks in both Mexico and Argentina. from marking up the value of fixed assets, or are If the lessons from the 1980s have been well they based on a solid stream of retained earnings learned, the current banking crises will have a much from a performing loan portfolio? shorter and less devastating effect on the region's Thus, it appears that the availability of informa- economy than the earlier crises. The current difficul- tion in Mexico will help the authorities determine ties provide an opportunity for Latin American au- the ability of individual banks to generate cash flow thorities to dispose of poorly run institutions that from their loan portfolios. This can make it easier for have become insolvent, thereby permitting their them to determine which banks ought to be forced banking systems to emerge from the crisis on a into bankruptcy. It also permits them to better deter- sounder footing. If they take advantage of this op- mine which banks are the best candidates to take portunity, the prospects for returning to the growth over failed institutions. path of the early 1990s look promising. V Long-Term Challenges to Franchise Value of Banks w ith the Mexican and Argentine financial cal series on the ratio of foreign reserve assets to a crises fresh in their minds, policymakers are narrow definition of money supply-currency plus acutely aware that capital inflows are transit~ry.~" transaction deposits--covering the period of capital The experiences of the crises of the 1980s and the inflows is presented in Chart 13for the sample coun- recent banking difficulties of 1994-95 have demon- t1ies.5~The chart shows Chile to be the most aggres- strated that a sound precrisis banking system and ex- sive sterilizer between 1990 and 1993, followed by perienced and skilled bank supervisors are important Colombia and Peru. Mexico shows a very erratic to the prompt success of crisis recovery programs. pattern of sterilization, and Argentina, showing a Therefore. in assessing the ability of Latin American downward trend in the ratio, can be characterized as countries to deal with future adverse shocks leading the least active sterilizer. to capital flight, it is necessary to consider the long- The evidence for the sample countries analyzed in term challenges to the franchise value of banks in this paper suggests that the experience with steriliza- these markets. tion during the period of capital inflows of the early Two of the policymakers' major concerns, which 1990s has been mixed. For example, in 1992, Chile, have been confirmed by events in Mexico and Ar- which was an active sterilizer, had the highest ratio gentina, are associated with the expansion of domes- of deposits to GDP among the five countries in the tic credit that follows large, unsterilized capital in- sample (see Chart 8).In contrast,Argentina, the least flows. The first is that the inflows may lead to an active sterilizer, had among the lowest ratios. Since unsustainable appreciation in the real exchange rate. Chile has one of the strongest banking franchises in The second is that the expanded bank loans may not the group, this indicator suggests that the quality of be sound enough to stand a reversal of the inflows. A the banking franchise is a more important determi- financial crisis may then follow a reversal of the nant of the ratio of deposits to GDP than the steril- capital inflows. This section focuses on the latter ization policy chosen by the central bank. concern, and issues related to the sustainability of the Consistent with the mixed evidence, there is no exchange rate regime are addressed in Section VI. consensus among policymakers and analysts on Before the Mexican financial crisis at the end of whether and under what circumstances sterilization is 1994, the authorities in several countries resorted to a useful tool to manage capital inflows. While sup- a policy of sterilization to ameliorate the impact of a porters of sterilization stress the risks associated with possible reversal of the capital inflows. Sterilization an expansion of domestic credit, critics of steriliza- was used as a tool to control the domestic credit ex- tion emphasize the costs and limitations associated pansion effects of capital inflows. The degree to which a country sterilizes can be measured by the ratio of foreign reserves to the supply of money. This liabilities to the nonbank public. that is, it sterilizes through open measure captures the intent of sterilization, which is market operations, it controls the growth of the money supply indirectly because it does not create bank reserve deposits to control the growth of credit by controlling the around which banks can increase their issuance of transaction growth of the supply of money as foreign currency accounts. assets are purchased by the central bank.s3A histori- 5%4nindicator of sterilization is constructed by using the ratio of foreign reserve assets to M 1 rather than to the monetary base because the latter ratio would be affected by how the sterilization 52Comprehensive analyses of the recent episodes of capital in- policy is carried out. If sterilization is carried out through imposi- flows in Latin America are included in Calvo, Leiderman, and tion of high reserve requirements, which are part of the monetary Reinhart (1993a. 199?b), Corbo and Hernandez (1993). and base. the ratio would be biased downward compared with a situa- Claessens and Gooptu (1993). tion in which the policy is carried out through the issuance di- '"f the central bank controls growth by imposing high reserve rectly to the public of central bank liabilities that are not part of requirements on deposits, the growth of the narrow money the monetary base. The choice of an indicator of sterilization supply relative to foreign currency assets on the balance sheet of using MI or any other monetary aggregate in the denominator. the central bank is controlled directly. If the central bank issucs however, seems arbitrary. this liability is issued to a domestic commercial bank sterilize. The third part considers some recent empir- in the form of a reserve deposit in the central bank, ical evidence on the relationship between the quality the commercial bank would use this deoosit to ex- of the bank franchise and the quality of capital flows pand domestic credit until the required or desired through the equity market. Finally, the last part turns ratio of domestic credit to reserve deposits in the cen- to another challenge that banking systems in some tral bank is restored. The central bank can conduct Latin American countries are currently facing: the sterilization either by increasing reserve require- securitization of financial instruments. Specifically, ments on commercial bank liabilities or by issuing li- this part analyzes how the recent developments in abilities to nonbanks, which do not have the authority domestic capital markets and the increased access to or the public confidence to use the central bank lia- world capital markets affect the franchise value of bility as a vehicle to expand domestic credit. Latin American banking systems. What role does the state of the franchise value of the banking system play in the policy decision to sterilize or not and in the choice of alternative steril- Sterilization and FranchiseValue of ization methods'? Clearly, if the financial system Banking System has strong mechanisms in place to control the qual- ity of credit, some of the dangers of credit expan- In carrying out a sterilization policy using reserve sion would seem to be reduced. First, sound banks requirements on bank deposits, the central bank con- would have incentives to evaluate properly the risks strains the expansion of bank credit following a cap- associated with extending loans, and, second-and ital inflow by directing banks to hold a high portion perhaps even more important--central banks would of the funds collected through the issuance of de- stand ready to deal with financial difficulties posits in the form of reserves held as vault cash or by using emergency loan assistance as an incen- deposits at the central bank rather than as loans. If tive for banks to establish credible loan workout the reserve ratios against deposits specified by the programs.57 central bank are a binding requirement, this ratio Turning to the choice of sterilization method, the must be greater than that which banks would volun- discussions in Sections I1 and 111highlight the trade- tarily hold at the interest rate the central bank offers off of carrying out a sterilization policy through the on reserve deposits that banks hold with the central imposition of high reserve requirements on bank de- bank.58 posits: although high reserve requirements are effec- If banks are subject to a binding reserve require- tive in preventing the expansion of domestic credit, ment, the spread between the interest earnid on they reduce bankers' opportunity to learn how to loans and the interest paid on deposits will be greater build a strong banking franchise by preventing them than in the absence of the requirement. This will act from making loans. Based on their experiences in as a tax on bank intermediation, reducing both the the 1980s. policymakers have begun to recognize the quantity of deposits and the quantity of loans rela- dangers of forcing the banks to bear the burden of a tive to the situation in which the banks are not sub- sterilization policy through high reserve require- ject to reserve requirements. Thus, the policy will di- ments. An important question is whether these costs rectly affect the quantity of credit. Because it is the of a sterilization program can be avoided; that is, can purpose of a sterilization policy to reduce domestic the central bank find a way to sterilize without in- credit, this aspect of a high reserve requirement is hibiting the development of the franchise value of beneficial to the overall policy objective.59 As indi- the banking system? cated in Section 111, however, reserve requirements The first part of this section considers alternative induce both central bankers and commercial bankers policy options to sterilize. The second part discusses to accept lax credit standards. how the state of the banking franchise might affect Is there a way that the central bank can achieve authorities' policy choice of whether and how to the perceived benefits of credit reduction while pre- venting a deterioration in the franchise value of the S'lf the franchise value of the banking system is strong, even banks through a binding reserve requirement policy? the adverse impact on banks of a sudden reversal of the real ex- For example, are the deleterious effects of the above change appreciation would be minimized. Sound banks would policy reduced if the central bank sterilizes through take into account the probability of such an occurrence when ex- tending credit to economic agents whose real net income is largely dependent on revenues from the nontradables sector of the 58Bankshold reserves for settling transactions among individ- economy. Thcrcfore, either credit to the riskier sectors would be ual banks. The demand for reserves depends on the overall liquid- limited or the risk would be properly reflected in an accumulation ity of an economy's money markets and the lending policies of of capital. Indeed, the discussion in Section 111 made it evident the central bank. See Garber and Weisbrod (1992). Chap. 13. that the methods used by central banks in some Latin American s9Moreover, because reserve requirements tend to lower the in- countries played a major role in the exknt and duration of thc cri- terest rate paid on deposits, they wouId constrain additional in- sis in the 1980s. flows of capital. a voluntary rather than a required reserve program? bank were riskier than liabilities issued by the U.S. If the banks are to hold reserves voluntarily, they government or U.S. chartered banks.6" must be compensated through interest payment on The central bank could cover these losses in sev- reserve balances held with the central bank.""he eral ways. First, it might attempt to roll the losses reserve tax would no longer be a tax because, by de- over by issuing additional liabilities. This policy finition, reserves pay whatever it takes for banks to would be unsustainable, however, if the central bank voluntarily hold them. Thus, thc central bank would continued paying market interest rates on its liabili- have to pay an interest rate on bank reserve deposits ties; it would only lead to higher and higher liability that would result in the same amount of domestic costs. Alternatively, the central bank might have ac- credit that is created under the reserve requirement cess to subsidized funding from the government policy. Since, by assumption, the amount of domes- budget, but this approach would only transfer losscs tic credit is the same as before, the interest rate on from the central bank to the fiscal authority. If this loans is the same as under the reserve requirement source of funding were closed, the central bank policy. Depositors, however, obtain a higher yield would have to fund its losses on bank reserve de- because the banks now earn interest on reserve de- posits through its seigniorage earnings on currency posits at the central bank.61Hence, the tax cost of issue, which does not pay interest. If these earnings the sterilization policy is shifted from depositors to were the central bank's only source of revenue, the the central bank.- losses the central bank could absorb would be lim- Central banks also have the option of duplicating ited by its willingness to tolerate inflation. the above policy results by issuing central bank lia- If the central bank must minimize its losses and bilities paying market rates of return that are not control inflation, its options are few: it must limit its held in the banking system, as in Chile and Colom- role of direct lending to domestic borrowers to bia. Instead of imposing reserve "requirements" on demonstrate to the market that it is serious about banks that pay market rates of return, the central controlling inflation, and it must strengthen its su- bank could issue an equivalent amount of liabilities pervisory procedures to control bank risks before a to nonbanks to equal the amount of liabilities it had banking crisis occurs. It follows, therefore, that only originally held in the form of reserve deposits of central banks with strong franchises should risk pay- banks. ing interest on liabilities. The central bank liabilities issued to nonbanks would compete with bank deposits for investor funds-as some investors would shift from bank de- Strength of the Franchise and the posits to the liabilities of the central bank. This Impact of Sterilization would increase bank funding costs and loan ratcs and would reduce bank credit.62 Hence. the central The strength of the banking franchise affects the bank can choose a quantity of liabilities issued to the costs and benefits of a sterilization policy in more nonbank public that reduces bank credit to the same ways than just the direct costs to the central bank, level as that achieved through a policy of either im- however. This subsection considers the costs and posing reserve requirements or paying interest on benefits of sterilization under four different scenar- reserves. ios-where both the central bank and the banks have When the central bank chooses a policy that relies a strong franchise, where the central bank is strong on investors' voluntarily holding its liabilities, the but the banks are weak, where the banks are strong interest rate it must pay depends on how investors- but the central bank is wcak, and where both the cen- whether banks or the general public-evaluate the tral bank and the banks have a weak franchise value. risk of lending to the central bank. If the central bank If both the banks and the central bank have strong holds foreign reserve assets. such as U.S. dollar trea- franchises, bank credit decisions are basically sound, sury bills or bank deposits earning an almost risk- and the central bank is well equipped to restore con- free interest rate, it would run losses if investors be- fidence in banks if a systemic crisis occurs. If the lieved that liabilities issued by the domestic central central bank should choose to engage in some steril- ization to build up a war chest relative to domestic credit to use in the event of a systemic crisis, it can 60They would; however, be willing to hold some reserves with- do so at relatively low cost. There would, however, out being compensated with interest payments because the re- serves are necessary for payments clearing purposes. h'Incontrast to a reserve requirement policy. a policy of paying interest on bank reserves, by raising the yield on bank deposits. 63Central bank losses in Chile and Colombia-two active ster- may create incentives for further capital inflows. ilizers-reached 2.2 percent and 0.8 percent of GDP in 1991, re- h2Like a policy that pays interest rates on reserves, sterilization spectively. By 1992. these ratios had decIined to 1.1 percent and through open market operations may also attract further inflows 0.5percent of GDP, respectively. However, these losses cannot be of capital. fully attributed to sterilization practices. seem to be little need to engage in sterilization to chise. The place to begin is probably with central limit domestic credit growth resulting from capital bank policies on supervision and lcnding because this inflows because bankers will not take undue risks. If is most directly in the control of policymakers. the capital inflow exceeds the availability of safe do- mestic loans, domestic savers will be induced to re- duce their supply of domestic savings. For examplc, Capital Inflow Challenge and Franchise if bankers believe that there is an excess supply of Value of Banks: Some Empirical capital that cannot be invested at an expected return Evidence commensurate with the risk, they have two alterna- tives: they can invest the excess funds in foreign as- Capital can flow into an economy through chan- sets or they can reduce the interest rate paid on de- nels other than directly through the banking sy~tem.~S posits, which reduces loanable funds in the domestic For example, in many Latin American countries, eq- market. In either case, the domestic supply of sav- uity markets have attracted the interest of foreign and ings declines, and the expected return on Ioans in- domestic investors in the early 1990s. The fact that creases for a given level of risk. investors have alternatives to banks is also a source If the central bank has a strong franchise, but the of concern for policymakers in countries receiving banks are weak, the central bank must be concerned large inflows of capital. After all, even if regulators about overexpansion of bank credit. If it chooses a know that banks are making sound credit decisions, policy of sterilization. it can do so at relatively low they may not have knowledge about how firms are cost because investors perceive that central bank using funds that they have raised in the stock credit decisions both in normal times and in a sys- market.66 If an adverse shock occurs, the authorities temic crisis will be sound. The central bank must, may face pressures to protect the real value of share howcvcr, design policies that build the franchise prices. These concerns have sometimes added to ar- value of banks. To this end, it is probably preferable guments advanced by those supporting sterilization for the central bank to sterilize by issuing liabilities policies. This subsection examines the relationship directly to the public to give investors a safe haven between the quality of the bank franchise and the from risky banks rather than imposing high reserve quality of capital flows through the equity market to requirements, which, as shown by the experience of determine whether the banking franchise affects how the 1980s, does not encourage bankers to improve funds raised in the equity market are invested. the franchise value. In Section 11, it was argued that the banking fran- If the banks are strong and the central bank is chise plays a very important role in Latin Amcrican weak, resources should be kept out of thc hands of economies because banks monitor and control bor- the central bank. Rank credit expansion is a lesser rower liquidity. In the highly uncertain economic en- risk than the accumulation of foreign reserve assets vironment of a developing economy, a good bank on the central bank's balance sheet that can be used cnsures that its borrowing customers invest their later to expand central bank domestic credit. Even funds in projects with immediate cash-flow benefits. without sterilization, however, a capital inflow has Banks have tools at their disposal, such as the ability the potential to increase the balance sheet of the cen- to demand imrncdiate payment of principal rather tral bank as long as the resulting foreign currency in- than to roll over a loan, that give them power to flow is purchased by the central bank.64 In this case, force borrowers to abandon unprofitable activities. dollarization-by which residents are freely permit- To use these tools, banks analyzc short-term cash- ted to hold dollar-denominated deposits-may help flow data and the day-to-day activities of the firm. In to safeguard the franchise value of banks. If low re- sharp contrast to bank loans, equity contracts en- serve requirements on foreign currency deposits are courage firms to invest in projects that do not have imposed, a significant proportion of thc capital in- immediate cash-flow benefits because they permit flow will not end up on the balance sheet of the cen- corporations to raise funds at a low cost relative to tral bank. current earnings. Equity contracts do not require the If the franchise values of both the banks and the repayment of principal; nor do they require the bor- central bank are weak, policy options are very few. It rower to promise the investor a fixed stream of inter- is certainly wise to control the expansion of bank est or dividend payments. credit, but it is unwise to put resources under the con- trol of thc central bank. Whether policymakers in this situation choose sterilization or not, the most impor- hSWhen foreign investors purchase equity, they must exchange tant policy objective is to begin building the fran- a hard currency deposit for a local deposit to pay for their pur- chase. As a result, the local bank obtains hard currency funds that can result in an expansion in the local banking system. "Unsterilized intervention results in an expansion of the cen- h6The role of credit extended through the equity markets in a tral bank's balance sheet. speculative attack on the exchange rate is discussed in Section VI. In Section 111, evidence was presented that, in more volatile in weak franchise markets, which economies where central bank policies encouraged should lead to greater volatility in stock prices. banks to perform their role of monitoring firms' cash Price-earnings ratios should also, on average, be flow, that is, in those countries with a relatively higher in weak franchise markets because expected strong franchise value of banks, the adjustment to future earnings should be high relative to current large capital outflows associated with the debt crisis, earnings. Of course, stock market variability should while difficult, did not lead to the extreme levels of also make price-earnings ratios more variable in inflation that occurred where the banking franchise weak franchise markets than in strong ones. was weakest. In weak franchise markets, inflation Evidence on stock market volatility, covering the was used as a device to spread the cost of credit period of capital inflows, is presented in charts 14 problems throughout the economy because neither and 15. Volatility is measured in terms of local mar- central banks nor commercial banks had the means ket indexes in local currency and in U.S. dollars. All to resolve credit problems by working with troubled of these measures indicate that, through the end of borrowers to resolve their cash-flow problems. Be- 1994,on average variability has been highest-albeit cause equity contracts do not encourage investors to declining-in Argentina, Colombia, and Peru and monitor cash flow, investments made through these lowest in Chile and Mexico. This pattern fits investor contracts create many of the same risks as loans perceptions of the franchise ranking of the banking made by banks with weak franchises: lenders have systems in the early 1990s, as measured by the di- no means of resolving cash-flow problems in the posit-to-GDP ratios (see Section 111). Investors in event of renewed capital flight. Chile and Mexico perceive these banking franchises This potential weakness of the equity contract cre- to be stronger than those in Argentina and Peru. ates a monitoring role for short-term lenders. Al- Data on price earnings ratios in the two relatively though corporations can raise funds through equity weak franchise systems are not available over the en- contracts, even in markets like the United States, tire period, so inferences from this evidence cannot be they rarely have the freedom to depend completely as conclusive as the volatility evidence. Nevertheless, on equity contracts to fund their investment projects. as indicated in Chart 16, Argentine price-earnings Other lenders, whether bondholders, commercial ratios have been, on average, substantially higher paper holders, or banks, issue contracts that require than those of Chile and Mexico. Colombian price- borrowers to pay principal and interest on a timely earnings ratios have also been, on average, above basis. In Latin America, these other lenders are pri- those of the two countries where the franchise is per- marily banks. Thus, it is important to investigate ceived to be strong. Price-earnings ratios for Peru are whether, in those markets where the banking fran- available only for the third and fourth quarters of chise is relatively strong, banks play an important 1993 and are not presented in a chart. However, for role in maintaining the liquidity of firms even when the third quarter of 1993,the Peruvian price-earnings they issue equity contracts. ratio was 34.95, cornpared with 39.27 for Argentina, If banks monitor and manage corporate liquidity, 18.05 for Colombia, 16.89 for Chile, and 14.11 for firms will not be able to disregard the demand for Mexico. In the fourth quarter, the price-earnings ratio current cash flow, even though they rely heavily on was 39.19 for Peru, 24.47 for Argentina, 24.90 for equity markets. Banks will insist that borrowers Colombia, 20.04 for Chile, and 19.70 for Mexico. meet stringent payment schedules, or they will resort Thus, the meager evidence available for Peru sug- to bankruptcy procedures. In contrast, where the gests that price-earnings ratios are also higher in that franchise is weak, loan covenants will not be en- market than in the strong franchise markets. forced, and firms' investment decisions will be de- Thus, it appears that the quality of the bank fran- termined largely by the demands of equity investors chise affects how funds are invested outside the for capital gains rather than by current cash banking system. Where the franchise is strong, firms As a result, in an economic downturn, firms that invest in projects that place more priority on current must satisfy the demands of strong franchise banks cash flow than where the franchise is weak. To the will be in a position to cut costs to absorb the shock extent that an environment in which firm cash flow- of reduced revenues, whereas firms monitored by and hence economic value-is uncertain, speculators weak franchise banks will not. For example, in may be attracted to that market. A strong banking strong bank franchise markets, a larger portion of franchise may therefore reduce speculative capital in- firms' costs will be variable than in weak franchise flows that enter outside the banking system.68 markets. Hence, corporate profit streams should be hxln this regard, it should be noted that Mexico did not experi- cncc a rapid increase i r ~block prices in the months before the cri- 67For evidence on this point from several East Asian sis. In fact, the Mexican Bolsa Index stood at 2600 on December economies, see Weisbrod and Lee (1993). I, 1994, the same level ;is of January 1, 1994. ri,- 100 12 Argentina n Chile Io - 60 50 16 Colombia n Mexico Source: International Finance Corporation. Emerging Markets Database. 'Definedas a six-month moving average of the standard deviations of the monthly percent changes in the index. Capital MarketThreat to Banking in major industrial countries: the securitization of FranchiseValue many financial instruments that heretofore had ap- peared mostly on bank balance sheets. In Latin In addition to the policy dilemma imposed by ster- America, capital markets alternatives to bank loans ilization, before the financial crisis at the end of had begun to be available on a fairly wide scale, 1994,the banking systems in several Latin American most notably the corporate bond market in Chile, countries had started to face a challenge well known money markets in Mexico, and equity markets in a Chart IS. VofatH'iofLocalEquity Markettnd~x,U.S. Doflarsl @ramfirstguo* Fmpuoneri 12 Argentina Chile n 7 30 18 Colombia Mexico Source: International F~nanceCorporatlon, Emerg~ngMarkets Database 'Defined as a SIX-monthmoving average of the standard devlat~onsof the monthly percent changes In the number of other countries. By and large, the most in- international capital markets and has hampered the ternationally well-known firms were able to raise further development of several domestic securities capital in the U.S. and Euromarkets. This raised con- markets. Notwithstanding recent developments, cerns that the franchise value of the banking system however, the significant changes in domestic capital may have weakened in Latin America. markets that occurred in the early 1990s cannot be To a large extent, the present crisis has severely ignored, and the question still remains as to whether, curtailed the access of Latin American countries to once current banking difficulties are resolved, do- Chart 16. Price EarningsRatfos (F$acprorcer@fka~~ O 1990 1991 I992 50 - -- Colombia Mexico Source lnrernat~onalF~nanceCorporar~on.Emerg~ngMarkets Database mestic securities markets will pose a permanent cipal on time. In addition, if there should be an oper- threat to the franchise value of banks. This section ational problem during the issuance of commercial evaluates such a potential threat. paper, the corporation can draw on its credit line As suggested by the evidence in the previous until the problem is cleared up. These credit lines are subsection as well as by the discussion on the nature not credit guarantees; they can be, and are, canceled of the bank franchise value in Section 11, a strong on very short notice. It is the cancellation of a com- banking franchise can improve the performance of mercial bank credit line that signals to other in- bond and equity markets because banks are in a vestors that the borrower's credit quality has deterio- better position than equity holders to monitor corpo- rated. Thus, the threat of a line cancellation gives rate liquidity; thus, a capital market and a strong banks tremendous power over a borrower's busi- banking system can be complementary rather than ness, even though the bank no longer makes a loan. competitive. Given the importance of the banking franchise, In highly developed capital markets, such as the even in highly developed financial markets, what United States, short-term money market instru- competitive threat do capital markets pose to the ments. primarily the commercial paper market, have franchise value of hanks? Because capital market replaced bank loans as the primary source of short- development creates alternative sources of funds term credit to corporations. Despite this develop- that are available to corporations, it leads to a reduc- ment, banks continue to play the role of liquidity tion in the role of bank loans in corporate finance. monitor. Commercial paper issuers obtain a credit The spread between the interest rate on loans to line from banks to ensure investors that they can corporations and bank funding costs declines. and meet their commitments to deliver interest and prin- bank profits decline. Under normal circumstances, profitability would be restored through shrinkage of the banker's acceptance market and the interbank the banking industry both through mergers and market.70In early 1994, before the outbreak of the through a decline in the percentage of financial in- financial crisis, the commercial paper rate was about struments supplied by banks. A scaled-down bank- 13.5 percent, and the banker's acceptance and inter- ing system would then continue to play its important bank rates were about 11.5 percent; thus, the spread role in maintaining borrower liquidity.69 between the commercial paper market and the It is very difficult, however, for policymakers to banker's acceptance or interbank markets was about scale down the banking system. Because of the im- 200 basis points, compared with only a few or no portant role that banks play in maintaining liquidity basis points in the United States (Chart 17).71 Con- in financial markets and in providing accounts used sider the following scenario: if the noninterest ex- as the means of payment, they are usually implicitly pense involved in a bank making a large corporate or explicitly insured by the government, either loan is assumed to be 75 basis points, which is high through direct deposit insurance or through access to by U.S. standards, the bank is still able to earn a central bank credit. This insurance is often not spread of 125 basis points between a loan priced at priced appropriately, preventing bank deposit costs the commercial paper rate and its marginal cost of from fully reflecting the risks that banks take. Con- funds. If the bank funds this loan with an 8 percent sequently, bankers who operate banks that should equity-to-loan rati0,~2it will earn a pretax return on exit with this shift have an incentive to maintain equity of 27 percent, which is high even for a market their profitability by taking increased risks (see where treasury bill rates are somewhat less than 10 Weisbrod, Lee, and Rojas-SuArez, 1992). The gains percent.7-us, it appears that developments in do- from using the insurance subsidy to generate profits mestic capital markets have not yet significantly increase as competition from capital market instru- threatened the profitability of Mexican banks.74 ments causes spreads on safe credits to decline. Besides Mexico, Chile is the other Latin Ameri- Thus, at some point even banks with formerly strong can country where the issue of competition to banks franchises will be tempted to take risks (see Weis- from domestic capital markets has arisen. Chile is brod, Lee, and Rojas-SuBrez, 1992). The recent perhaps the only Latin American market that has de- banking crisis in the industrial world-Japan, the veloped a corporate bond market. As in the case of United States, and Scandinavia-indicates the diffl- the Mexican commercial paper, the impact of this culty of timing regulatory actions to prevent the in- market on the bank franchise must be measured in crease in bank risk. terms of the spread between corporate bond interest In Latin America, therefore, the main issue in rates and the marginal cost of funds facing banks. evaluating the capital market impact on the franchise Unfortunately, this spread cannot be measured as di- is whether the spread that banks can earn by holding rectly as the spread in the commercial paper market short-term loans to corporations on their balance in Mexico, but there is strong direct evidence that, if sheets has fallen to the point where banks seek alter- the spread between bond interest rates and the mar- native, risky investments. The spread available to banks holding corporate loans can be approximated 7UBanker's acceptances were an important funding instrument by comparing the cost to corporations of borrowing for banks in 1989 when rescrve requirements on deposits reached in the short-term money market with the marginal 100 percent. When these requirements were eliminated, the mar- cost to banks of funding corporate loans. ~ a n k s ' ket for banker's acceptances shrank substantially, and bank de- marginal funding costs are the costs that banks face posits regained their importance as a funding source for banks. 7'Moreover, the spread between commercial paper and short- in raising liabilities in the money markets, such as term deposits was about 400 basis points in early 1994. the wholesale market for certificates of deposit and 72This equity ratio is chosen to conform to the capital guide- the interbank market. If this spread is relatively high, lines of the Bank for International Settlements. banks can still make money by lending to large cor- 7'This calculation assumes that 92 percent of the loan is funded at the bank's marginal cost of funds, and the remaining 8 percent porations, and the balance sheet franchise is rela- of funds is raised in the equity market. Banks earned 13.5 percent tively safe. on the entire loan, which is the commercial paper rate. They paid The only market for which there are published 11.5 percent, the banker's acceptance rate, on 92 percent of the data on commercial paper rates and money market funds used to make the loan, as well as 75 basis points o l nonin- bank funding rates is Mexico. In Mexico, two terest expenses on the entire amount of the loan. Based on this revenue stream and the funding and noninterest costs, banks wholesale funding markets are available to banks: earned about 27 percent on the remaining 8 percent of funds used to make the loan, which represents the equity funding. 6Wie lunciion of providing liquidity is still important, but the 741t is also noteworthy that during the financial crisis in Mex- demand for bank loans to provide liquidity declines as capital ico, when interest rates on repurchase agreements secured by markets develop. Capital markets develop because markets are government paper rose to more than 80 percent, interest rates on liquid enough to settle payments at the end of the day with securi- bank deposits remained below 60 percent. Of course, only the ties rather than with "good funds." This argument is presented in most secure banks in the system were able to raise bank deposits detail in Section 11. under these conditions. Chart 17. Mexico: InternstRateSpreads on Selected Financial Instruments, March 1991-)rlatrh 1994 (In wj Corporate Bonds GDP ' (1) (2) (1)1(2) OoO Year (Billrons of pesos) (Percent of GDP) Commercialpaper minus --+- banker's acceptances 800 1980 2.0 1,075.3 0.18 1981 3.7 1,273.1 0.29 1982 21.1 1,239.1 1.70 1983 21.0 1,557.8 1.35 1 984 20.0 1,893.4 1.06 1985 29.1 2.576.6 113 1986 12.3 3,246.1 0.38 1987 32.6 4,159.8 0.78 1988 107.8 5.4 1 1.O 1.99 1989 242.2 6,778.4 3.57 1990 41 1.7 8.477.9 4.86 1991 659.9 10,939.2 6.03 1992 758.1 13,740.0 5.52 - Source: Banco de Mexico, lndicadores Econom~cos(November Sources: Ch~le.Superlntendenc~adevalores y Seguros, and IMF 1991 and March 1994). staff estimates. ginal funding costs of banks were available, it would be relatively large. TaMe II. ChiktRRSSOR Assetsby Fund As a percentage of GDP, the corporate bond mar- T pofIsrdngAgency, December 1993 ket grew rapidly during 1988-91 and has slowed (Pwceni~f pemion@n& wets) down recently (Table 10). In 1992, corporate bonds, which equaled about 5.5 percent of GDP, were al- most entirely held by pension funds and insurance Five Largest companies, with two-thirds of the volume held in IssuingAgency Funds' pension f~nds.~"ncontrast, pension funds and in- surance only held 17 percent of total equities. In Total government institutions Central Bank of Chile 1993, the top five pension funds held 74 percent of Treasury the total assets of all pension funds as well as 75 per- Other cent of the corporate bonds held by pension funds Financial institutions (Table 11).These figures indicate that the market for Enterprises corporate bonds is quite narrow and concentrated Investment in foreign entities Other and imply that the market is relatively illiquid-that Total assets is, a single trade probably moves the market price significantly. The illiquidity of the market should Source: Chile. Superintendenciadevalores y Seguros. tend to keep interest rates on corporate bonds rela- 'Cuprum, Habitat. Provida, Santa Maria, and Summa. tively high compared with a more liquid instrument like bank deposits. This would suggest that bank spreads and bank profits are still relatively immune to domestic capital market competition. U.S. market through ADRs, which trade on the A possible further threat to the franchise value of major stock exchanges.76In addition, some firms is- the banking system in Latin America is large corpo- sued medium-term notes, which are instruments rate access to the capital markets of the United similar to bonds but are not subject to the disclosure States and Europe. For example, before the 1994 cri- sis, large, well-known firms with stable revenue 76Foreign firms use A D R s t o become listed o n a U.S. stock ex- bases, such as utilities, raisea equity funds in the change without being subject to the Security and Exchange Com- mission's disclosure requirement. To issue ADRs, however, for- eign firms must h o l d a deposit w i t h a U.S. chartered bank to 7sFor a relative comparison, in the U n i t e d States, corporate guarantee payment o f dividends. Receipts verifying the existence bonds are about 19 percent o f GDP. of these deposits are the instruments that are actually traded. requirements of bonds, in the U.S. market. Despite America, because money markets are relatively illiq- their name, these instruments have varying maturi- uid, banks may not be willing to price a contract off ties, sometimes quite long77 the interbank rate. If, for example, in Mexico, a While access to U.S. markets for new funds is peso-dollar swap contract were priced off the Mexi- quite limited under current conditions, access at can commercial paper rate rather than off the inter- some future date cannot be ruled out. How much of a bank rate, the discount on pesos at maturity of the threat does U.S. market access pose to the franchise contract would be relatively large because the com- value of banks in Latin America? Since ADRs are mercial paper rate is higher than the interbank rate. similar to equity instruments, a strong domestic This would erode a large portion of the gain to rais- banking franchise is an important complement to ac- ing short-term funds in the dollar market.78 cess to the U.S. domestic market as it is to access to In Chile, the corporate bond market is a medium- domestic equity markets. The threat posed by access term market, which might seem to provide an inter- to the medium-term note market is best evaluated by est rate that can be used to price currency swaps for considering why Latin American firms are not signif- firms that issue medium-term notes in the United icant users of U.S. short-term money markets, such States. Again, the partial evidence presented above as the commercial paper market, unless they have a indicates that, like the Mexican commercial paper dollar-denominated cash flow. U.S. money markets market, this market is illiquid, implying that Chilean are highly liquid, and large U.S. corporations use medium-term interest rates include a liquidity pre- money market instruments rather than bank loans mium, which reduces the cost advantage of raising when cash outflows do not quite match cash inflows. funds in the more liquid U.S. market. It might seem relatively simple for a Latin American Based on the evidence of how fixed-income and telephone company, for example, to issue commer- equity markets in Latin America operate, it seems cial paper in the U.S. market and to engage in a cur- that a strong threat to the banking franchise from rency swap to cover its cash outflow in its local mar- capital markets is some years away. The most liquid ket. In this way, it would seem to be able to lock in a capital markets in the region are equity markets; low borrowing cost from a highly liquid market. even where fixed-income markets are operational- This transaction, however, is not as simple as it namely, the commercial paper market in Mexico and might first appear. A currency swap, like any futures the corporate bond market in Chile-spreads are contract, is priced according to the relative nominal high relative to bank costs and the paper is held by a interest rates between two markets. The currency in few investors. The growth of equity markets, how- which nominal interest rates are higher will trade at ever, poses some dangers for policymakers worried a discount at maturity of that contract relative to the about unstable capital inflows. currency where nominal interest rates are lower. Swap contracts are usually priced off money market 78During the recent financial crisis in Mexico, a Mexican firm interest rates, such as the interbank rate. In Latin experienced a well-publicized default on its dollar-denominated commercial paper, perhaps damaging other Latin American ??SeeCrabbe (1993) firms' access to this market as well. VI Financial Soundness and Macroeconomic Stability: A Bank Balance Sheet Approach Thisconsequences section focuses on the macroeconomic tacks,79 the second has not received sufficient of banks' performance. Specifi- attention in the economic literature. In addition, this cally, it analyzes the extent to which the sustainabil- section addresses the issue of the appropriate ity of a widely implemented type of stabilization holdings of foreign exchange reserves by central program in Latin America, namely, an exchange- banks. rate-based program, depends on the performance of As discussed in Section I1 and reinforced in Sec- the banking sector. This discussion complements tion V, although banks are a dominant feature in the those in previous sections, which used a sample of financial landscape of Latin America, there are also five countries to analyze how alternative policy significant differences in the degree of importance of environments in those countries affected the behav- banks and in the structure and organization of finan- ior and performance of banks, as well as the op- cial markets among countries in the area. The analy- tions available to the authorities to handle financial sis that follows, therefore, centers on two cases that difficulties. may best represent the alternative financial struc- Recent literature on the sustainability of adjust- tures of the region: (1) an economy where the finan- ment programs has focused on the destabilizing ef- cial sector is largely dominated by banks and where fects associated with the lack of credibility in the dollar-denominated deposits are an important com- permanence of the announced policies (e.g., see ponent of the banks' balance sheets (such as Ar- Calvo, 1991; Obstfeld, 1986; Flood and Garber, gentina and Peru); and (2) a financial sector where 1984a; Calvo and VCgh, 1990; and Rojas-SuBrez, some other long-term financial institutions, such as 1992).An often-cited example is that even when a pension funds, are important players in absorbing fi- long-run consistency exists between macroeco- nancial flows and where dollar-denominated de- nomic fundamentals and the announced exchange posits are not an important fraction of total bank de- rate policy, a lack of confidence in the persistence posits (such as Chile).80 By considering the specific of the announced policy may result in a speculative effects of dollarization, the analysis will also derive attack on the domestic currency. Because the attack conclusions for bank-dominated systems in which itself changes the fundamentals-notably, by ex- banks' deposits are largely denominated in the do- panding domestic credit if the authorities accommo- mestic currency. date the demands of speculators, or by raising do- In both cases, stock exchanges are assumed to mestic short-term interest rates if the authorities be a small, albeit growing, component of the finan- attempt to defend the exchange rate parity by in- cial sector. In conformity with the current policy creasing the cost of speculation-the government stance in most Latin American countries, it will be may find it too costly to defend the exchange rate assumed that the government does not allow the on a sustained basis. In such cases, the authorities nominal exchange rate to adjust freely to market may validate the attack and abandon the announced conditions.*' exchange rate. In the remainder of this section, it is argued that the degree to which a Latin American central bank may be able to withstand a speculative attack on 79See, for example, Flood and Garber, 1984b; and Krugman, 1979. its domestic currency-whether it involves a rever- 80Although the analysis could incorporate the use of any for- sal of the capital inflows or not--depends crucially eign currency, U.S. dollars are, by far, the most commonly used on two factors: (1) the extent of the commitment currency in which bank deposits in foreign currency are denomi- of the central bank to stabilize prices in the finan- nated in those countries where foreign currency deposits are cial sector; and (2) the strength of the banking allowed. 81Thisis true even in those Latin American countries with the sector. While the first factor is a natural extension lowest degree of government intervention in foreign exchange of the well-known literature on speculative at- markets, such as Peru. Bank-Intermediated System Facing it would trigger what is commonly known as capital Significant Dollarization flight.85 Expectations of an exchange rate devalu- ation could also induce capital flight in a sound In the financial system to be analyzed here, the bank-dominated financial system where deposits de- regulatory framework andlor domestic economic nominated in a hard foreign currency-the U.S. dol- conditions prevent banks from facing significant lar-are not allowed. competition from other financial intermediaries. Although the stock of foreign exchange reserves This system, which characterizes many Latin Ameri- in the central bank is important for assessing the ef- can countries, can persist, even when constraints on fects of a possible reversal of the capital inflows on credit allocation or interest rate controls are re- the exchange rate, its importance is associated with moved. An explanation for this occurrence is that the degree of dollarization in the economy. In the ex- high volatility in the net revenues of domestic firms treme case of a fully dollarized economy, an attack complicates their issuance of commercial paper and on the domestic currency is meaningless because the bonds or their placement of equity share~.~2As dis- domestic currency no longer plays any significant cussed in Sections I1 and V, commercial paper and role in the system. However, agents holding U.S. corporate bonds are practically nonexistent in most dollar deposits in domestic banks may demand to Latin American countries (Chile being the exception convert them into liabilities issued by U.S. authori- for corporate bonds), whereas stock markets are ties that are held by the central bank.86In contrast, in booming, although the number of listed shares is an economy that is less than 100 percent dollarized, very limited. As a result, bank credit remains the the stock of foreign exchange reserves plays a role in sole source of credit for the majority of firms. maintaining an announced exchange rate regime: an Dollarization is defined here in its broadest form, attack on the foreign exchange reserves of the cen- namely, to indicate that the U.S. dollar is used not tral bank would involve short-term bank liabilities only as a store of value, but also as a unit of account denominated in domestic currency. As in the former and the medium of exchange.83 example, agents may also demand to convert U.S. Assets issued by residents of a particular country dollar-denominated deposits into U.S. dollar bills. are subject to two kinds of risk: (1) the risk of large As will be discussed below, however, a speculative losses in the real value of assets denominated in the attack on the central bank foreign exchange reserves domestic currency as a result of economic policies involving bank liabilities denominated in U.S. dol- that lead to rapid inflation or to large exchange rate lars will occur only if the public fears that its dollar- depreciations; and (2) the default risk associated denominated deposits are no longer safe in the do- with solvency problems in the issuing institutions or mestic banking system. with the expropriation of domestic assets.84If the The decision of how large the stock of foreign ex- banking sector is perceived as sound and bank prof- change reserves should be is perhaps one of the key itability is viewed as having a high priority among policy decisions of central bankers in Latin America. the authorities' goals-effectively ruling out the The discussion in Sections 111and V made it clear probability of expropriation of deposits because that a large net foreign asset position held,by a cen- such an action would hamper banks' stability-a tral bank carries with it two risks: first, it increases lack of confidence in the stability of the exchange the temptation of governments to delay the correc- rate would trigger a switch into bank liabilities de- tion of policy inconsistencies; second, it provides nominated in U.S. dollars. By comparison, a lack of central banks facing an adverse shock in the domes- confidence in the soundness of the banking system tic economy with the option of simply selling those would induce depositors to move their deposits de- foreign assets and expanding credit rapidly, with ad- nominated in either the domestic currency or U.S. verse consequences for the soundness of the banking dollars outside the domestic financial market; that is, system. This is particularly true when the central bank has a weak franchise value, as exemplified by the Peruvian experience during the mid-1980s. In 82SeeWeisbrod and Lee (1993) for a detailed discussion of this issue. other words, too many foreign exchange reserves in 83The need to clarify the use of the concept is important be- the hands of central banks may limit the amount of cause there is no agreement on the meaning of the term, and some market discipline that authorities feel the need to authors (Calvo and Vkgh, 1992) would refer to t h ~ sbroad use of the foreign currency as "currency substitution." g4See Rojas-Suirez, 1991. Although the two types of risk are 85Notice that funds need not flow outside the country; they'may distinguished for analytical purposes, a situation may arise in well remain in an informal financial market dealing in U.S. which both risks become indistinguishable. For example. expec- dollars. tations of a devaluation-which increases the first kind of risk- 86lnvestors may also want to convert other assets denominated may increase the expectation that the authorities, in an attempt to in U.S. dollars into U.S. dollar bills. They would be successful, defend the domestic currency, would freeze dollar-denominated however, only to the extent that the central bank guaranteed con- bank deposits-which increases the default risk. vertibility of those assets. exert in making policy decisions. Moreover, when Lack of Confidence in the Exchange Rate Policy, domestic interest rates (adjusted for changes in the but BanksAre Perceivedas Solvent exchange rate) are higher than international rates for In an economy where U.S. dollar bank accounts risk-free assets, holding foreign exchange reserves is are allowed and the banking system is perceived to costly for the authorities because the marginal cost be sound, an exchange rate policy that lacks credibil- of central bank debt is higher than the marginal rev- ity does not need to generate flight from the domes- enue of holding foreign exchange. tic financial system; rather, the public will be in- The discussion above suggests that there is a clear duced to convert its domestic-currency-denominated trade-off associated with holding foreign exchange deposits into deposits denominated in U.S. dollars. reserves: on the one hand, they provide the resources In addition, speculators taking positions against the to defend an exchange rate parity; on the other hand, domestic currency will increase their demand for they give rise to some risks and costs for sound poli- loans denominated in domestic currency in the ex- cymaking, including delaying reforms of the bank- pectation that the devaluation will bring about a cap- ing sector. Aware of this problem, authorities in ital gain. However, if the central bank is committed many Latin American countries are taking a policy to keeping the monetary base fully backed by for- stance regarding their desired holdings of foreign eign exchange reserves, the central bank will not exchange reserves in a way that, according to their validate the expansion of credit denominated in do- perceptions, minimizes such a trade-off. Indeed, mestic currency (i.e., it will not provide banks with their position regarding sterilization practices is a re- reserves to allow the expansion of credit);sg instead, flection of these choices. In the following two sub- domestic currency interest rates will increase.90 sections, these choices are addressed. Although central bank policy will prevent the ex- tension of credit denominated in domestic currency, Monetary Base Fully Backed by Foreign the public may be able, to a large extent, to convert Exchange Reserves its domestic-currency-denominated de~ositsinto U.S. dollar deposits.9' In the extreme, the public One of the central bank's choices is to maintain a may want to have all its deposits (and cash holdings) ratio of the stock of (gross) foreign exchange re- denominated in U.S. dollars.g2 Assuming that re- serves to the monetary base equal to or greater than serve requirements on domestic-currency-denomi- one. This decision was made explicit in Argentina's nated deposits equal those on foreign currency de- convertibility law of March 1991, but it has been im- posits, banks will need to convert the value of their plicitly followed in many other Latin American cash assets (for reserve requirements purposes) de- countries during the early 1990s, including three of nominated in domestic currency into U.S. dollars. As the countries analyzed in this paper.87 The question a result, the domestic monetary base will be extin- to be addressed is to what extent the solvency of a guished, but the reduction in the stock of foreign ex- banking system is important for determining the sta- change reserves held at the central bank will be lim- bility of a preannounced exchange rate in a dollar- ited to the initial stock of domestic currency in ized economy where the monetary base is fully circulation (i.e., currency held by the public andout- backed by foreign exchange reserves. To analyze the importance of a solvent banking system for the stability of a preannounced exchange 89In a fractional reserve requirement system, the monetary base rate, assume that an adverse shock-which will re- includes reserve requirements on deposits denominated in domes- tic currency, other bank deposits in the central bank, and domestic main unspecified here-generates a lack of confi- currency in circulation. dence in the sustainability of the announced poli- gOLikewise,by not providing an expansion of domestic cur- cies. Two scenarios will be considered. First, the rency to satisfy an increase in banks' reserve requirements, the shock produces a generalized expectation of a de- monetary authorities would make it too expensive to reschedule valuation, but banks are perceived to be sound.88 existing loans denominated in U.S. dollars into loans denomi- nated in domestic currency. Second, the lack of confidence in the exchange rate 9'Converting domestic-currency-denominated time and saving is accompanied by a lack of confidence in the bank- deposits into U.S. dollar deposits usually carries a cost. It is as- ing system. sumed here either that those costs are negligible or that they are outweighed by the expectation of large losses in the real value of those deposits arising from a devaluation. 9?Although this is a highly unlikely case if interest rates are allowed to fluctuate freely, the example is used here to evaluate 87Bythe end of 1993, the ratio of gross foreign exchange re- the capacity of the system to withstand flight from assets de- serves in Colombia, Mexico, and Peru was about 1.5. In contrast. nominated in domestic currency. In a more realistic scenario, the Chile kept a smaller ratio of 0.8. people induced to shift into U.S. dollar assets would be those 8% the remainder of this subsection, it is assumed that the ex- whose expectations of the size of the devaluation exceed that change rate equals one. This simplifies the exposition greatly contained in the interest rate on domestic-currency-denominated without affecting the analysis. deposits. side the banking system).93 The monetary system, crease in the demand for domestic-currency-denom- that is, currency and deposits held by the public, will inated loans will reverse because the real interest become, at least temporarily, fully dollarized. rate will decline. Even in an extreme case when people prefer to Alternatively, uncertainty regarding expectations move out of the domestic currency altogether, full may lead people who are convinced of the authori- dollarization will not persist if (1) there are some ties' commitment to believe that others remain un- preexisting private contracts requiring payments in convinced and that a further speculative attack may domestic currency; or (2) the authorities demand occur in the future (resulting in additional increases that certain payments take place in domestic cur- in the interest rate on domestic loans). These percep- rency, such as taxes and trading involving govern- tions will affect the process of dollarization. The ini- ment bonds denominated in domestic currency. Al- tial increase in the interest rate, following a specula- though the demand for domestic currency to make tive attack, hits not only speculators but also payments on preexisting domestic-currency-denom- borrowers in domestic currency whose loans mature inated contracts may be temporary (it may vanish as at the time of the attack. Borrowers convinced of the the contracts expire), the demand for domestic cur- monetary authorities' commitment may fear the pos- rency to comply with authorities' regulations will be sibility of unanticipated future speculative attacks, permanent. Therefore, if conditions (1) and/or (2) because the resulting increase in interest rates will hold, the monetary base will not vanish (or may do affect their maturing loans. Those borrowers may so for a very brief period) following the sudden loss find it beneficial to shift the currency composition of of credibility in the announced exchange rate policy; their liabilities to the banks into U.S. dollar loans. it will, however, be reduced significantly. That is, in this plausible outcome, different percep- An example of the changes in the balance sheets tions regarding the credibility of the monetary au- of banks and the central bank that may follow imme- thorities' defense of the domestic currency will re- diately after a sudden loss of confidence in the ex- sult in a strengthening of the dollarization process. change rate regime is shown in the appendix. Fol- In this example, the persistence of dollarization re- lowing the conversion of domestic currency deposits sults from a lack of information regarding expecta- into U.S. dollar deposits, banks may find themselves tions. That is, dollarization can strengthen even if with a mismatch in the currency denomination of everybody individually believes in the authorities' their assets (domestic currency and dollars) and lia- commitment while thinking that others are unlikely bilities (only dollars in the example under considera- to believe in such a commitment. Although this situ- tion). However, the same factors that would make ation cannot persist in the long run, it can lengthen the dollarization a temporary process would also the period during which borrowers prefer dollar- make the currency mismatch temporary: as preexist- denominated loans. ing stocks of loans denominated in domestic cur- The discussion above assumes that the authorities rency mature and as the need to undertake transac- are able to use the interest rate defense as a tool to tions that must be effected in domestic currency respond to speculative attacks. Some empirical ex- leads to the conversion of some U.S. dollar deposits amples suggest, however, that in some cases the au- into domestic currency, the currency mismatch thorities perceived the costs associated with such in- would disappear. The correction of the currency mis- creases in the real interest rates as exceeding the match is shown in the example presented in the ap- perceived benefits from defending the exchange rate pendix. That example also shows a plausible new parity, and, as a result, a devaluation f0llowed.9~ equilibrium where, compared with the above exam- In general, the costs to the authorities of increas- ple, the monetary base has fallen drastically and the ing interest rates on domestic currency assets are dollarization process has strengthened significantly. well known: the government's and businesses' fi- An interesting question remains, however: once nancing costs increase, which may lead to a decline the overhang of loans denominated in domestic cur- in output and employment. In addition, the increase rency matures, will borrowers engage in further bor- in interest rates exerts a downward pressure on the rowing denominated in domestic currency for busi- price of nonbank assets, including the equity market. ness purposes? Although there is no definite answer In the case under consideration, to the extent that to this question, the dynamics of expectations behav- some transactions need to be carried out using do- ior sheds light on this issue. Clearly, if the authori- mestic currency, there could be a short-term decline ties are able to convince the public at large of their in real activity following the temporary rise in real commitment to the exchange rate, the initial de- 'jThe exchange rate crisis of the European Monetary System 93Thisis SO because reserve requirementson domestic currency during the fall of 1992 offers a good example of the costs associ- deposits held at the central bank will be converted into reserve re- ated with using the interest rate defense to protect the exchange quirements on foreign currency deposits also held at the central rate parity. For a detailed analysis of this episode, see Interna- bank. tional Monetary Fund (1993). interest rates. Moreover, the default risk faced by currency, dollarization minimizes the adverse impact banks may increase following the speculative attack of an increase in domestic currency interest rates on as the increase in domestic currency real interest the financing costs of firms and, therefore, mini- rates may adversely affect some borrowers' ability to mizes the pressure on the authorities to expand pay. If, as assumed, banks are sound, they will be credit in domestic currency. well capitalized and will be able to withstand the temporary increase in the default rate without the Lack of Confidence in the Exchange Rate Policy support of the monetary authorities. and in the Soundnessof the BankingSystem The exchange rate parity can be preserved only if the authorities stand ready to accept the possible ef- In this case, an adverse shock generates a lack of fects on the real economy of an increase in domestic confidence not only in the exchange rate system, but real interest rates and avoid the temptation to expand also in the stability and soundness of the banking sys- credit in domestic currency, thereby abandoning the tem. Now, the public has an incentive to shift all of full backing of the monetary base with international its deposits-in domestic and foreign currency--out reserves. Dollarization can, however, help minimize of the domestic banking system. Because, in the case the well-known trade-off between real activity and considered here, there are no domestic alternatives in exchange rate stability. The faster firms can adjust which the public can keep its financial savings, there the currency composition of their liabilities-that is, is an incentive to transfer the funds outside the coun- the faster they can borrow in U.S. dollars rather than try-that is, an incentive for capital flight. in domestic currency-the more limited the effect If perceptions are right and the banking system is on the real sector.~~ollarizationwould be particu- not sound, the authorities can do little to prevent a larly important for those firms whose access to fi- sustained flight from the banking system (with the nancial markets is limited to the domestic banking exception of freezing deposits). If banks are not sol- system. In the absence of dollarization, these firms vent, they will not be able to use their assets to sat- would have to face the higher domestic real interest isfy the public demand to cash in its deposits-non- rates: the higher the financing costs, the more con- performing loans cannot be liquidated to match the tractionary the effect of using the interest rate to de- closing of deposit accounts. In this circumstance, the fend the exchange rate parity.96 central bank would face strong pressure to bail out The discussion in this section has made it apparent banks by extending credit to them. If the central that in a bank-dominated system where liquid assets bank were to expand the monetary base to extend are largely bank liabilities, the authorities may be credit to banks, however, the central bank would de able to successfully defend its parity against a specu- facto be abandoning its commitment to fully back lative attack if the banking system is sound and is the monetary base with international reserves; this, perceived to be sound and if the monetary base is in turn, would significantly hamper the central fully backed by foreign exchange reserves. In this bank's ability to satisfy the demand for foreign ex- process, a plausible outcome is that dollarization is change reserves at the announced exchange rate. strengthened either if the authorities fail to convince In contrast, a solvent banking system has a better the public of their commitment to the announced ex- chance of withstanding a "lack of confidence crisis." change rate policy or if some economic agents be- First, loans in good standing can be sold to pay bank lieve that further speculative attacks could occur in depositors if there is a secondary market for bank the future." Allowing U.S. dollar-denominated ac- loans, or used as collateral to obtain credit from for- counts in the banking system is crucial for these re- eign institutions, or, if there is room for monetary sults: by providing a source of financial investment expansion without violating the assumption that the denominated in foreign currency, dollarization pre- monetary base is fully backed by foreign exchange vents capital flight following a sudden loss of confi- reserves, used to obtain credit from the central bank. dence in the exchange rate. Also, by providing a When the system is solvent, the degree of maturity source of domestic credit denominated in foreign matching between assets and liabilities as well as banks' access to additional sources of liquidity are 9Thispossiblity is of course related to the maturity structure of important to the success of the banks faced with a the existing stock of domestic-currency-denominated debt at the run.98Second, if banks are able to match the outflow time of the speculative attack. of deposits with bank assets, and if the central bank Wikewise. the net fiscal effect of a temporary increase in inter- est rates on domestic assets will be minimized if the government is does not expand domestic credit further, the maxi- able and willing to shift the currency composition of its debt. "The authorities' alternative is to increase the number of trans- yWbviously, banks would not keep a perfect currency and ma- actions that have to be carried out in domestic currency to the ex- turity match between bank assets and liabilities. Likewise, how- tent that they force de-dollarization. As discussed above, how- ever, it would be extremely unlikely that the run would dry up ever, this alternative would motivate capital flight even when the bank liquidity 100 percent, especially in an economy fully depen- banking system is perceived as sound. dent on banks for financial transactions. mum amount of demand for central bank foreign ex- Would the results be affected by the inflow of change reserves would equal the monetary base plus large amounts of foreign capital before the crisis? If dollar-denominated reserve requirements.99Because the banks were sound and stayed solvent and prof- the monetary base is fully backed by foreign ex- itable during the inflows of capital-that is, if they change reserves, the central bank will be able to sat- did not increase the risk of their loan portfolio- isfy the demand for foreign currency at the estab- there should be no significant difference with the re- lished exchange rate.Io0 As the discussion so far sults discussed above. The size of the banks' balance indicates, bank soundness complements exchange sheets would be larger, as would the size of the tem- rate stability. porary reversal of flows, but the analysis would re- As in the first scenario, the central bank's commit- main unchanged. ment to back the monetary base with international If banks were not sound (or had a weak franchise reserves would imply an increase in the domestic value, to use the terminology of the previous sec- currency real interest rate following the increased tions), the previous capital inflow would make a dif- demand by speculators for liabilities denominated in ference: bankers who were not prepared to price risk domestic currency. Also, as in the first scenario, the correctly might have used the inflows of capital to increase in real interest rates may increase the de- extend risky loans that, in the presence of an adverse fault rate faced by banks. If banks are initially well shock, would turn into nonperforming loans. If a capitalized, they can absorb the losses; otherwise, bank run were to occur, the authorities would have banks that were solvent before the run would be- to face the additional pressure of bailing out failing come insolvent, inducing a further run out of the do- banks; the amount of central bank credit necessary mestic banking system. The key, therefore, is appro- to do so would increase in proportion to the size of priate bank capitalization such that a sound bank the capital inflows used to extend bad loans. At this would remain sound following an unexpected in- point, it is useful to recall the discussion on steriliza- crease in the default rate. tion contained in Section V. As noted there, steriliza- There is an additional similarity with the first sce- tion seems unnecessary if the banking system is sol- nario: if banks are indeed solvent, the authorities will vent because the size of the capital inflows should be able to maintain the exchange rate parity only if make no significant difference. However, a steriliza- they can resist the temptation to expand the monetary tion policy seems to deserve strong consideration if base to avoid the costs associated with the increase in banks are not yet strong enough to make sound the interest rates (even when the increase is tempo- credit decisions and the authorities fear a bank run rary). There is, however, an important difference with before bank restructuring is completed. In this case, the first scenario. Because the disintermediation from as discussed in Section V, curtailing the expansion of the domestic banking sector would include both do- banks' balance sheets through sterilization seems ap- mestic and foreign deposits, firms that depend com- propriate. Although the costs of sterilizing are well pletely on domestic banks would not be able to rede- known,lO' they may be outweighed by the costs of a nominate the currency of their loans because bank run in the context of bad credit decisions. Ster- liquidity in both currencies would have dried up fol- ilization could then be used as a temporary measure lowing the speculative attack; those firms would then to allow enough time for banks to consolidate their have to pay the increased financing costs in full. If operations. It is important to stress, however, that the process is understood to be temporary, these firms sterilization will yield the desired results only if the would have a greater incentive than before to delay monetary authorities are in a better position than the production plans to avoid the increased financing banking sector to allocate financial resources. lo2 costs. The temporary effects on output would, there- fore, be greater than in the first scenario. Likewise, Monetary Base Partially Backed by Foreign government domestic financing costs would be Exchange Reserves greater. Notwithstanding these increased costs, if banks proved to be sound, the process of bank disin- The discussion above suggests that a stock of for- termediation would tend to reverse itself. eign exchange reserves smaller than the monetary base may weaken the central bank's ability to defend the exchange rate if a lack of credibility crisis 99Total net liabilities of the central bank can be approximated emerges. This is the typical case analyzed in the tra- as the monetary base plus other net liabilities denominated in for- eign currency, which, for the purpose of this discussion, is as- ditional literature of speculative attacks (see Flood sumed to equal U.S. dollar-denominated reserve requirements and Garber, 1984a; and Krugman, 1979) and does (see the example in the appendix). 1MFor this to be true, foreign exchange reserves need to be ac- counted for net of U.S. dollar bank deposits in the central bank to 'OISeeCalvo (1991). satisfy reserve requirements because investors may convert the '020n the appropriate choice of sterilization method, see the entire stock of domestic currency into dollars. discussion in Section V. not need an extended discussion. Two main conclu- lending side, banks remain the most important source sions are appropriate to the example analyzed here. of credit; with the exception of Chile, very few firms First, even if there is no lack of confidence in the in other countries issue bonds in the domestic mar- solvency of the banking system-such that the pub- kets. Moreover, the stock exchanges in Latin Amer- lic would want only to change the currency denomi- ica, although showing rapid increases in share prices, nation of its deposits-a speculative attack on the have a limited number of listed companies.'% domestic currency may force the monetary authori- It is nonetheless important to recognize that the ties to abandon the exchange rate because, in a mas- quick expansion of equity markets may create pres- sive attack on the currency, the central bank may not sures on the government to protect the real value of have enough foreign exchange reserves to satisfy de- share prices, especially if the development of more mands from the public to convert domestic currency sophisticated and diversified capital markets be- holdings into U.S. dollars. The higher the stock of comes a government priority. Indeed, a current pol- domestic currency assets held by the public (includ- icy discussion among policy regulators in Latin ing cash and bank deposits), the greater the pressure America concerns the extent of government involve- on the central bank's foreign exchange reserves.10" ment in developing capital markets. To keep the Defending the exchange rate would, therefore, analysis relevant for Latin America, this example most likely involve a greater increase in the domes- will assume that banks remain the most important tic currency real interest rate than if the monetary source of liquidity in the system but that long-term base were fully backed by foreign exchange re- assets, such as bonds and equity, compete with bank serves. As discussed above, this would increase the loans as sources of credit. Also, bonds are assumed costs of defending the parity and might increase the to be denominated in domestic currency. authorities' temptation to abandon such efforts.104 Second, if there were to be a loss of confidence in Lack of Confidence in Exchange Rate Policy the domestic banking system, the pressures for a de- valuation would probably be greater if the stock of Assume, as before, that banks are sound and are foreign exchange reserves were smaller than the mon- perceived to be sound, but that an adverse shock re- etary base, even if the banks were solvent and could sults in a speculative attack on the domestic cur- use their assets to cash out the closing of deposits.105 rency. What difference does the introduction of fi- nancial assets competing with bank loans make to the analysis? The key difference is that it imposes A More Diversified Financial System additional constraints on the monetary authorities and may therefore weaken their capacity to defend The limitations on the diversification of the finan- the exchange rate. cial system in Latin America were discussed in Sec- The public would want to shift from domestic- tion 11. Indeed, as mentioned above, in countries currency-denominated assets to foreign-currency- where banks' relative importance in the financial sys- denominated assets. In addition to bank deposits, the tem has declined, the only significant competition in public now has a more diversified portfolio of assets absorbing savings comes from pension funds. On the that it may want to convert into foreign-currency- denominated assets. Pension fund managers, as well 10"f the reserve requirement on U.S. dollar deposits is smaller as the general public, would then have an incentive than that on domestic currency deposits, banks may find them- to transform bonds and, probably, equity into foreign selves with excess reserves following the conversion of domesfic- currency-deposits into U.S.dollar deposits. Banks may, therefore, currency assets (cash or foreign currency deposits).l07 be willing to extend further credit in domestic currency. and the central bank may find that it is asked to convert these new domes- tic currency deposits into U.S. dollars. Further pressure on the ex- "KInternational placement of bonds and equity in the interna- change rate would then follow. tional capital markets by private Latin American companies is 104Notwithstandingthese additional pressures, as in the case improving in terms of both the number of issues and the con- when the monetary base is fully backed by foreign exchange re- tracts. See International Monetary Fund (1992). Once again, serves, the possibility of substituting among the domestic and for- however, these markets remain small and confined to a select eign currencies may minimize the impact of the increase in do- number of large companies with international standards. mestic currency real interest rates on the firms' and government's I07If there are bonds denominated in foreign currency, the pub- financing costs; this may, therefore, minimize the costs of defend- lic may be satisfied to change the currency composition of its as- ing the exchange rate. sets and there may not be a shift toward bank deposits denomi- Io5Asdiscussed above, if (1) the banks can liquidate assets to nated in foreign currency. Because equity is a real asset, it is not match the closing of deposits, and (2) the authorities do not ex- clear that the demand for it would decline following adverse ex- pand domestic credit further, the maximum demand for foreign pectations about the exchange rate. The most affected issuing exchange reserves would equal the monetary base plus reserve re- firnls, however. would be those whose liabilities are mostly de- quirements on U.S. dollar deposits. This demand would not be nominated in U.S. dollars but whose earnings are denominated in fulfilled if the monetary base exceeded the stock of foreign ex- domestic currency (an example could be a firm producing a non- change reserves. tradable good using imported inputs). The outcome of this attack on the domestic cur- Lack of Confidence in Both Exchange Rate rency depends once more on the monetary policy Policy and Soundness of Banking System followed by the central bank. If the monetary base is Can the availability of assets competing with required to be fully backed by foreign exchange re- banks' liabilities help prevent capital flight during a serves at all times, the public as a whole will not be bank run? Only if the alternative assets are denomi- able to convert its bonds and equity holdings into nated in foreign currency. This is just an extension U.S. dollar-denominated bank deposits. To under- of the discussion above and needs no further com- stand this result, it is necessary to realize that for an ment."() However, even if asset alternatives to bank economic agent to sell a bond or equity denominated deposits were denominated in foreign currency and in domestic currency, some other agent must give up capital flight was avoided, the stability of the ex- a bank deposit denominated in foreign currency. change rate would still depend on the strength of the This scenario is not possible in the aggregate be- banking system. Just as in the case where banks were cause it contradicts the fact that, in the aggregate, the the only financial institutions, the decline in the de- public wants to reduce its holdings of domestic cur- mand for assets denominated in domestic currency- rency assets. Two outcomes are then possible: (1) cash and bank deposits-would imply a reduction in the monetary base will not expand. and bond prices, the stock of foreign exchange reserves held at the and perhaps equity prices, will fall, or (2) the mone- central bank and, therefore, would put pressure on tary base will expand to prevent a decline in asset the announced parity. In other words, the pressure on prices. The latter would imply a de facto abandon- the exchange rate does not depend on whether the ment of the full backing of the monetary base and foreign currency stays in the economy or leaves it would validate the attack on the exchange rate; this, through capital flight; rather, it depends on whether in turn, would significantly weaken the central international reserves leave the central bank. bank's ability to defend the parity. The outcdme described in ( I ) would increase the probability that the monetary authorities could main- tain their exchange rate policy. The downside of this Results - outcome, however, is that the decline in asset prices, This section has used a simple balance sheet ap- by reducing the real wealth of their holders, could proach to analyze the role of a sound banking system increase the probability of default faced by banks if in allowing the central bank to maintain the an- asset holders were also banks' debtors.lo8 If the au- nounced exchange rate policy following a specula- thorities perceived that the decline in asset prices tive attack on the domestic currency. A summary of could lead to problems in the banking system, they the results follows: could have an incentive to abandon the exchange Dollarization may be an ally for governments rate policy. That is, a more developed financial sys- pursuing exchange-rate-based stabilization pro- tem imposes greater obstacles to the defense of an grams but only if the banking system is sound. Dol- exchange rate even if the banking system is sound larization may prevent capital flight if a lack of con- and if the monetary base is fully backed by foreign fidence in the sustainability of the economic exchange reserve^.^()^ This can be interpreted either program leads agents to substitute away from bank as an argument against fixed exchange rates. or as deposits and other assets denominated in domestic one to keep most financial transactions under the currency into those denominated in foreign currency. roof of banks, at least during the transition period to- Dollarization also minimizes the impact on output ward macroeconomic stabilization-when loss of and employment resulting from a possible increase confidence crises are likely to occur. in the domestic currency interest rate that may fol- low the speculative attack, because domestic firms IoEBanks would also be directly affected if they were holding a may find it advantageous to shift from bank loans significant proportion of the assets whose relative prices had denominated in domestic currency to bank loans de- declined. nominated in U.S. d~llars.~'' 10qThelarger the stock of international reserves relative to the In sharp contrast, if a bank-dominated financial monetary base, the higher the probability of success for a central system is not sound but the public does not have in- bank defending a parity in a speculative attack. In the extreme. if foreign exchange reserves equal the monetary base plus the value of bonds and equity, a speculative attack is unlikely to bc huccess- ful. However, as discussed at the beginning of this section and in "(The currency denomination of capital markets instruments Sections 111 and V, keeping a high ratio of foreign exchange re- may deter capital flight onIy to the extent that the lack of confi- serves to the monetary base, which could be achieved through dence includes only the banking system and not the entire finan- high reserve requirements on U.S. dollar deposits or through ster- cial system. ilization, may entail high costs both for the real seclor-through "lor, to the extent that it is permissible by the legislators, from the effects of higher real interest rates-and for the banking sys- issuing bonds denominated in domestic currency to issuing bonds tem, as it may weaken the franchise value of banks. denominated in foreign currency. formation about banks' financial difficulties,dol- sound, sterilization is not only unnecessary but it larization may be a problem for governments under- also increases the financing costs for firms and the taking serious adjustment efforts. In this case, authorities. although a speculative attack on the domestic cur- Increasing the domestic real rate o f interest to rency may not be followed by capital flight, the defend the exchange rate may weaken the banking problem would not vanish as depositors would just system because it may lead to an increase in the de- shift the currency composition of their deposits fault rate faced by banks. I f banks are well capital- within the same troubled banks.112In the event o f a ized, however, the authorities may not face signifi- run on the banking system, pressures to rescue cant pressures to extend credit in order to rescue problem banks would increase the government's in- banks in trouble-that is, the better the banks' capi- centives to abandon the announced exchange rate talization, the greater the probability that the author- policy. ities will be successful in preserving the exchange Sterilization may be a useful policy in bank- rate parity. dominated financial systems facing a large inflow of A more diversified financial system imposes ad- foreign capital, but only as a temporary measure to ditional constraints on a government defending the give time to banks to strengthen their balance sheets. exchange rate after a speculative attack. For the ex- As discussed in Section V, if the banking system is change rate defense to be successful, asset prices would have to decline. I f a fall in asset prices is viewed as having a serious impact on the real sector '"The problem would compound if the reserve requirement on of the economy, the authorities may have a greater dollar-denorninated deposits were lower than that on domestic currency deposits. In that case, the balance sheets of troubled in- incentive to expand domestic credit and abandon stitutions would expand further. their exchange rate policy. Appendix Mechanics of a Speculative Attack on a Sound, Dollarized Banking System Assume that the authorities impose reserve require- Assets Liabilities ments on the following: domestic-currency-denominated deposits = k Banks Reserve requirements Deposits = 10percent Domestic currency = 0 Domestic currency = 0 foreign-currency-denominated deposits = j U.S. dollars = 30 U.S. dollars = 200 = 20 percent Loans Domestic currency = 90 Assume also that transactions that must be ef- U.S. dollars = 80 fected in domestic currency (payment for taxes, for Central Bank example) are such that the public needs to hold a Foreign exchange Monetary base = 0 minimum of deposits denominated in domestic cur- reserves = 30 Bank deposits in U.S. dollars = 30 rency equal to $10. For simplicity, assume that the public holds no cash, so that its entire financial wealth takes the form of bank deposits. The ex- Because the monetary authorities will not accom- change rate between U.S. dollars and domestic cur- modate a speculative attack on the domestic cur- rency is assumed to equal l. rency, the interest rate on domestic-currency-denom- An initial position can be characterized as follows inated loans will rise. The increase will prevent a (where bank capital has been netted out in the currency restructuring of domestic currency loans. accounts): In the above balance sheets, banks are not satisfy- ing reserve requirements in U.S. dollars, and, there- fore, that position is a disequilibrium one that could Assets Liabilities exist for only a brief time. The position also shows a mismatch in the currency composition of banks' as- Banks Reserve requirements Deposits sets and liabilities. As the existing stock of loans de- Domestic currency = 10 Domestic currency = 100 nominated in domestic currency expires and as the U.S. dollars = 20 U.S. dollars = 100 public need to hold a minimum of $10 in domestic- Loans currency-denominated deposits becomes binding, a Domestic currency = 90 U.S. dollars = 80 possible outcome may be: Central Bank Foreignexchange Monetary base = 10 Assets Liabilities reserves = 30 Bank deposits in U.S. dollars = 20 Banks Reserve requirements Deposits A lack of confidence in the announced exchange Domestic currency = I Domestic currency = I0 rate may lead to a shift away from the domestic cur- U.S. dollars = 29 U.S. dollars = 145 Loans rency. Hypothetically, if no other change occurs, the Domestic currency = 9 balance sheet may look as follows in the instant im- U.S. dollars = 1 16 mediately after the shift away from domestic Central Bank deposits. Foreign exchange Monetary base = I reserves = 30 Bank deposits in U.S. dollars = 29 In this outcome, the dollarization process has 'The dynamics toward the final position could take a variety of forms. The example presented here is chosen only for illustrative strengthened and will remain strong unless the pub- purposes. lic becomes convinced of the monetary authorities' commitment to the exchange rate. Also, owing to a posits, the new equilibrium involves a lower level of reserve requirement on U.S. dollar deposits higher total loans to the economy. Notice that if k =j, total than that on domestic-currency-denonlinated de- loans will have remained unchanged. References Baliiio, Tomiis J.T., "The Argentine Banking Crisis of Corrigan, E. Gerald, "Balancing Progressive Change and 1980," in Banking Crises: Cases and Issues, ed. by Caution in Reforming the Financial System," Federal Tomas J.T. Baliiio and V. Sundararajan (Washington: Reserve Bank of New York, Quarterly Review, Vol. International Monetary Fund, 1991). 16 (Summer 1991), pp. 1-12. Blommestein, Hans, and Michael Spencer, "The Role of Crabbe, Leland E., "Anatomy of the Medium-Term Note Financial Institutions in the Transition to a Market Market," Federal Reserve Bulletin, Vol. 79 (August Economy," in Building Sound Finance in Enlerging 1993), pp. 75148. Market Economies, ed. by Gerard Caprio, David Deutsche Bundesbank, Zahlenubersichten und Methodis- Folkerts-Landau, and Timothy D. Lane (Washington: che Erlauterungen Zur gesanltwirtschaftlichen Fi- International Monetary Fund, 1994). nanzierungsrechnung der Deutschen Bundesbank Calvo, Guillermo A., "The Perils of Sterilization," Staff 1960/89 (Frankfurt am Main, 1992). Papers, International Monetary Fund, Vol. 38 (De- Fama, Eugene, "What's Different About Banks?" Journal cember 1991), pp. 921-26. of Monetary Economics, Vol. 15 (January 1985), ,Leonardo Leiderman, and Carmen M. Reinhart, pp. 29-39. "Capital Inflows to Latin America: The 1970s Feldman, Robert A,, and Manmohan S. Kumar, "Emerging and the 1990s," IMF Working Paper No. 92/85 Equity Markets: Growth, Benefits, and Policy Con- (Washington: International Monetary Fund, October cerns," IMF Paper on Policy Analysis and Assess- 1992). ment No. 9417 (Washington: International Monetary (1993a), "Capital Inflows and Real Exchange Rate Fund, March 1994). Appreciation in Latin America: The Role of External Flood, Robert P., and Peter M. Garber (1984a), "Collaps- Factors," Staff Papers. International Monetary Fund, ing Exchange Rate Regimes: Some Linear Exam- Vol. 40 (March), pp. 108-51. ples," Journal of International Economics, Vol. 17 (1993b). "The Capital Inflows Problem: Concepts (August), pp. 1-13. and Issues," IMF Paper on Policy Assessment and (1984b), "Gold Monetization and Gold Disci- Analysis No. 93/10 (Washington: International Mon- pline," Journal of Political Economy, Vol. 92 (Febru- etary Fund). ary 1984), pp. 90-107. Calvo, Guillermo A., and Carlos A. VCgh, "Currency Sub- Garber, Peter M., and Steven R. Weisbrod, The Economics stitution in Developing Countries: An Introduction," of Banking, Liquidity, and Money (Lexington, Massa- Revista de Analisis Econcimico, Vol. 7 (June 1992), chusetts: D.C. Heath, 1992). pp. 3-27. International Finance Corporation, Emerging Stock Mar- ,"Credibility and the Dynamics of Stabilization kets: Factbook (Washington: International Finance Policy: A Basic Framework," IMF Working Paper Corporation), various issues. No. 901110 (Washington: International Monetary ,Quarterly Review of Emerging Markets, Fourth Fund, 1990). Quarter 1993 (Washington: International Finance Claessens, Stijn, and Sudarshan Gooptu, eds., Portfolio Corporation, 1993). lnvestment in Developing Countries, World Bank International Monetary Fund, lnternational Capital Mar- Discussion Paper No. 228 (Washington: World Bank, kets: Developments, Prospects, and Policy Issues 1993). (Washington: lnternational Monetary Fund, Septem- Corbo, Vittorio, and Leonardo Hernandez, "Macroeco- ber 1992). nomic Adjustment to Portfolio Capital Inflows: Ra- , Private Market Financing for Developing Coun- tionale and Some Recent Experiences," in Portfolio tries (Washington: lnternational Monetary Fund, lnvestment in Developing Countries, ed. by Stijn 1993). Claessens and Sudarshan Gooptu, World Bank Dis- , World Economic Outlook (Washington), various cussion Paper No. 228 (Washington, December issues. 1993). pp. 353-71. ,lnternational Financial Statistics (Washington), Corbo, Vittorio, and others, El Sistema Bancario Chileno: various issues. Desarrollos Recientes y sus Perspectivas, Instituto James, Christopher, "Some Evidence on the Uniqueness de Economia, Pontificia Universidad Catolica de of Bank Loans," Journal of Financial Economics, Chile (1992). Vol. 19 (1987), pp. 217-36. Krugman, Paul, "A Model of Balance-of-Payments Singh, Ajit, and Javed Hamid, Corporate Financial Struc- Crises," Journal of Money, Credit and Banking, Vol. tures in Developing Countries, IFC Technical Paper 11(August 1979), pp. 311-25. No. 1 (Washington: International Finance Corpora- Mathieson, Donald I., and Liliana Rojas-SuBrez, Liberal- tion, 1992). ization of the Capital Account: Experiences and Is- Sundararajan, V.. and TomBs J.T. Baliiio, eds., Banking sues, IMF Occasional Paper No. 103 (Washington: Crises: Cases and Issues (Washington: International International Monetary Fund, 1993). Monetary Fund, 1991). Mexico, Comisi6n Nacional Bancaria, Boletin Estadistico United States, Board of Governors of the Federal Reserve de la Banca Mdltiple (Mexico City, 1994). System, Balance Sheets for the U.S. Economy Morris, Felipe, M. Dorfman, J.P. Ortiz, and M.C. Franco, 198691 (Washington, 1992). Latin America's Banking Systems in the 1980s-A Velasco, Andres, "Liberalization, Crisis, Intervention: The Cross-Country Comparison, World Bank Discussion Chilean Financial System. 1975-85," in Banking Paper No. 81 (Washington: World Bank, 1990). Crises: Cases and Issues, ed. by TomBs J.T. Baliiio Obstfeld, Morris, "Rational and Self-Fulfilling Balance- and V. Sundararajan (Washington: International Mon- of-Payments Crises," American ~ c o n o m i cReview, etary Fund, 1991). Vol. 76 (March 1986). pp. 72-8 1. Weisbrod, Steven R., and Howard Lee, "Role of Financial Rojas-SuArez, Liliana, "Risk and Capital Flight in Devel- Intermediaries during Asset Deflations: Case Studies oping Countries," in Determinants and Systemic Con- of Japan, Korea, and Taiwan," in Portfolio Investment sequences of International Capital Flows, IMF Occa- in Developing Countries, ed. by Stijn Claessens and sional Paper No. 77 (Washington: International Sudarshan Gooptu, World Bank Discussion Paper No. Monetary Fund, 1991). 228 (Washington: World Bank, Howard Lee, 1993). ."An Analysis of the Linkages of Macroeconomic ,and LilianaRojas-Suilrez,"Bank Risk and theDe- Policies in Mexico," in Mexico: The Strategy to clining Franchise Value of the Banking Systems in Achieve Sustained Economic Growth, ed. by Claudio the United States and Japan," IMF Working Paper Loser and Eliot Kalter, IMF Occasional Paper No. 99 No. 92/45 (Washington: International Monetary (Washington: International Monetary Fund, 1992). Fund, June 1992). Recent Occasional Papers of the International Monetary Fund 132. Financial Fragilities in Latin America: The 1980s and 1990s, by Liliana Rojas-SuBrez and Steven R. Weisbrod. 1995. 131. Capital Account Convertibility: Review of Experience and Implications for IMF Policies, by staff teams headed by Peter J. Quirk and Owen Evans. 1995. 130. Challenges to the Swedish Welfare State, by Desmond Lachman, Adam Bennett, John H. Green, Robert Hagemann, and Ramana Ramaswamy. 1995. 129. IMF Conditionality: Experience Under Stand-By and Extended Arrangements. Part 11: Background Pa- pers. Susan Schadler, Editor, with Adam Bennett, Maria Carkovic, Louis Dicks-Mireaux, Mauro Mecagni, James H.J. Morsink, and Miguel A. Savastano. 1995. 128. IMF Conditionality: Experience Under Stand-By and Extended Arrangements. Part I: Key Issues and Findings, by Susan Schadler, Adam Bennett, Maria Carkovic, Louis Dicks-Mireaux, Mauro Mecagni, James H.J. Morsink, and Miguel A. Savastano. 1995. 127. Road Maps of the Transition: The Baltics, the Czech Republic, Hungary, and Russia, by Tapio 0 . Saavalainen, Biswajit Banerjee, Mark S. Lutz, Thomas Krueger, Vincent Koen, and Michael Marrese. 1995. 126. The Adoption of Indirect Instruments of Monetary Policy, by a Staff Team headed by William E. Alexan- der, Tomis J.T. Baliiio, and Charles Enoch and comprising Francesco Caramazza, George Iden, David Marston, Johannes Mueller, Ceyla Pazarbasioglu, Marc Quintyn, Matthew Saal, and Gabriel Sensen- brenner. 1995. 125. United Germany: The First Five Years-Performance and Policy Issues, by Robert Corker, Robert A. Feldman, Karl Habermeier, Hari Vittas, and Tessa van der Willigen. 1995. 124. Saving Behavior and the Asset Price "Bubble" in Japan: Analytical Studies, edited by Ulrich Baumgart- ner and Guy Meredith. 1995. 123. Comprehensive Tax Reform: The Colombian Experience, edited by Parthasarathi Shome. 1995. 122. Capital Flows in the APEC Region, edited by Mohsin S. Khan and Carmen M. Reinhart. 1995. 121. Uganda: Adjustment with Growth, 1987-94, by Robert L. Sharer, Hema R. De Zoysa, and Calvin A. McDonald. 1995. 120. Economic Dislocation and Recovery in Lebanon, by Sena Eken, Paul Cashin, S. Nuri Erbas, Jose Martelino, and Adnan Mazarei. 1995. 119. Singapore: A Case Study in Rapid Development, edited by Kenneth Bercuson with a staff team compris- ing Robert G. Carling, Aasim M. Husain, Thomas Rumbaugh, and Rachel van Elkan. 1995. 118. Sub-Saharan Africa: Growth, Savings, and Investment, by Michael T. Hadjimichael, Dhaneshwar Ghura, Martin Miihleisen, Roger Nord, and E. Murat U~er.1995. 117. Resilienceand Growth Through SustainedAdjustment: The Moroccan Experience, by Saleh M. Nsouli, Sena Eken, Klaus Enders, Van-Can Thai, Jorg Decressin, and Filippo Cartiglia, with Janet Bungay. 1995. 116. Improving the International Monetary System: Constraints and Possibilities, by Michael Mussa, Morris Goldstein, Peter B. Clark, Donald J. Mathieson, and Tamim Bayoumi. 1994. 115. Exchange Rates and Economic Fundamentals: A Framework for Analysis, by Peter B. Clark, Leonardo Bartolini, Tamim Bayoumi, and Steven Symansky. 1994. 114. Economic Reform in China: A New Phase, by Wanda Tseng, Hoe Ee Khor, Kalpana Kochhar, Dubravko Mihaljek, and David Burton. 1994. 113. Poland: The Path to a Market Economy, by Liam P. Ebrill, Ajai Chopra, Charalambos Christofides, Paul Mylonas, Inci Otker, and Gerd Schwartz. 1994. 112. The Behavior of Non-Oil Commodity Prices, by Eduardo Borensztein, Mohsin S. Khan, Carmen M. Reinhart, and Peter Wickham. 1994. 111. The Russian Federation in Transition: External Developments, by Benedicte Vibe Christensen. 1994. 110. Limiting Central Bank Credit to the Government: Theory and Practice, by Carlo Cottarelli. 1993. 109. The Path to Convertibility and Growth: The Tunisian Experience, by Saleh M. Nsouli, Sena Eken, Paul Duran, Gerwin Bell, and Zuhtu Yiicelik. 1993. 108. Recent Experiences with Surges in Capital Inflows, by Susan Schadler, Maria Carkovic, Adam Bennett, and Robert Kahn. 1993. 107. China at the Threshold of a Market Economy, by Michael W. Bell. Hoe Ee Khor, and Kalpana Kochhar with Jun Ma, Simon N'guiamba, and Rajiv Lall. 1993. 106. Economic Adjustment in Low-Income Countries: Experience Under the Enhanced Structural Adjustment Facility, by Susan Schadler, Franek Rozwadowski, Siddharth Tiwari, and David 0.Robinson. 1993. 105. The Structure and Operation of the World Gold Market, by Gary O'Callaghan. 1993. 104. Price Liberalization in Russia: Behavior of Prices, Household Incomes, and Consumption During the First Year, by Vincent Koen and Steven Phillips. 1993. 103. Liberalization of the Capital Account: Experiences and Issues, by Donald J. Mathieson and Liliana Rojas- SuLez. 1993. 102. Financial Sector Reforms and Exchange Arrangements in Eastern Europe. Part I: Financial Markets and Intermediation, by Guillermo A. Calvo and Manmohan S. Kumar. Part 11: Exchange Arrangements of Previously Centrally Planned Economies, by Eduardo Borensztein and Paul R. Masson. 1993. 101. Spain: Converging with the European Community, by Michel Galy, Gonzalo Pastor, and Thierry Pujol. 1993. 100. The Gambia: Economic Adjustment in a Small Open Economy, by Michael T. Hadjimichael, Thomas Rumbaugh, and Eric Verreydt. 1992. 99. Mexico: The StrategytoAchieve Sustained Economic Growth, editedby Claudio Loser and Eliot Kalter. 1992. 98. Albania: From Isolation Toward Reform, by Mario I. Blejer, Mauro Mecagni, Ratna Sahay, Richard Hides, Barry Johnston, Piroska Nagy, and Roy Pepper. 1992. 97. Rules and Discretion in Intemational Economic Policy, by Manuel Guitihn. 1992. 96. Policy Issues in the Evolving Intemational Monetary System, by Moms Goldstein, Peter Isard, Paul R. Masson, and Mark F'. Taylor. 1992. 95. The Fiscal Dimensions of Adjustment in Low-Income Countries, by Karim Nashashibi, Sanjeev Gupta, Claire Liuksila, Henri Lorie, and Walter Mahler. 1992. 94. Tax Harmonization in the European Community: Policy Issues and Analysis, edited by George Kopits. 1992. 93. Regional Trade Arrangements, by Augusto de la Torre and Margaret R. Kelly. 1992. 92. Stabilization and Structural Reform in the Czech and Slovak Federal Republic: First Stage, by Bijan B. Aghevli, Eduardo Borensztein, and Tessa van der Willigen. 1992. 91. Economic Policies for a New South Africa, edited by Desmond Lachman and Kenneth Bercuson with a staff team comprising Daudi Ballali, Robert Corker, Charalambos Christofides, and James Wein. 1992. 90. The Internationalization of Currencies: An Appraisal of the Japanese Yen, by George S. Tavlas and Yuzuru Ozeki. 1992. 89. The Romanian Economic Reform Program, by Dimitri G. Demekas and Mohsin S. Khan. 1991. 88. Value-Added Tax: Administrative and Policy Issues, edited by Alan A. Tait. 1991. 87. Financial AssistancefromArab Countries and Arab Regional Institutions,by Pierre van den Boogaerde. 1991. 86. Ghana: Adjustment and Growth, 1983-91, by Ishan Kapur, Michael T. Hadjimichael, Paul Hilbers, Jerald Schiff, and Philippe Szymczak. 1991. 85. Thailand: Adjusting to Success-Current Policy Issues, by David Robinson, Yangho Byeon,'and Ranjit Teja with Wanda Tseng. 1991. 84. Financial Liberalization, Money Demand, and Monetary Policy in Asian Countries, by Wanda Tseng and Robert Corker. 1991. 83. Economic Reform in Hungary Since 1968,by Anthony R. Boote and Janos Somogyi. 1991. Note: For information on the title and availability of Occasional Papers not listed, please consult the IMF Publications Catalog or contact IMF Publication Services.