w (Dco EE) ED n 23 n v ow sS o Esn z&~25 111k __________0___3___M_ _U_ ED Ge~ne 1°qX- Black December Banking Instability, the Mexican Crisis, and Its Effect on Argentina . ,a ,I. tA 6; f sF ti"~~ - U 6 WORLD BANK LATIN AMERICAN AND CARIBBEAN STUDIES Viewpoints Black December Banking Instability, the Mexican Crisis, and Its Effect on Argentina Valeriano F. Garcia The World Bank Washington, D.C. Copyright © 1997 The International Bank for Reconstruction and Development/THE WORLD BANK 1818 H Street, N.W. Washington, D.C. 20433, U.S.A. All rights reserved Manufactured in the United States of America First printing June 1997 This publication is part of the World Bank Latin American and Caribbean Studies series. Although these publications do not represent World Bank policy, they are intended to be thought-provoking and worthy of discussion, and they are de- signed to open a dialogue to explore creative solutions to pressing problems. Comments on this paper are welcome and will be published on the LAC Home Page, which is part of the World Bank's site on the World Wide Web. Please send com- ments via e-mail to laffairs'worldbank.org or via post to LAC External Affairs, The Aorld Bank, 1818 H Street, N.W., Washington, D.C. 20433, U.S.A. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author(s) and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Di- rectors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publi- cation and accepts no responsibility whatsoever for any consequence of their use. The boundaries, colors, denominations, and other information shown on any map in this volume do not imply on the part of the World Bank Group any judgment on the legal status of any territory or the endorsement or acceptance of such boundaries. The material in this publication is copyrighted. Requests for permission to reproduce portions of it should be sent to the Office of the Publisher at the address shown in the copyright notice above. The World Bank encourages dissemination of its work and will normally give permission promptly and, when the reproduction is for noncommercial purposes, without asking a fee. Permission to copy portions for classroom use is granted through the Copyright Clearance Center, Inc., Suite 910, 222 Rosewood Drive, Danvers, Massachusetts 01923, U.S.A. ISBN: 0-8213-3960-5 Cover: The painting on the cover, El Adorador Solar by Mexican artist Pedro Coronel, was provided by Christie's. Permission to reproduce it was granted by the Society of Mexican Authors of the Plastic Arts. Valeriano F. Garcia is principal economist in the Office of the Chief Economist in the Latin America and the Caribbean Regional Office of the World Bank. The author wishes to acknowledge helpful comments made by Saul Lizondo and V. Hugo Juan-Ramon. He also thanks Suman Bery, Jorge Canales, Allan Meltzer, and Guillermo Perry for their useful suggestions. Library of Congress Cataloging-in-Publication Data Garcia, Valeriano F. Black December: banking instability, the Mexican crisis, and its effect on Argentina / by Valeriano F. Garcia. p. cm. "April 1997." ISBN 0-8213-3960-5 1. Banks and banking-Latin America. 2. Capital movements-Latin America. 3. Balance of payments-Latin America. 4. Financial crisis-Mexico. 5. Mexico-Economic conditions-1994- 6. Argentina-Economic conditions-1983- I. Title. HG2710.5.A6G37 1997 330.982'064-dc2l 97-19841 CIP CONTENTS INTRODUCTION .....................................................1 Banking Blues .....................................................2 Capital Inflows .....................................................3 Changing Capital Flows and the Real Exchange Rate ..................................................... 3 I ANATOMY OF BANKING DISTRESS AND CRISES ................7 Financial Boom: Increased Demand for Money and Capital Inflows .....................................7 Financial Bust: Reduced Demand for Money and Capital Outflows .....................................8 High Interest Rates Affect Bank Portfolios .....................................................8 Changes in Relative Prices .....................................................8 Brewing the Crisis: Fractional Reserve Requirements and Deposit Insurance .......................9 The Perverse Asymmetry .....................................................9 ll DETERMINANTS OF CAPITAL FLOWS .......................................... 11 Increment in International Interest Rates: Unlikely Cause for the Mexican Crisis ............... 11 III SIZE AND COMPOSITION OF CAPITAL FLOWS .................... 13 Size of the Flows ................................................................... 15 Composition of the Flows ................................................................... 15 IV CURRENT ACCOUNT DEFICITS: NEITHER CURSE NOR BLESSING .................................................... 19 The Sustainability of Current Account Deficits .................................................... 20 V THE MEXICAN CRISIS .................................................... 23 The Exogeneity Hypothesis .................................................... 23 The Endogeneity Hypothesis .................................................... 24 Which is the Best Hypothesis? .................................................... 24 VI AFTERSHOCKS OF THE MEXICAN CRISIS: ITS IMPACT ON ARGENTINA .................................................... 27 What to Do in the Future? .................................................... 31 Unemploynment and the Real Exchange Rate .................................................... 31 VII PEGGED VERSUS FIXED EXCHANGE RATES: ARGENTINA AND MEXICO COMPARED .................................... 33 Summary and Conclusions .................................................... 36 NOTES .................................................... 37 BIBLIOGRAPHY .................................................... 39 111 TABLES Table 1. Ratio of Short-Term Debt Outstanding to GNP and to Export for Selected Latin American Countries, 1982 and 1994 .............................. 13 Table 2. Ratio of Total Debt and Debt Service Paid and (Due) to GNP for Selected Latin American Countries 1982 and 1994 ............................... 14 Table 3. Selected Debt and Financial Indicators for Selected Latin American Countries, 1981 and 1994 .................. 14 Table 4. Debt and Nondebt Flows to Selected Latin American Countries, 1975-93 ............. 15 Table 5. Average of Changes in the Current Account Balance, National Savings, and Domestic Investment in Selected Latin American Countries, by Period, 1975-93 ............. 20 Table 6. Average Change in Current Account Balance, National Savings, and Domestic Investment in Selected Latin American Countries, by Course of Change, 1975-93 .20 Table 7. Argentina: Fiscal Panorama, 1994 .28 Table 8. Mexico: Stock of Foreign Assets and Stock of Molnetary Base Causality Tests ............. 33 Table 9. Mexico: Changes in the Monetary Base and Changes in Foreign Assets Causality Tests .34 Table 10. Correlations .34 CHARTS Chart 1. Libor: Annual Rates .2 Chart 2. Mexico: Domestic Credit .24 Chart 3. Argentina:Total Deposits and Cash .29 Chart 4. Argentina: Portfolio Substitution .29 Chart 5. Argentina: Stock of Money in Dollars and Ratio MI to M3 .30 Chart 6. Argentina: Real Exchange Rate/Dollar Exchange Rate,Weighted by Argentinian Trade Pattern .................. 32 Chart 7. Mexico and Argentina: Ratio of Central Bank's Foreign Assets to Monetary Base ........... 34 Chart 8. Argentina and Mexico: Current Account Financed with Changes in CB's Foreign Assets ............... 35 Chart 9. Mexico: Income Velocity of Money and Growth of Money Base ....... 35 iv INTRODUCTION IN DECEMBER 1994 Mexico shocked the world and stunned the international financial community. A nation considered a model of economic reform and good financial health was suddenly bankrupt: Its international reserves had depleted, its currency was on free fall, it was about to default on its sovereign debt, and its banking system was on the verge of collapse. Uncertainty gripped the whole Latin American region. The international financial community burden to the taxpayer, particularly through the feared the effects of the Mexican crisis on other inflation tax. The result was that capital inflows Latin American countries. Earlier experiences reverted to open capital outflows, and inflation with large capital inflows into Latin America had surged. Moreover, income distribution worsened not concluded happily, either. The capital inflows because the inflation tax is highly regressive. In of the 1920s ended with the economic crisis of many instances, the same business sector that had the early 1930s, and the large capital inflows of been bailed out caused the capital outflows. the late 1970s ended with the debt crisis that Despite the history of banking instability in began in 1982. That crisis, marked by the Mexi- Latin America, the depth and scope of the can debt default, left many other countries in 1994-95 Mexican crisis caught many people, Latin America unable to pay their external debt. including economists, by surprise.' Mexico had This debt overhang increased country risk and made considerable reforms and its government reduced foreign investment. In many cases, addi- economic managers were highly trained and tional fiscal difficulties arose because the external highly regarded professional economists. Due to debt was socialized. Under strong pressures from fiscal strengthening and general economic reform international commercial banks, governments in Mexico, neither this country in particular, nor took over the private debt, bailing out the debt- the region in general, appeared vulnerable to ridden business sector. sudden and drastic changes in capital flows. By socializing the debt the governments Although structural, long-run, capital flows can created a fiscal problem, shifting the financial change through a combination of changes in 1 2 m BLACK DECEMBER national savings and in domestic expenditures, etary stock was reduced in nominal (and real) those functions are deemed stable. terms by almost 20 percent in a four-month HoNvever, the most recent Mexican crisis span.3 To put this figure into perspective, it is illustrates the crucial role in economic destabiliza- worth noting that during the world economic tion played by short-term policies that result in an depression of the 1930s, it took nearly four years excess supply of money. In Mexico excess money (from August 1929 to March 1933) for the U.S. caused reserve losses, additional current account money stock to decline by 35 percent. According deficit, exchange rate instability, and a reduced to Friedman (1963), this contraction of the mon- demand for money. In the context of a fixed ey stock was the main reason for the length and exchange-rate regime, the balance-of-payment severity of the worldwide depression. deficit resulting from the excess supply of money In Chile during the 1982-83 banking deba- could have been predicted by the monetary cle, the government took over more than 50 per- approach to the balance of payments. In the case cent of the nation's banking assets. In the 1982 of Mexico the prediction was correct. 2 Mexican crisis, the whole banking system was The Mexican crisis caught observers by sur- nationalized by the Lopez-Portillo government, prise in part because the 1990s had been a time and in early 1995, the newly privatized banking of sweeping structural reforms in Argentina, system was again at the brink of collapse. In Mexico, and Peru. Most Latin American coun- Venezuela's 1994 banking breakdown, the cost to trics, including Mexico, had adjusted their the nation of solving its banking crisis was esti- economies by privatizing important sectors, mated at about 14 percent of its gross domestic deregulating many others, improving their fiscal product (GDP); in addition, the crisis directly stanccs, and opening their borders to the bencfits affectcd 55 percent of the country's banking sys- of international trade. Brazil had also gone a long tem and more than 6 million people. way in reforming trade, while Chile continued It is important to remember that recent to bc stable politically and economically, and the banking crises have not been confined to Latin Brady external debt program had given some America. In the 1990s the Baltic countries, Esto- relief. Interest rate increases in the United States nia, Latvia, and Lithuania, have experienced during 1994 were moderate, and the source of severe crises.4 Developed countries that many capital inflows into Latin America, coming from Latin American countries saw as models of good the net savers in East Asia, remained stable. regulation and efficient supervision were them- selves hit by similar problems. During 1995-96, BANKING BLUES Japanese banks' non-performing loans were esti- mated at between U.S.$400 billion5 and The 1994-95 Mexican crisis was not the U.S.$800 billion. In 1995,Japan's largest crcdit first banking crisis in Latin America. Historically, cooperative, Kizu;Tokyo's largest bank, Cosmo; this region has had a large share of banking insta- and the nation's biggest regional bank, Hyogo; bility. In the last two decades, Argentina, Chile, collapsed as a result of their bad portfolios. The Mexico, andVenezuela have experienced the Japanese government responded by raising taxes most resounding crises, while other countries, and strengthening deposit insurance. Finland, such as Bolivia, Brazil, Peru, and Uruguay, have Sweden, and Norway have also recently experi- also suffered their own set of banking problems. enced large banking losses. In the U.S., the The size of these crises has been staggering. breakdown and subsequent government bailout In Argentina during 1982-83, the real value of of U.S. savings and loans cost taxpayers several deposits declined by 58 percent from the previ- hundred billion dollars. ous year's levels, and some leading banks, like The Mexican banking system was particu- Banco de Intercambio Regional and Banco de larly affected by the 1994-95 crisis. Aggregate Italia, among others, had to be liquidated. In a past due loans increased by 31 percent in a one- second crisis during early 1995,Argentina's mon- month period January to February 1995).6 INTRODU CT IO N 1 3 According to estimates, the Mexican bail-out heavily dollarized economies. In this case the will cost taxpayers about 9 percent of GDP.7 It is inflows will not cause current account deficits. If commonly believed that most banking crises the country is open and has a fixed exchange have mainly been caused by macroeconomic rate, part of its capital inflows will be generated imbalances, coupled with structural weaknesses by changes in its demand for money and supply in the financial system. In turn these banking of domestic credit; this is the "domestic" compo- crises have feedback to the economy. Also, capital nent of capital flows. Other capital inflows- inflows (and outflows) are given a prominent investment opportunities, for example-are role in explaining the recent crisis. "exogenous." Both domestic and exogenous forces produce capital flows that alter the under- CAPITAL INFLOWS lying equilibrium real rate of exchange. The ability of Argentina, Chile, and Mexico The structure of capital inflows in the '90s (until 1994) to keep inflation in check-and was much different from that of the previous even to reduce it-in the presence of large capi- decade; there was a much larger share of non- tal inflows suggests an increase in the demand for debt portfolio flows, longer-term debt, and direct nmoney in those economies. The best example is foreign investment. The World Bank has been Argentina, where the flows have boosted both instrumental in supporting these structural international reserves and the supply of money, changes. Some economists were indeed con- but inflation has dropped to international levels. cerned about the sustainability of large current- A highly dollarized economy, Argentina has used account deficits, but their worries were sub- part of its capital inflows to meet the growing sumed in the overall atmosphere of optimism. demand for international money. Long-term capital inflows adjust for the Under a floating exchange-rate regime, capi- difference between desired domestic savings and tal inflows do not necessarily produce inflation. desired investment. This adjustment has mone- Capital inflows change the real exchange rate and tary and exchange rate implications. If the consequently change relative prices, but they cause country has a floating exchange rate, the nomi- inflation only to the extent that the central bank, nal exchange rate adjusts in response to increas- acceding to political pressure to "protect" the es in capital inflows. In this system, the balance export sector, increases its holdings of international of payments is always balanced in the sense that reserves in order to reduce exchange rate apprecia- the current account is equal to the capital tion.When the central bank cannot sterilize the account, and there is no change in international increase in the money supply by reducing other reserves. sources of monetary expansion, the accumulation If the country has a fixed exchange rate, the of reserves produces inflation (Calvo, Leiderman, excess supply of dollars generated by the initial and Reinhart, 1992; Corbo and Hernandez, 1993). capital inflow goes into the coffers of the central bank, which issues domestic currency. Initially, CHANGING CAPITAL FLOWS AND there is a balance-of-payments surplus, measured THE REAL EXCHANGE RATE by the increase in international reserves. Later, the excess supply of domestic currency will work In most Latin American countries, the itself out through a current account deficit, and domestic currency has appreciated over the past the central bank's foreign exchange holdings will few years (Dooley, Fernandez-Arias, and Kletzer, return to their initial level.8 1994; Calvo, Leiderman, and Reinhart, 1992 and In a fixed exchange-rate system, the real 1993). Argentina and Mexico have experienced exchange rate will adjust through a rise in the the sharpest appreciation recent years. Chile's real domestic prices of nontradables. exchange rate is much more stable than those of Capital inflows can also be associated with the other countries, but it still has tended to rise. increased demand for money, particularly in In Brazil appreciation gradually subsided be- 4 m BLACK DECEMIBER tween 1992-93 but picked up again under the increase the benefit of importing high-technolo- Real Plan. gy goods. In this way, appreciation could benefit An increase in capital inflows causes appre- the export sector.This will be especially impor- ciation in the domestic currency. In other words, tant as broad market reforms and trade liberaliza- it increases the relative price of nontradable tion take effect, because these reforms may goods. Several studies have noted that capital increase the need for new investment. inflows have the same effect on exchange rates as The above general description of banking a mineral discovery or a permanent increase in crises and capital flows leads to the purpose of the terms of trade (this phenomenon is some- this paper-to discuss in general the main causes times called "Dutch-disease") (Corden and of banking distress and crises and, in particular, Neary, 1982; Corden, 1984; Corbo and Hernan- the Mexican crisis and its aftershocks in dez, 1993).This occurs regardless ofthe Argentina. exchange-rate regime (Calvo, Leiderman, and First, we discuss the anatomy of recent Reinhart, 1993; Corbo and Hernandez, 1993). episodes of financial distress and crisis. Most of With a fixed exchange-rate regime, the the banking sector crises are preceded by a bank- exchange-rate appreciation will occur through ing boom. The banking expansion goes hand-in- price increases and, if the country has a "clean" hand with financial liberalization, reduced rate of float, through appreciation of the nominal rate. inflation, increased capital inflows, increased The dollar price of tradable goods will not demand for money, and credit expansion. Then, change in either case, because for small coun- financial distress occurs, caused by a combination tries, it is given. of macroeconomic imbalances, changes in rela- Argentina, which has had a truly fixed tive prices (including exchange rate appreciation nominal exchange rate since 1991, has had no and high real interest rates), poor enforcement of trouble adjusting to increasing capital inflows, regulations, and a perverse asymmetry generated because under this type of exchange-rate regime, by the joint effect of fractional reserve require- the equilibrium exchange rate requires that the ments and deposit insurance. domestic price of nontradables increase. But if Second, we discuss the traditional determi- the rate of capital inflows slows, a more depreci- nants of capital flows, emphasizing profitability ated domestic currency would result. If the nom- (interest rates) and risk (interest-rate differen- inal exchange rate is fixed, deflation (not just a tials). We claim that to understand long-term reduction in inflation) will occur. This deflation trends in capital flows, it is very important to would result in substantial recession and unem- scrutinize savings and investment functions, but ployment, in particular if the labor market is to understand short-term swings, it is crucial to rigid. look at the balance of payments as a monetary Exporters dislike appreciation of the domes- phenomenon (Johnson, 1958). tic currency rate because it affects their ability to Third. we provide background material sell their goods in international markets, and they describing the size, composition, and use of will lobby against competition from abroad. If recent capital flows to Argentina, Chile, Brazil, they are successful, the government could reverse and Mexico.We compare the size and structure trade liberalization. But if the government main- of the current capital inflows to those of the late tains liberalized trade, these producers may post- 1970s, assessing different financial indicators. In ponc invcstment in the export sector becausc of addition, we show that these financial indicators their diminished international competitiveness. pointed to a substantial improvement in each Appreciation of the real exchange rate, country's creditworthiness before the 1994-95 however, could reduce the cost of investment crisis. goods for local business. In particular, to the Fourth, we discuss whether current account extent that an economy cannot produce capital deficits are a curse or a blessing. Current account goods and must import them, appreciation will deficits allow a country to profit from investment IN TRO DUCT IO N * D opportunities, but they also allow a country to crisis were mainly endogenous, meaning that spend on consumption beyond its means. Cur- even though there were some exogenous shocks rent account deficits have been blamed for the to Mexico, including political violence, domestic Mexican and other banking crises and are now policy mainly caused the country's economic in low regard. All the same, if there are surplus breakdown. countries, there have to be deficit countries. Not Sixth, we discuss the aftershocks of the crisis all deficits are bad and some of them, in fact, and its impact on Argentina, which was greatly might be quite good for long-term growth. affected by the Mexican collapse.The Mexican Fifth, we attempt to explain the Mexican fiasco triggered a full-fledged run on Argentina's crisis.We stylize three different hypotheses banking system, with deposits reduced by 18 explaining the causes of the crisis as completely percent and the money supply by 20 percent exogenous, completely endogenous, and hybrid. between January and May 1995. During that We conclude that "Mexico's crisis can be same year, real income in Argentina fell by 4.5 summed up as the classic case of a pre-deter- percent.We contrast the policy response of mined exchange rate that becomes unsustainable Argentina, which followed the rule of a currency due to the expansion of domestic credit and the board, with that of Mexico, which followed a reduction in money demand." The causes of the discretionary policy. ANATOMY OF BANKING DISTRESS AND CRISES THE DESCRIPTION AND ANATOMY of recent world experiences with financial crises have been amply illustrated in the literature, by, among many others, Giorgio (1996), Gorton (1986), Karninsky and Reinhart (1995), Meltzer (1995), and Rojas-Suarez and Weisbrod (1966).There is rough agreement about the stylized facts describing the path of eco- nonuc variables before and during these crises, but there is less agreement about the weight given to those variables in determining causalities and the length and depth of these crises. FINANCIAL BOOM: INCREASED During these phases, an increased demand DEMAND FOR MONEY AND for money did not produce a recession, because CAPITAL INFLOWS it was fed from the open capital account or, to a lesser extent, from the central bank.This was par- Before the crises we have generally ticularly the case in exchange-rate-anchored sta- observed a financial boom due to deregulation of bilization plans in which money supply was the financial sector, opening of the capital demand-determined or endogenous, as in Chile account on the balance of payments, and macro- (1979-81), Argentina (1991), and Mexico economic stabilization. (1990-94). During these episodes, the capital These financial booms coincided with high account contributed greater capital inflows than income growth and an improved business envi- desired by the excess demand for money (those ronment, coupled with high real interest rates.The inflows had an important exogenous compo- latter did not produce an immediate impact on nent), resulting in current account deficits and a banks' portfolios, because at their onset, they coin- boom in expenditures.9 cided with the expansionary phase of the business Alternative plans, monetary based stabiliza- cycle. Business cycle expansion was propelled by tion (MBS), anchored by the monetary base, promising expectations of stability, deregulation, often produced early recessions as a result of very and increased capital inflows, among other factors. high interest rates. Those rates were created by 7 8 BLACK DECEAM BER central bank failure to feed enough money to ture bust, resulting in a deep recession and high satiate the initial increase in demand for money unemployment. (Mundell, 1971). This occurred in Latin America except for Peru, which in 1991 launched and HIGH INTEREST RATES AFFECT ambitious and successful stabilization plan BANK PORTFOLIOS anchored in the monetary base, coupled to a freely floating exchange rate, very tight fiscal pol- During the downturn, the deleterious icy, and vast structural reform. effect of high interest rates could no longer It is noteworthy that Peru did not experi- remain hidden by the expansionary phase of ence a recession, but in fact experienced the the business cycle. High interest rates and kind of boom usually associated with exchange- changes in relative prices that occurred during rate-based stabilization plans (ERBS). This anom- the early phase of the stabilization plan affected aly might well be explained by the large share of the market value of the banks' portfolios. The this country's money supply that was endoge- capital basis of banks was eroded and became nous due to the large dollarizationi of the niegative for some of them, helping to trigger economny. the crises. High interest rates and increasingly appreci- FINANCIAL BUST: REDUCED ated domestic currencies have been found to DEMAND FOR MONEY AND precede many banking crises (Giorgio, 1996). CAPITAL OUTFLOWS Interest rates have been high for many reasons, including increased demand for money and cred- In many ERBS plans, the expansionary it to finance a expenditure boom; increased risk process comes to an abrupt end as a result of in credit operations; increased risk of devalua- domestic policies that are inconsistent with the tion, increased country risk, and sterilization of fixed exchange rate.The most common factor capital flows by the central bank.12 shaking confidence in specific countries has been In Chile's crisis during 1981-89, its a time-inconsistent fiscal-cum-exchange rate and exchange-rate-based stabilization period coincid- monetary policy, as seen in Argentina o during ed with an appreciating dollar and a fall in cop- 1979-81 and 1985-86 and in Mexico during per prices (de la Cuadra andValdez, 1992).The 1981-82 and 1994. interest rate had a floor determined by the Libor In Argentina during 1979-81, increased fis- rate, which had been very high.When the price cal expenditures increased the stock of debt to of copper plummeted, putting downward prcs- unsustainable levels. By the end of 1989, sure on domestic prices, real interest rates sky- Argentina's domestic debt had again reached a rocketed affecting banks' portfolio. ceiling, and the government again confiscated a large share of its citizens' financial wealth.The CHANGES IN RELATIVE PRICES other clear example of this phenomenon occurred in Mexico in 1994, when official Many recent crises in Latin American coun- development banks increased domestic credit to tries have been instigated by sharp changes in the private sector in a way that was inconsistent relative prices related to stabilization plans. The with the fixed exchange rate, finally causing the irony is that even good economic policies have crisis. been the source of banking problems due to In both the Argentine and Mexican crises their effect on relative prices. The banks' pre- 1994-95, there was a reversal of the exogenous stabilization portfolios may have been profitable comnponent of capital inflows, which aggravated with a pre-stabilization set of relative prices, but the crises, although it did not cause them. Ulti- through relative price changes, stabilization sub- mately, the sharp reduction in capital inflows stantially reduced the real value of banks' asset reversed the expenditure boom to an expendi- portfolios. ANATOMY O F BANKING DI STR E SS AND CR I SES * 9 Furthermore, many stabilization plans, wvith THE PERVERSE ASYMMETRY their emphasis on deregulation and privatization, have often modified the government's role as a In all of the banking crises, which affected major contractor (mostly of infrastructure), ren- very different countrics and diverse banking sys- dering old government-related business unprof- tems, the common factor has been the important itable. Some firms adjusted and others perished, role of moral hazard and what we call the per- while bank portfolios suffered.This source of verse asymmetry problem. Perverse asymmetry banking problems, for example, was particularly refers to the tendency of the public, because of important in Bolivia during its 1994-95 banking fractional reserve requirements and deposit insur- distress. ance, to disassociate the quality of a banking sys- tem's assets from its liabilities. BREWING THE CRISIS: In the early phases of stabilization, high FRACTIONAL RESERVE interest rates attract more depositors who, due to REQUIREMENTS AND DEPOSIT the moral hazard introduced by deposit insur- INSURANCE ance, believe that their funds are insulated from the market value of bank assets. Later, in the Banking crises in Latin America cannot be contractionary phase of the cycle, deposits con- understood without comprchcnding the trou- tinue to increase independent of the banks' eco- blesome nature of fractional reserve require- nomic situation. ments. Fractional reserve requirements imply an Consequently, the constant dollar value of inverted pyramid, with a small reserve base sup- the liability side of the banking system continues porting a large quantity of deposits and credit. to increase. By contrast, during the subsequent In the early phase of the stabilization process, downturn, bank portfolios deteriorate due to when there is expansion of that base, there is high real rates of interest or changes in relative euphoria. prices (mainly in the exchange rate). This in turn Fractional reserve requirements mean that affects the market value of the bank's asset port- small changes in the reserve base expand deposits folio. Distress borrowing by businessmen contin- and credit by many times. Conversely, a small ues, along with unconcerned funding by the reduction in the monetary base reduces credit public and other investors. Depositors enjoy and money supply many times.'3 increasing rates of return With little risk because Deposit insurance14 has been widely used to of the deposit insurance warranty. avoid the domino effect caused by both fraction- In other words, the banking system enters al reserve requirements and an increment in the into an explosive Ponzi scheme. Banks find them- cash-to-deposits ratio (a run on the banks). In selves locked in with significant assets that are not the short run, deposit insurance became a crucial paying interest due or principal. Interest due is instrument in halting ongoing crises. In the long capitalized in a process also called evergreening. run, as explained below, it combined w-ith some Consequently, to repay interest (and some princi- macroeconoiuc fundamentals to set the stage for pal) to its old depositors, banks have to rely on a full-fledged crisis. new depositors. In this game, interest rates con- tinue to go up, weakening banks' asset portfolios. DETERMINANTS OF CAPITAL FLOWS THERE ARE THREE EX-POST ACCOUNTING identities that shed light on capital flows by focusing on their ex-ante determinants: the investment-savings gap, the expenditure-income gap, and the money supply-money demand gap. The investment-savings gap emphasizes the investment-savings functions and is an useful approach for long-term pre- diction of capital flows:We know that if a country has a low savings rate and high investment opportunities, it probably has a relatively high interest rate and would finance part of its investment with a deficit in its current account. The expenditure-income or absorption approach emphasizes the determinant of production and expenditures, and the monetary approach emphasizes the excess ex-ante flow demand (or supply) for money. The monetary approach provides the most useful model for analyzing the kind of shock experienced by Mexican international reserves since October 1994, which ultimately caused the nation's banking and exchange-rate crisis. INCREMENT IN INTERNATIONAL ferentials (Dooley, Fernandez-Arias, and Kletzer, INTEREST RATES: UNLIKELY 1994; Fernandez-Arias, 1994; Calvo, Leiderman, CAUSE FOR THE MEXICAN CRISIS and Reinhart, 1992 and 1993).These studies claimed that in the 1990s, capital flowed to Latin Several studies have found that the main America because interest rates dropped in the determinant of capital flows are interest-rate dif- United States and other industrial countries, 11 12 * BLACK DECEMB ER while returns remained high in Latin American An important structural feature underlies countries. Thus, they hypothesize that if interest Latin America's capital flows: Its low savings rates in industrial counties rise again, capital will ratio, coupled with its higher investment ratio. flow out of Latin America. These studies stated Relatively large gross savings from Japan and the that there was room for a significant rise in United States have been an important determi- interest rates that would sharply reverse capital nant of capital flowvs to Latin America. Had the flows and cause great hardship in the region. region had a savings ratio similar to Japan's, it Chart 1 shows the annual Libor percentage would not have received net capital inflows. rate. During 1994 there were indeed increments In the short run, however, the Mexican cri- in the Libor rate, but notice the difference sis has highlighted dramatically the importance between this situation and that of the late '70s. of the domestic policy underlying the volatility The problem did not originate in higher inter- of capital flows. In the Mexican case, the prob- national rates, but in low domestic rates due to lem was the inconsistency of a fixed exchange both a policy of sterilizing some initial capital rate with expansive, non-passive monetary and outflows and expanding domestic credit in the credit policy. context of the reduced demand for money. Argentina between 1979 and 1981 provides Long-term real interest-rate differentials another historical example of the importance of responld to and are formulated by long-term domzestic policy. Its cornbiniation of tight dorries- structural forces that are determined by the mar- tic credit, loose fiscal expenditures, and fixed ginal propensities to save and to invest. Econo- (pre-determined) exchange rates led to very high mists generally accept, for example, that the main interest rates, luring large amounts of hot capital. determinant of the difference in the ex-ante A small, higher-than-scheduled devaluation in interest-rate differentials and current account February 1981 acted as a warning signal that flows between Japan and the United States has triggered the outward stampede of capital, thus been the difference in these countries' savings causing the collapse of the nation's economic rates. In the short term, howvever, sharp changes strategy. That poorly managed devaluation resem- in interest-rate differentials can also arise due to bles the one that triggered the recent Mexican mistakes in monetary policy. debacle. A final example is Chile during 1982-83, Chart 1 when a sharp deceleration in capital inflows to Libor: Annual Rates that country caused an extremely severe crisis. This deceleration was not due to an increment 2b in international interest rates; in fact, interest 20 rates in the United States collapsed after 1981. Within the context of a pegged cxchange ratc, 1D- 1\1 1v\ Chile had increased domestic credit to the pri- g0 \vate sector by 50 percent in 1980 and again by no X < /t 50 percent in 1981.15 As in Argentina during 5 1981 and in Mexico during 1994, the Chilean crisis was mainly due to endogenous forces, o ..... . . . ..... namely, very large increases in domestic credit, 78 80 82 84 86 88 90 . ~ 92 9496 coupled with a pegged exchange rate and signifi- SoLr.e: riernatiorK = nrKoaIa Sra:istics cant appreciation of the domestic currency. E AND COMPOSITION OF CAPITAL FLOWS RECENT FOREIGN INVESTMENT in Latin America has had a contractual nature very different from that of the late 1970s. A large share of recent capital inflows has taken the form of nondebt portfolio investment and direct investment. Gradual increment in international interest rates would probably have caused a deceleration of capital inflows to Latin America, but this phenomenon did not cause the Mexican stampede. It is relevant to note that, from the middle GNP fell from 21 percent in 1982 to only 3 '80s through the early '90s,Argentina, Chile, and percent in 1994. Mexico faced an improved financial situation The following tables present some tradi- with regard to short-term debt (Table 1). tional indicators of the relative burden of exter- Between 1982 and 1994, debt-to-GNP and nal debt: debt-to-export ratios declined in all these coun- Ratios of total debt and debt service to tries. Argentina experienced the greatest GNP also improved for all four countries (Table improvement; its ratio of short-term debt-to- 2). Argentina's debt to GNP ratio fell by almost Table 1 Ratio of Short-Term Debt Outstanding to GNP and to Export for Selected Latin American Countries, 1982 and 1994 Ratio to GNP Ratlo to Export 1982 1994 1982 1994 Argent na 0.21 0.03 1.70 3.34 Braz 0.07 0.06 0.75 0.63 Chi e 0. 5 0.10 0.65 0.33 Mexico 0.16 0.09 0.95 0.56 Sour-e:e World Deb- Tab es 1996. 13 14 * BLACK DE CEMIBER Table 2 Ratio of Total Debt and Debt Service paid and (due) to GNP for Selected Latin American Countries 1982 and 1994 Debt Debt Servicc Paid anc (Due) 1982 1994 1982 1994 Argentina 0.55 0.28 0.06 0.02 (0.02) Brazdl 0.35 0.28 0.07 0.03 (0.04) Chi e 0.78 0.46 0.6 0.06 (0.05) Mexico 0.53 0.35 0.10 0.05 (0.05) Source: Wor lc Debt Tab es, 1996 Table 3 Selected Debt and Financial Indicators for Selected Latin America Countries 1981 and 1994 Argentina Brazil Chile Mexico 1981 1994 1981 1994 ,981 1994 1981 1954 InterestjExp 29.1 19.6 38.V 12.9 3 .6 8.0 35.2 14.2 Interest/GNP 4.5 1.5 4.1 1.2 5.7 2.4 '-.1 2.2 Reserve/GNP 6.5 5.8 2.9 7.1 ^2.5 27.4 21 1.8 Resefve/lMP 3.0 51.0 2.3 70.7 4.5 87.1 1.4 7.5 Short-term 36.2 6. 19.0 21.0 19.1 21.9 32.0 24.6 Debt/Tota Debt Average 11.9 7.8 15.C 8.0 5.0 7.5 15.0 5.7 nterest A\erage 13.8 9.3 10.0 8.5 10.6 13.4 8.2 7.5 Matur ty half between 1982 and 1994, and its debt service ing and economic crisis with massive ratio fell by two-thirds. Chile and Mexico expe- devaluation. rcienced similar improvements in these ratios. In the '90s long-term capital flows in Latin Brazil had modest improvement in its debt ratio America increased relative to short-term flows. but, like the other countries, it also experienced Nevertheless, one lesson from the Mexican crisis significant improvement in its debt service ratio. is that capital flows usually classified as long The preceding tables show that, until 1994, term can be just as hot as short-term flows. As most financial indicators made it unlikely to long as there are secondary markets providing foresee the scope and depth of the recent Mexi- liquidity, the classification of long- and short- can crisis.The crisis showed, however, that Mexi- term is not very useful. The most important co's strengthened financial status was no guaran- classification, with regard to the stability of tee against collapse. The lesson is that it takes a flows, is the ratio of foreign direct investment to long time to build up econormic and financial total flows. strength, but it takes verv little time to throw it Just as in the late 1 970s, Latin American overboard, which is as true for individuals as it is countries in the 1990s had been experiencing for countries. surpluses in the capital account of the balance of In 1994, therefore, three events brought to payments.'6 But the size and the coniposition of life a scenario that most people had previously the 1990s flows were different from the earlier believed to be very unlikely: a sharp increment ones. The most important change has been in the in Mexican domestic credit, an important reduc- contractual nature of the flows; by the 1990s tion in money demand, and a drying up of inter- debt flows had declined relative to portfolio and national reserves. The consequence was a bank- foreign direct investment. S IZ E AND C O M P OS IT IO N OF CAP I TA L FLOWS * 15 During 1995 the rate of capital inflows to the again became a net recipient of capital flows. In region was substantially reduced. Although the 1992 and 1993, inflows were above 4 percent of inflows did not change to outright capital outflows, GNP, roughly half the level of inflows received they grew at a much slower pace. It is noteworthy in 1979. that much of the reserve loss experienced by the In Chile, inflows peaked in 1981, when they Mexican Central Bank did not reflect reserve losses reached a formidable 15 percent of GNP. But in of the Mexican citizens. During the recent crisis, 1982, the flows dropped to only 5 percent. of millions of small national investors switched their GDP, triggering a serious crisis. In 1992 and portfolios, substituting dollars for pesos. Rather, the 1993, flows to this country were above 6 percent. loss to Mexican citizens will come at the increment Finally, for Brazil the new inflows have of their future tax liabilities. reversed the downward trend of the 1980s, but Table 4 Debt and Nondebt flows to Selected Latin American Countries, 1975-93 (average ratios to GNP) Debt Flows Nondeot Flovls Country 1975-81 1982-89 1990-93 1975-81 1982-89 1990-93 Argentina 0.037 0.013 0.014 0.004 0.006 0.022 Brazi 0.035 0.012 0.006 0.010 0007 0.070 Chie 0.053 0.066 0.027 0.008 0.010 0.026 Mexicc 0.044 0.017 0.032 0.009 0.011 0.034 SIZE OF THE FLOWS they remain moderate relative to GNP compared wvith those of the other countries in the sample We express capital flows as a ratio to GNP and with Brazil's earlier experience. In Brazil, to highlight their relative importance.While inflows peaked in 1974 at 6 percent of GNP and Brazil, Chile, and Mexico have received larger stayed above 3 percent until 1985. The flows capital inflows in the 1990-94 episode than in then fluctuated around zero until 1991, before the 1975-81 episode, these flows, when finally chmbing to about 2 percent of GNP in expressed as a ratio to GNP, are only about half 1992 and 1993. those received by Chile in the first episode (see Table 4). COMPOSITION OF THE FLOWS In the first episode, the inflows to Mexico peaked in 1981 at 7 percent of GNP. Between The composition of the flows to Latin 1983-85, the country accumulated capital out- America has changed drastically since the late flows equivalent to 4.2 percent of GNP. Since 1970s and early 1980s when medium- and long- then, Mexico recovered spectacularly; its capital term commercial bank loans predominated. The flows became positive again in 1989, and its cap- emphasis has shifted from debt flows to nondebt ital inflows exceeded 8 percent of GNP in 1991, flows, and in Argentina and Brazil, nondebt capi- 1992, and 1993. During 1995 capital flows were tal now accounts for a larger share of capital dramatically reduced. inflows than does debt (see Table 4). Much of For Argentina and Chile, the inflows as a the capital has been allocated to portfolio invest- ratio to GNP recently reached their highest lev- ment and direct investment-foreign investors els since the 1975-81 episode, but remained well have become partners. And most important, a below the peaks in that episode. In Argentina, large share of the inflows has been directed to the flows peaked in 1979, reaching 7 percent of the private sector rather than to the government. GNP. They then fluctuated annually between 2 Nondebt flows in Argentina reached about and 3 percent of GNP, finally becoming outflows 1 percent of GNP in 1980 and 1981, and then in 1989 and 1990. In 1991, however, Argentina fluctuated throughout much of the decade. But 16 * BLACK DECEMBER in 1988 they began a fairly steady upward climb, During the two episodes of large capital inflows reaching 1.5 percent of GNP in 1989 and con- in the past twenty years,Argentina, Brazil, Chile, tinuing to increase. In Chile nondebt flows rose and Mexico already were running deficits in to about 1.8 percent of GNP in 1982. Although their current accounts; the capital inflowvs these flows then dropped below 1 percent in allowed these economies to finance larger some years during the debt crisis, they have since deficits, that is, to use more resources fromn recovered, rising above 2 percent of GNP in the abroad. 1990s. The most impressive growth in nondebt When explaining the current account as a flows occurred in Mexico, where such flows gap between expenditures and income, it is increased from about 1 percent of GNP in important to know if that gap is due to a fall in 1980-82 to more than 3 percent of GNP in permanent or transitory income. In the case of a 1992-93.The exception to this trend of expand- transitory gap, foreign resources could be used to ing nondebt flows has been Brazil, which has smooth consumption over time following a tem- exhibited a downward trend in these flows as a porary adverse shock to production. For exam- share of GNP ple, faced with a natural disaster, a country might Equity financing has been aided by the cre- lower its savings to smooth consumption and use ation of depository receipts which permit trad- resources fronm abroad to maintain investment at ing in securities not listed on local stock an optimal level, generating a deficit in the cur- exchanges. These receipts, which take two rent account. forms-American depository receipts (ADRs) For Latin American economies, the possibil- and global depository receipts (GDRs)-repre- ities for smoothing consumption are limited sent claims on, for example, Latin American because their creditworthiness deteriorates when securities and can be traded in the United States they are hit by a large shock (Mathieson and and Europe.This mechanism has expanded the Rojas-Suarez, 1992; Gertler and Rose, 1991). investor base for developing country securities. Moreover, historically, most countries have bor- In the United States institutional investors are roxved more when their economies have been permitted to hold ADRs because they are con- strong than when they have suffered a shock sidered U.S. securities (Chuhan, Claessens, and (Calvo, Leiderman, and Reinhart, 1992). Maningi 1993; El-Erian 1992; World Bank The decision by governments to run cur- 1994). rent account deficits to smooth the effects of Although it remains small, foreign direct negative shocks deemed temporary could be bad investment has increased in Latin America. It has if those shocks turn out to be more permanent. grown in part because these countries have been In a classic example of a government failure (as friendlier in recent years toward foreign investors opposed to a market failure), Brazil, in response and because foreign investors have confidence in to the oil shocks of the 1970s, followed a two- the steps that all these countries have taken since pronged but inconsistent strategy. First, it initiat- the first debt crisis in 1982. As part of those poli- ed a costly substitution of alcohol-based fuel for cies, Brady-type debt restructuring deals have oil, a measure consistent with a perception that helped to boost country creditworthiness in the oil shock was permanent. Second, it bor- Latin America. rowed vast foreign resources, running large cur- Capital inflows allow the recipient country rent account deficits. to increase expenditures in domestic and foreign This strategy of borrowing was consistent markets for goods, services, and assets. In terms with a perception that the oil shock was tempo- of the balance of payments, this means that capi- rary. Eventually oil prices fell by roughly half, but tal inflows enable the economy to run a deficit they never returned to their original levels. The in its current account (a surplus in its capital government's strategy of using alcohol substitu- account), spending more than it currently earns. tion as a long-run solution and huge debt as a SI Z E AND CO MP OS IT IO N OF CAP I TA L FLOWS * 17 short-run palliative proved ill-fated. It financed ing 1990-94 investment increased, but savings rapid growth during the late 1970s and early continued to decline. Thus, Mexico used foreign 1980s, but this artificial growth later collapsed. real resources to dampen the effect of reduced Argentina, Brazil, Chile, and Mexico have savings on investment. This situation deteriorated used their economic capacities to draw on for- when Mexico used its own stock of reserves to eign resources for different objectives. These finance domestic expenditures. objectives are revealed in the trends in invest- Between 1976-81 Chile also substituted ment and gross national savings for each country; foreign for domestic savings at an increasing rate, the gap between these trends is reflected in each to sustain investment and to increase consump- country's current account balance. In the tion. This episode ended in severe crisis, due not 1990-94 episode of capital flows, none of these only to the savings-investment gap, but also to countries used its capacity to borrow foreign other factors like its credit policy. resources to smooth the effects of a shock. Argentina has shown an important decline Rather, their borrowing stemmed from improved in its savings ratio between 1978 and 93. Its sav- investment opportunities, coupled with the low ings have begun to pick up during the past three savings ratio and relatively high interest rates. years, but it is too early to detect a change in the From the perspective of the investment-sav- trend. Argentina's savings-investment gap is, nev- ings gap, until 1982 Mexico borrowed abroad to ertheless, far smaller than Chile's was in the per- finance increased investment. This foreign- od around 1980 and also smaller than Mexico's financed rate of investment proved unsustainable, in the recent crisis. however, and the rate declined until 1990. Dur- CURRENT ACCOUNT DEFICITS: NEITHER CURSE NOR BLESSING CURRENT ACCOUNT DEFICITS are many times discussed either as a curse or a blessing: In fact, they are neither. Both curses and blessings are God-given, exogenous, and this is not the case for current account balances. When a country runs a current account deficit, its foreign sector is usually described as weak, and, when the current account deficit increases, as deteriorating. The implication is that current account deficits are bad and current account surpluses good, which is a mercantilist anachronism. An increase in the current account deficit before the 1982 crisis), 1982-89 (the post-crisis can be good or bad depending on the source of years), and 1990-93 (the recovery years before change. For example, a U.S.$1 billion increase in the Mexican crash).The same information is the current account deficit could be due to presented in both tables, but arranged in differ- investment falling by U.S.$200 million and sav- ent ways to make comparisons easier. ings falling by U.S.$1.2 billion, or it could be Notice that Mexico is the only country in due to investment increasing by U.S.$1.2 billion, 1990-93 showing both a negative change in and savings by U.S.$200 million.The source of national savings and a positive change in invest- financing of the current account is also impor- ment.This resulted in a large negative change in tant. As is clearly shown by the recent Mexican its current account. experience, a current account deficit (a flow) Argentina's current account appeared to financed with a given stock of international have deteriorated in both 1975-81 and 1990-93 reserves is unsustainable. and improved in 1982-89. But when we look at Tables 5 and 6 show the sources of change the source of the changes, we realize that this in the current accounts of the four countries for interpretation is wrong. In the high-debt period three periods, 1975-81 (the high-debt years of 1975-81, the savings ratio decreased an aver- 19 20 * B L A C K D E C E-M B E R Table 5 Average of Changes in the Current Account Balance, National Savings, and Domestic Investment in Selected Latin American Countries, by period, 1975-93 (as a percentage of GNP) 1975-8} 1982-89 199C-93 Country rCA dSV di/'V dC.4 dSV dlNV oCA dSv d/NV Argentina -n 89 -0.67 C 2 -0.54 -0.34 -0.88 -0.30 0 70 1.00 Br-az I 0.37 -0.,1 -0 79 -0.6C0 -r019 -O 79 C 47 -O 27 -0.73 Ch e - .77 -1.41 0.36 1.68 2.86 1.28 -0.52 -0.58 -0.05 Mexico -0.41 0.51 0.93 0.59 - .14 -.73 -<.25 -0.33 0.93 Note: cCA s chance n cu,renr acccint balance dSV s change n nat onal sav ngs and cO INV is change ir domestic imvestment. Soorce; WVorld Bark age 0.67 percent of GNP per year while the THE SUSTAINABILITY OF investment ratio increased 0.21 percent of GNP CURRENT ACCOUNT DEFICITS per year. Therefore, the decline in the savings ratio was the important factor accounting for the There is a lack of general agreement on the negative change in the current account. Between significance of current account deficits. Some 1982-89 the current account apparently authors, for example Edwards (1995) maintain improved, but we see that the improvement was that the main cause behind the Mexican peso due to a fall in investment of almost 1 percent of crisis was an unsustainable current account deficit GNP per year coupled with a fall in savings of that, starting in 1992, was financed by very large about a third of a percent of GNP per year. By capital inflows. In this vein, capital flows are the contrast, in 1990-93 the increase in the current cause of the problem, not the effect. account deficit was good because investment Although it is seldom expressed, there is a increased and savings also increased, although very important distinction to be made regarding more slowly. the source of current account deficits that bears Not all of the investment-caused current on their sustainability. Current account deficits accounts deficits are good. For example, an can be fed by two sources: capital inflows and a increment in the current account deficit that reduction in the stock of the country's interna- increases investment could be unsustainable tional reserves. A current account deficit or flow if it is not based on fundamentals (for example, a that is financed by a given stock of international higher domestic interest rate due to higher reserves is unsustainable because the stock is expected rates of returns to investment), being depleted. (When the reasons for the deple- but is based in increased domestic tion are deemed temporary, the International credit. Monetary Fund and/or the World Bank come to Table 6 Average Change in Current Account Balance, National Savings, and Domestic Investment in Selected Latin American Countries, by Course of Change, 1975-93 (as a percentage of GNP) Change ,n current account Sea/ance Chance in nationa! sa/ings Change in domestic 'rvestment (dCA) (dS/) (/N V) Country 1975-1 1982-89 1990-93 19758 I 1982-89 1990 93 1975-81 1982-89 1990 93 Argent na -0.89 0.54 -0.30 -0 67 -0.34 070 C.21 -0.88 1.00 Brazil C 37 0.60 0.47 -0.41 -0.19 -0 27 -0.79 -0 79 -0.73 Ch e -'.77 1.58 -053 -b1 2.86 -0.58 C.36 1.28 -0.05 Mex co -0.41 0.59 ->.25 0.51 -1.10 -0.33 0.93 -1.73 0 93 Source. World 3ank C U RRE N T A cc o UN T DE F IC ITS: N E IT HE R C U RS E No R B LESS ING m 21 the rescue.) In the case of Mexico, an increasing- deficit to unsustainable levels and that a devalua- ly large share of its current account deficit tion is imminent, as in the recent case of Mexi- became financed with its reserves and, hence, co. Capital inflows are a function of profit, risk, became unsustainable. and the underlying savings-investment relation- A current account deficit can be permanent ships. If these relationships are stable, there is no (the U.S. has had a deficit for many years and so risk of stampedes. In the case of Mexico, the risk have Malaysia and Thailand), unless there is a increased both because of the large increment in perception by foreign investors that increment in domestic credit and the real exchange-rate domestic credit or money supply is pushing the appreciation. THE MEXICAN CRISIS "NO ONE PREDICTED the sheer size of the debacle in Mexican financial markets.A peso at 4.2 to 4.5 was conceivable; but the peso below 7-even if only for a day or so-was beyond anyone's expectations. And although most people foresaw that the prices of financial assets wvere heading for a fall or a 'correction' no one beheved that they would plummet.They did." (Sir Alan Walters, AIG World Markets, April 1995) THE EXOGENEITY HYPOTHESIS geneity hypothesis (Wall StreetJournal,January 31, 1995).According to Mancera, in 1994 the Under this hypothesis, the main causes of Central Bank followed a monetary policy that the crisis were political events, especially violent changed domestic credit each time there was a ones, and investors' herd instincts, coupled with change in the international reserves, which rising U.S. interest rates.The political events altered with each event of political violence. began with the Zapatista revolt ofJanuary 1994 Accordingly, all changes in the balance of pay- and exploded with the Zapatista resurgence at ments were due to exogenous political events, the end of December 1994. During this period and the Central Bank only reacted to these the government and the Zapatistas held peace changes. talks; nevertheless, there were other political The part of this hypothesis outlining the upheavals, including the assassinations of a presi- herd instincts of investors was articulated by dential candidate and other important public fig- Mexican Foreign Minister Jose Angel Gurria. In ures. This unrest had an important impact, touch- his words, the market could not be taken seri- ing off the speculative attack against the peso and ously because the "market" was nothing more the ensuing crisis. than fifteen guys in tennis shoes in their 20s The governor of the Central Bank of Mex- (Wall Street Journal, January 20, 1995). Capital ico, Miguel Mancera, has articulated this exo- flows were very large and, by nature, speculative. 23 24 * B LACK DEC EM BE R Suddenly the interest rate differential was not During 1994 the Central Bank increased enough to counter the perceived increased risk domestic credit every time an episode of political from the worrisome Zapatista uprising. One violence reduced international reserves, and this large institutional investor withdrew; the herd voided the automatic, passive mechanism that instinct prevailed, and others followed. should be implicit in a sustainable pegged exchange-rate regime. The Central Bank pre- cluded any automatic adjustment by trying to THE ENDOGENEITY HYPOTHESIS sterilize reduction in reserves induced by increas- es in the income velocity of money. The Central The endogeneity hypothesis is as follows: Bank focused on holding interest rates down in By the second half of 1994, domestic credit, order not to affect commercial banks' portfolios especially credit granted by official development and economic activity. Economic theory proved banks, increased dramatically. An excess supply of right, and the Central Bank could not control money caused a loss of reserves and a higher both the rate of interest and the nominal deficit in the current account. exchange rate. The endogenous explanation's paradigm is The next chart shows the log of domestic the monetary approach to the balance of pay- credit. Domestic credit increased by 27 percent ments. Changes in international reserves are from 1993-III through 1994-III, while the explained by changes in the demand for money exchange rate increased by 9 percent during this and changes in domestic credit. In Mexico polit- same period. During the last two quarters of ical violence did affect the demand for money, 1994, the level of domestic credit was growing but more important, official development banks at an increasing rate, with a 31 percent annual increased domestic credit to finance the private rate increase for the period between August 1 sector. Ultimately, changes in domestic credit and November 30. This huge increment in caused changes in international reserves. domestic credit was taking place in a context of both a pegged exchange rate (the nominal rate WHICH IS THE BEST HYPOTHESIS? was on the upper bound of the band) and a reduced demand for money. This huge disequi- Clearly, the exogenous explanation is not librium could only be sustained for a short time, convincing. By September 1994 the monetary as long as international reserves did not reach a base was 23 percent higher than in September critical point. By December 1994, it had 1993, and the credit of development banks had exploded. risen 32 percent. At the same time, the income velocitv of circulation of money was increasing. Chart 2 ' ~~~~~~~~~~~Mexico: Domestic Credit From February through August 1994, the ratio of international reserves-to-base money 6.5 declined steadily, but the reserves were still larger 6.4 than the base. After September 1994, the same ratio nose-dived, while the ratio of short-term 6. debt to total debt climbed. 6.2 The significant increment in domestic cred- it caused the reserve losses that ultimately 61 doomed the stabilization effort, but even assum- 6 C ing that political violence and herd instincts wvere r9 also important factors, the fact is that in each 931 93:3 94:1 94:3 95:1 95:3 96:1 episode of political violence and flight away from domestic money, the Central Bank did not allow the monetary base to contract. Source: IMF's l=S, sc2era nu-nbers ire 32. Data is ir ocs THE MIEXICAN CRISES * 25 To sum up, Mexico's crisis can be explained only a symptom of these underlying problems. In as the classic case of a pegged exchange rate that this endogenous scenario, the crisis was triggered becomes unsustainable due to an expansion of by the market's sudden perception that a devalu- domestic credit and a reduction in money ation was imminent. demand. The large current account deficit was AFTERSHOCKS OF THE MEXICAN CRISIS: ITS IMPACT ON ARGENTINA THE MEXICAN CRISIS produced an economic shock in Mexico and aftershocks in Argentina, Brazil, and other major Latin American countries. During 1995 real income in Argentina fell 4.5 percent. Between December 20, 1994, and March 1995, both the Argentine and Brazilian stock exchange indexes lost approximately 40 percent, and Argentina faced a banking crisis. However, different countries showed different levels of resilience to the eco- nomic aftershocks. The 1995 aftershocks of the Mexican crisis If the exogenous explanation of the crisis included: takes hold, we would expect pressures for capital- flow controls.We would also expect pressures for 1. Reduced rate of capital inflows higher trade barriers and subsidies and a general 2. Higlher domlestic interest rates backslide of many structural reforms. 3. Lower demand for money and need for The Mexican pegged exchange-rate regime tighter fiscal policy was totally different from the Argentine currency 4. Reduced growth board arrangement. For example, in 1994 Mexi- 5. A more fragile financial system co prevented the automatic adjustment mecha- nism that would have been consistent with a Economists disagree as to the causes of the fixed exchange rate by increasing the monetary crisis, and consequently, about capital controls. base when the peso wvas under pressure. Those economists who view the crisis as mostly The outstanding feature of a currency board due to exogenous factors would advise capital is its automatic adjustment rule, which is the controls. Economists who believe mostly in opposite of the discretion exercised by central endogenous causes will emphasize domestic bankers regarding sterilization and other mone- policies. tary control mechanisms. Most importantly, a 27 28 * BLACK DECEIMBER currency board imposes a hard budget constraint 1995 of $14.86 billion. on the treasury. Under this system, the treasury Furthermore,Argentina's government stead- can finance its deficit only with debt, either fastly maintained the currency board, basically domestic or foreign, but it cannot finance its played by its rules (accepting that money is deficit with an inflation tax. Also, the system's endogenous and there was little room for basis in rules allowvs it to withstand the kinds of increasing domestic credit), and weathered a pressures that central banks cannot. Independent sharp recession. After one year, the banking sys- central banks are not the answer.17 tem had recovered all its pre-crisis deposits. Argentina was greatly affected by the Mexi- The sharp reduction in capital inflows- can crisis. The main reason wvas the general per- and the fear that the government would repeat ception, mostly outside Argentina, that the the 1991 forced swap of government bonds for Argentine peso was grossly overvalued, and most time deposits-caused both a run away from important, that the nation's fiscal situation had banks in search of cash and a run on the peso in been deteriorating. It seemed that the time of search of dollars. The former is an increase in the reckoning for Argentina had arrived. Investors desired cash-to-deposit ratio, the latter an not wanting to be caught in an exchange-rate increase in desired velocity. Each of these events crisis began moving capital out of the country. has a different effect. A sharp increase in the Markets have good memory, and since both cash-to-deposit ratio causes bank failures and an Araentina and Brazil had confiscated financial exogenous reduction in the money supply. By xvealth, the fiscal accounts became a crucial vari- contrast, an increase in desired velocity of circu- able for investors to gauge. The "tequila effect" lation means decreases the money demand. hurt Argentina, not only because its currency The golden rule of a currency board is that had appreciated considerably since 1991, but also the ex-post money supply is completely endoge- because its fiscal accounts had deteriorated sub- nous (or demand-driven). In other -words, an stantially since the second quarter of 1994.18 excess demand for, or excess supply of, money is The next table shows the Argentine pre- automatically accommodated by changes in the crisis fiscal panorama. nominal money supply.19 There could be several The reaction of Argentina xvas an attempt sources of ex-ante exogenous changes in the to improve its fiscal stance; total revenues money supply; which are not associated with increased from a 1994 average of $14.88 billion changes in the demand for money. in the third and fourth quarter to a 1995 average For example, let's assume that the World of $17.16 billion, while total expenditures wxere Bank provides Argentina with the resources to kept at the samc quarterly average level during financc its fiscal deficit. In this casc, the Argen- Table 7 Argentina: Fiscal Panorama 1994 Quarter I i 111 IV Cocrent Revenues 12.50 12,87 12,3L 12,55 Current Expend tures 11,23 11,27 12,18 12.80 Surplus 1,27 1,60 0,½6 -0.25 Tctal Revenues 14,87 559 1t,5 ½,?6 Tota Expend rures 1415 13,49 1668 5,04 Tota Surp us 0.72 2 10 -0 03 -0,07 Total XP Surp us 0.70 1.63 -0.03 -0,17 In b ] on of pesos xP Surplus does not nclude pr va-izatio-o proceeds SoLurce: Bo etin Oficia , Secretaria de Hacienda AFT E RSH O C KS OF THE M EXI CAN CR IS IS 29 tine treasury deposits those dollars in the central 1994, total deposits in the banking system bank in exchange for pesos. The treasury, when increased by 200 percent. The crisis, which began spending those pesos, creates an exogenous in December, had by March 1995 reduced excess supply of money. The market adjusts this deposits by about 13 percent. Furthermore, the disequilibrium through a current account cash-to-deposit ratio had increased by about 50 deficit-and through inflation if the inflow of percent. Both events created great distress in the foreign capital is not a one-time event.20 financial community and posed a threat to the Another example of an exogenous change payment system. in money supply is changes in the cash-to- Despite its benefits, the currency board deposit ratio.The increase in the cash-to-deposit framework does not have a lender-of-last-resort ratio in Argentina, for 1994-V through 1995-1 mechanism. For example, during the existence of caused both a liquidity squeeze on the banking currency boards in colonial Africa, banks were system and a reduction in the money supply.This foreign-owned and, if necessary, they had a type of reduction in the money supply is in fact lender of last resort in the major financial cen- exogenous (driven by changes in the multiplier ters. One way to prevent liquidity crises is to of the monetary base). The only way that a implement a narrow banking system whereby change in the cash-to-deposit ratio would have banks have 100 percent reserves against demand no exogenous effect on money supply would be deposits. All other deposits are really "certificates in a banking system with 100 percent reserves, of participation" (similar to mutual funds), and and this was not the case in Argentina. A curren- the investors do not have an ex-ante fixed return cy board regime coupled with a banking system bearing the risk of the banks' investments. This, that has fractional reserve requirements and very in fact, was the thrust of the Friedman-Simons limited deposit insurance has little defense proposal to avoid banking crises in the U.S. against a liquidity crisis because there is no Chart 4 shows the important shifts in port- lender of last resort. folio composition in Argentina between 1992 The main reason for the Argentine banking and 1996. From January 1993 until October problems were a shift away from money (reduc- 1994 there is a gradual substitution of peso tion in money demand) and a reshuffling of bank deposits in favor of dollar deposits. After Novem- portfolios to-ward cash and away from deposits. ber there is indeed a run against peso deposits The next graph shows both features-the fall in both relative to dollar deposits and to cash. Given total deposits and the increment in the cash-to- the fear of devaluation, the substitution against deposit ratio. peso deposits made the banks potentially weaker Notice the impressive growth in total because of the peso-to-dollar structure of their deposits. From January 1992 through November Chart 4 Chart 3 Argentina: Portfolio Substitution Argentina: Total Deposits and Cash 0.60 1.4 0.60 60000 \ Rato of peso to odollar depos ts >.3 0 55 Tota Deposits 50000 055 ' 1.2 0.50 1.1 n c,) X X 0 ~~~~~~~30000 t 10l 0.45 0-4 200000. Caso Deposits Ratio Ratio of oash -Lo peso depositsi 0 40 ,I. C , _____, _., ___C _., ____, _ . , . 10000 0.40 , , , ,,,.,,, ______,, ________.,,,,,',,, _ 08 92:0i 92.07 90:01 93:07 94:01 94:07 95:51 95:07 95.0 1992 I393 1994 1995 9C96 Source: Carta Econo-ica Source: Caeta Economica 30 * BLACK DECEVMBER assets. This structure was rigid in the short run. ly. Under the 1844 Act currency issue in England The next chart shows the real quantity of was subject to 100 percent gold reserves, so that money and the ratio of liquid money to total the Bank of England could not always satisfy the money. It shows that the Argentine banking sys- demand for Bank of England notes during a cri- tem has been hit by both a moderate shrinking sis unless the government suspended convertibili- of intermediation resources and a liquidity ty. Bagehot's advice to the Bank of England in crunch. 1873 has been summarized by Meltzer (1994) as follows: Chart 5 Argentina: Stock of Money in Dollars and Ratio Ml to M3 1. The central bank is the only lender of last resort in the monetary system. 0.64 35000 2. To prevent illiquid banks from closing, 0.62 real moneo, - the central bank should lend on any collateral 0.60 that is marketable in the ordinary course of busi- 0.58 25000 ness. It should not restrict lending to paper eligi- ble for discount at the central bank in normal 0.56 20000 periods. 0.54 M1/M3 3. Central bank loans, or advances, should 1500C be made on demand at a rate of interest above 0.52 the market rate. This discourages borrowing by 0.50 10000 those who can obtain accommodation in the 92:01 92:07 93:01 93:07 94:01 94:07 95:01 95:07 96:01 market. Source: Car-a Econom ca 4. The above three principles of central bank behavior should be stated in advance and Notice the sharp increment in the liquidity followed in a crisis. ratio during the first quarter of 1995.This incre- ment is much higher than the expected seasonal- Under its convertibility law, Argentina could ities observed before. Also, notice that the liquid- only increase the monetary base if backed by ity shift coincides with a significant fall in the U.S. dollars and a small share of dollar-denomi- real quantity of money. nated debt. In February 1995 the government Responding to the dramatic increase in the changed that rule, allowing the Central Bank to desired cash-to-deposit ratio, the Argentine open a rediscount window.21 Bagehot's rule for Central Bank opened a limited rediscount facili- crisis management and the rediscount window of ty. Some interpreted the bank's move as aban- the Argentine Central Bank are compatible with doning the core elements of its currency board, both the gold-standard rule and the currency- which it in fact did not do. If the rediscount board rule. This holds true as long as central window is used only to offset exogenous banks restrict their rediscount activity to the changes in money supply due to changes in the management of banking liquidity crises. money multiplier, there will be no violation of The objective of the new rediscount facility the golden rule of currency boards. This golden to offset changes in money supply sprang from rule, as stated above, makes changes in the mon- changes in the cash-to-deposit ratio. In fact, the etary base an exact reflection of changes in the Argentine Central Bank could be subject to the income velocity of money. This behavior ensures same criticism that Bagehot launched against the a new stable equilibrium due to its effect on Bank of England a century ago. In his book interest rates. Lombard Street, Bagehot criticized the Bank of During the 1995 banking crisis the Argen- England, not because it lent in crisis but for hesi- tine Central Bank seems to have followedWalter tating and doing it too late (Meltzer, 1994). Bagehot's advice, albeit belatedly and unwitting- Under its new rule the Argentine Central Bank A F TE R S HO C KS OF T HE ME XICAN C R IS IS * 31 still continues to back up its issue of cash with Following the currency board framework, U.S. dollars. the central bank should be closed and a Curren- The change in the composition and quanti- cy Board PLC should be created. The Argentine- ty of the financial portfolio experienced during peso interest rate is greater than the Argentine- the crisis had little effect on total credit to the dollar rate (due, of course, to devaluation private sector due to the limited rediscount expectations), and the Argentine-dollar rate is granted by the Central Bank, and most impor- greater than the U.S. Treasury-bill rate (due to tant, to the reduction of quite high (45 percent) country risk); lending real rates are falling but reserve requirements. Nevertheless, in the after- still very high, and this affects the banking port- math of the crises, new lending was mostly to folio and long-term growth. Closing the Central the government (a lot to the states, with a federal Bank, along with fiscal reform measures, would guarantee) and to AAA borrowers. boost confidence in the government's claim that the convertibility law will not be repealed. WHAT TO Do IN THE FUTURE? In 1996 the central bank, aware that it UNEMPLOYMENT AND THE REAL might not be able to act as a lender of last resort, EXCHANGE RATE negotiated with foreign banks a commitment to lend to it about U.S.$5 billion in case of a run Argentina's strength-its currency board-is on the banks. This is an step in the right direc- also its weakness, evident in the inflexibility of its tion, but it would be not enough to fend off a exchange rate to real depreciation. The way serious run (the total money supply M3 is about Mexico handled its devaluations during the $55 billion). 1994-95 crisis has not helped to promote the In most countries, including Argentina, the idea that it is much less painful to devalue than banking system alters money supply through to deflate. So far, Argentina has chosen to deflate, changes in credit and vice versa; consequently, keeping its nominal exchange rate unchanged. If money creation is inextricably linked to credit the nominal exchange rate is fixed, the only way creation. Furthermore, in this system any signifi- to achieve a real depreciation is by a fall in prices cant increment in the cash-to-deposit ratio cre- and nominal wages. If this is not achieved, the ates the threat of a collapse of the whole banking price of real devaluation will be larger unem- sector, no matter how solid the sector actually is. ployment. Argentina already has a historically A much more resilient financial system for high unemployment rate; therefore, the situation Argentina would be one that separates the two could become socially difficult. functions of the banking sector-administering Keeping the nominal exchange rate and the payment system and intermediating across adjusting through deflation has some benefits. savings and investment. This system was proposed For one, to keep afloat, businesses have to incor- in the '30s by the "Chicago School" of banking porate new technology and cost-saving measures. reform, whose most important proponents were In other words, they have to increase productivi- Simons, and later, Friedman. ty to survive. For another, the hard budget con- The basic tenet of the proposal is for the straints imposed by the currency board have banking system to have 100 percent reserves on prompted the Argentine government to slash demand deposits and a credit department where more federal expenditures, to privatize its minor- the only depositors are risk-taking investors.This ity stake in business enterprises, to put a cap on system is invulnerable to changes in the cash-to- large pensions, and to force some provinces to deposit ratio. Naturally, no institutional set-up privatize their provincial banks. Had the govern- could prevent a run away from the whole bank- ment faced a soft budget constraint implied by a ing system, but a narrow banking system would floating exchange rate, these measures would minimize the damage. probably have not taken place. 32 * B LAC: K DEC: E MB ER Argentine domestic currency appreciation depreciated significantly. Nevertheless, its value can be offset by increased total factor productivi- still is 60 percent relative to that of the second ty and by real devaluation of the dollar. Produc- quarter of 1990. tivity has been increasing, but mostly in the labor sector. Current evidence suggests that total-factor productivity has increased only modestly. At the Chart 6 same time, the dollar has devalued in the last Argentina: Real Exchange Rate same time, the doRar has devaiued in thelastDollar Exchange Rate, Weighted by Argentinian three years (1994-96) about 40 percent relative Trade Pattern to the yen and 22 percent relative to the Deutsche mark.23 This has surely helped the peso. which 0.6 because of its fixed value to the dollar, has deval- ued in a similar fashion as the dollar. o60 The next chart shows the Argentine real exchange rate.23 Although the domestic currency 0ss - has appreciated substantially since 1991, we do not know how far it is from equilibrium if any. 0.50 \ Most importantly, the chart shows that the domnestic currencv hlas experienced an irnportant reversal to steady'appreciation. By the end of 91:1 91:3 92:i 92:3 931 93:3 94:1 94:3 95.1 95:3 96:1 1993 the real appreciation of the peso reached its * Souce: lte,rat oral Firanc al StafisTics 2nd Tirecticr of Trade S-at st cs troughl and from then until the end of 1995, it ncpx Iq4 i= on PEGGED VERSUS FIXED EXCHANGE RATES: ARGENTINA AND MEXICO COMPARED AS WE HAVE DISCUSSED, Argentina's and Mexico's institutional frameworks were completely different. Argentina had a currency board, while Mexico had an exchange-rate crawling-band. How do the two regimes compare? In a true fixed and automatic exchange rate show any significance.The results are presented regime the changes in foreign exchange are in Table 8 and 9. experienced at the same time as the changes in Table 8 which relates stocks is not conclu- the monetary base. No lag nor lead would be sive but Table 9, which relates flows, shows that detected. In an active monetary policy, in the context of fixed exchange rate, we would find Table 8 ,from money (or domestic Mexico: Stock of Foreign Assets and Stock (Granger) causality from money tor uomestlc of Monetary Base Causality Tests credit) to foreign assets if the Central Bank took Null hypothesis F-statistic Probability the initiative and increased the monetary base or ME-H is not G!anger Caused money supply. by MEFAUS 0.46 0.64 In the case of an active sterilization policy MEFA is not Granger Causeo we would find a (Granger) causality from foreign by MEH 1.09 0 34 assets to the monetary base. For example, if there was an exogenous shock causing reserve losses ags: two nonths and the Central Bank reacted increasing domes- MEH: Stcck of Monetary Base tic credit to sterilized it. Both effects could, or MEFAJS: Stock of Central Bank s hooding of foreign exchange could not, cancel each other out. It is interesting Period 19336-1994 1 to find out if these lead-lag relations in Mexico 33 34 a BLACK DECEMBER Table 9 Table 10 shows that, in Argentina, the mon- Mexico: Changes in the Monetary Base and etary base and foreign assets of the Central Bank Changes in Foreign Assets Causality Tests Null hypothesis: F-statistic Probability were highly correlated. Also, the real quantity of MEDH s not Granger CEUSECmoney was highly correlated with the foreign MEDH is not Granger Caused by MEDFA 0.021 0.99 assets of the Central Bank. This is to be expected MEDFA rs not Granger Caused in a fixed exchange-rate regime. Mexico shows a by MEDH 4.182 0.03 negative correlation between base money and foreign assets and very small correlation between MEDH = Changes n -e Monetary Base movements in real money and movements in MEDFA - Cnanges In Bank of Mex co's nternat oral reserves foreign assets.This is additional evidence that L.ags: two months Mexico followed a loose rule regarding passive monetary policy. changes in the monetary base (Granger) caused changes in international reserves. The results of Table 9 do not "prove" anything. They simple Table 10 Correlations add more circumstantial evidence to the claims (Coefficient of determination) that endogenous changes in domestic credit had Base Money and FA Real Money and FA a relevant role in explaining reserve losses. Argent na .84 .77 Another test highlighting the differences Mexico -.16 .18 between the Mexican-managed pegged FA = Fo,eign Assets ho dings of the Central Bank exchange rate and the Argentine institutionally Money is Ml fixed rate is to look at the correlation between the monetary base and the foreign assets of the central banks. For Argentina we expect that Next, Chart 7 shows the ratio of foreign changes in the Central Bank's holdings of foreign assets held by the Central Bank to the monetary assets would lead to changes in the monetary base in Mexico and Argentina. base in the same direction-a positive correla- tion. If Mexico had an active monetary policv Chart 7 on average, we would expect that changes in Mexico and Argentina: Ratio of Central Bank's ' ~~~~~~~~~~oreign Assets to Monetary Base money supply would cause changes in the oppo- site direction in foreign assets-a negative 2.0 correlation. Under a true fixed exchange-rate regime, 6 E Mexico -" we would also expect that movements in the foreign assets of the central bank would be 1.2 strongly correlated with movements in the real demand for money. Under this regime, the 0.8 Argent na transmission mechanism is the following: an increment in the real demand for money changes interest rates; this affects capital flows, 04 93:3 941 94:3 95:1 95:3 96:1 and foreign assets are channeled to the central Source: In'ernational Financ aStatistics bank in exchange for domestic currency. Table 10 presents the correlations among base money, real quantity of money, and foreign-asset hold- Mexico clearly shows four distinct periods: ings of the central banks oflArgentina and Mex- . June 1993 through February 1994: The ico forJune 1993 through November 1994.We Central Bank has a very comfortable reserve have omitted December 1994, because of the position. Its foreign exchange holdings are twice Mexican devaluation. the monetary base. PEGGED VE RS US FIXE D EXC HAN GE RATES * 35 2. March through April 1994: In a two- bottom part of the chart mean that capital month period, foreign reserves plunge, becoming flows are larger than the current account roughly equal to the monetary base. deficit. 3. May through October 1994: Foreign This chart shows that, between 1993-I and reserves are kept stable and equal to the mone- 1994-I, Mexico's current account deficits were tary base. indeed smaller than its capital inflows, and the 4. November through December 1994: The Central Bank was accumulating reserves. This collapse occurs, and foreign exchange becomes a picture changed drastically in 1994, with the fatal small fraction of the monetary base. quarters being the second and the fourth. (As shown in Chart 8, the crucial months were Argentina shows the pattern forced on it by March through April and November through the convertibility law, whereby the Ccntral Bank December, with some recovery in the third is required to hold almost 100 percent interna- quarter and a final debacle in the fourth. In the tional reserves. Notice that from April to Octo- second quarter of 1994, the current account ber 1994, Mexico's Central Bank had dollar deficit was about U.S.$7 billion and the loss in backing equal to 100 percent of its monetary international reserves was about U.S.$12 billion. base.This was the reason many analysts were still This was clearly the definition of insustainability. confident about the future of Mexico. Also, and The next chart shows the income veloci- most importantly, the data available in November ty of money and the rate of growth in the and early December 1994 dated back to March monetary base for January through November and April of that year. 1994. Chart 8 shows the changes in the interna- Chart 9 tional reserves of Argentina and Mexico relative Mexico: Income Velocity of Money and Growth of to the flow of their current accounts. Positive Money Base figures mean that changes in foreign assets of 03 1.to the central banks have been financing current account deficits.26 Negative figures seen at the 02 Veloity l1.05 Chart 8 Argentina and Mexico Current Account Financed with Changes in CB's Foreign Assets 0.095 200 - Percentages B.0ase Crth ' 150 C Mexico -0.2 . . .._.._ 0.85 CA Deficits and Reductions in FA 93:01 93:04 93:07 93:10 94:01 94:04 94:07 94:10 95:01 100 - In this area CA def cits are arger Source: Internationa F nancia Statist cs than capital inflows 50- ,"-the /\Chart 9 makes clear that, after May 1994, o -7,,' /====_ \ the Central Bank of Mexico was actually Argentina increasing the rate of growth of the monetary -50CA Deficits and Increments in FA base at the same time that the money demand Ir this area CA deficits are smaller than capital inflovs was decreasing. The reaction of a system dedicat- 93.1 93.2 93.3 93.4 94.1 94.2 94.3 9 4 ed to supporting a fixed exchange rate should CA: Currert Account have been the opposite. In particular, notice the CB: Central Bank | MEFACA -ARDFACA explosive increment in the rate of growth of the FA Foreign Asse's illmonetary base during October and November Sourcea nternational Financial Statast cs and WVcrld Bank 1994. 36 * B LA C K DEC E MNB E P SUMMARY AND CONCLUSIONS income during 1995.The banking bail-out had a large fiscal cost that hiked interest rates and The main cause of the Mexican crisis was expected inflation. not an unsustainable current account deficit. The Mexican crisis hit Argentina with Although the current account was indeed unsus- force, mainly because the latter's exchange rate tainable, this was a symptom of the problem and was deemed overvalued relative to its weakening not a cause. The main cause of the Mexican cri- fiscal stance. The Argentine currency board sis was not, as some analysts have suggested, responded to the sharply reduced capital inflows political violence and investors' herd instincts. and according to its automatic rules reduced the Political violence certainly did not help the situ- monetary base. This action, combined with the ation, with each episode causing a one-time banking crises, reduced money supply by 20 per- increment in the income velocity of money. cent during the first five months of 1995, causing These increments, however, were not a crucial a drastic reduction in real income of about 4.5 determinant of the crisis. percent. The main cause of the Mexican crisis was Argentina needs a "confidence shock" in an inconsistent monetary and exchange-rate pol- order to increase its capital inflows and reduce icy that caused both an increment in the current the dollar-peso interest-rate gap and the country account deficit and a deficit in the balance of risk. These problems could be addressed by the payments.The increment in the current account following actions in specific sectors: deficit (a flow) was increasingly financed by 1. Fiscal: Argentina should tackle its fiscal international reserves (a stock), which caused a weakness by reducing expenditures by a perma- balance-of-payment deficit that in turn made the nent and significant amount,25 and not by exchange-rate peg unsustainable. When the stock increasing the tax burden. In fact, a fiscal reform was depleted, the crisis ensued with full force. should reduce the very highVAT tax rate and Because of lack of timely disclosure of key increase the tax base. niarket inforniation-mainly about Central Bank 2. Institutional: Argentina should close its reserves, domestic credit, and current account Central Bank, which under a currency board deficits-disequilibriums in the system were arrangement has no role, and instead form a allowed to build up. Currency Board PLC. The Mexican banking sector suffered 3. Banking: Argentina should direct its because of a lack of proper capitalhzation that banking system to the narrow banking alterna- could not absorb the increment in a non-per- tive, thus separating money from credit and get- forming portfolio. The portfolio deteriorated ting rid of deposit insurance. because of the large changes in relative prices, 4. Labor: Argentina should carry on its labor mainly the exchange rate; the sharp increment in deregulation, which is crucial to reducing the interest rates; and the drastic 7 percent fall in real current rate of unemployment. NOTES l Due to a significant time-lag in the disclosure of cal expenditures.The model left out a crucial variable the Mexican Central Bank data, economists were not axware needed to restrain government expenditures: A ceiling on of the large increase in domestic credit, the reduction in government debt. money demand, and the fall in international reserves expe- " For an excellent review of the literature concern- rienced by October through November and December ing different explanations for the high interest rates 1994. However, early during that year, some economists observed during stabilization periods see Philip L. Brock, were already concerined about the sustainability of the cur- "High Interest Rates and Guarantor Risk in an Open rent account deficit and by the rate of increase in domnestic Economy: A Case Study of Chile, 1975-1 983" (paper writ- credit. The number of economists who claimed that they ten for the World Bank and presented at the first meeting had correctly predicted the Mexican crisis increased sub- of Central Bank Research Centers, Mexico, 1996). stantially ex-post. 12 Sterilization of capital inflows has been widely 2 Mexico did not have a strict fixed exchange-rate used in many Latin American countries (particularly in regime, but for analytical purposes, we can assume that it Brazil), although it has been proven to be perverse in two had one. Mexico's debt crisis of the early '80s was very dif- ways. First, it increases the fiscal deficit, since the cost of ferent from that of the 1994-95 crisis. In the late '70s most domestic debt is higher than the return on international of the outstanding debt was contracted under floating inter- reserves. Second, it increases the incentives for continued est rates, with financing costs mounting as world interest capital inflows, causing a vicious circle. rates skyrocketed. In addition, the worldwide oil-price col- 13 In a world of no exogenous capital flows and a lapse helped trigger the crisis by drying up petro-dollars, the fixed exchange rate, there would be no problem on the so-called source of capital flows. money market, because money demand determines money 3 Central Bank of Argentina, Boletin Mensual, several supply. Nevertheless, when there are exogenous flows, even issues. though it is still true that money demand ultimately deter- 4 For a description, see Alex Fleming, Lily Chu, and mines money supply, the money multiplier complicates the Marie-Renee Bakker. "The Baltics-Banking Crises process and imposes undesired variability and noise to the Observed," (Policy Research Working Paper Number 1647, money and credit markets. World Bank,Washington, D.C., September 1996). 14 See Larry Sjaastad, "Deposit Insurance," (In Pre- S In this report United States currency will be designat- venting Banking Sector Distress and Crises, proceedings of ed as "U.S.$"; otherwise, the "S" symbol will refer to pesos. a World Bank conference, 1996) for an insightful presenta- 6 World Bank estimates. tion on the deleterious effects of fractional reserve require- International Monetary Fund country reports. ments and deposit insurance. 8 Here we are considering a one-shot capital inflow. I5 Some researchers have claimed that an additional If the capital inflow continues at the same constant rate, all endogenous force in Chile was its indexation rule. Chile things being equal, the current account deficit also will had a backward indexation of wages, and due to falling continue at the same pace, and the initial change in the inflation, the indexation rule would have caused a over- international reserves of the central bank will remain shooting of real wages. This explanation is not satisfactory constant. because employment was increasing and unemployment 9 In the case of Argentina, the current account deficit had reached very low figures. financed a large share of new, imported, capital goods. 16 We define capital flows in terms of the capital 10 Argentina in 1979 had a tablita that was a pre- account balance. A capital inflow would be associated with announced path of the nominal exchange rate. It also had a a surplus in the capital account and a capital outflow with a domestic credit tablita to make both the exchange rate deficit.We estimate the balance in the capital account as the tablita and the targeted increment in international reserves change in international reserves less the balance in the cur- consistent.This straightjacket proved not enough to halt fis- rent account. 37 N o T F S U 38 '7 Larry Sjaastad of the University of Chicago has 22 During 1997-I the dollar has appreciated many times claimed that an independent central bank does substantiallv. not necessarily mean independent central bankers. 23 The real exchange rate is the nominal exchange 1 The Argentine 1994 fiscal accounts were much rate for the United States, weighted by the Argentine trade better than its 198(0 fiscal accounts, but experience had pattern and multiplied by the ratio of the respective con- taught Argentines a harsh lesson, and they had learned it sumer price indexes of Argentina's major trading partners well: Income velocitv of money (MI) -was 10 in 1979, and to the Argentine consumer price index. These countries, after three years of stability, it wvas still 17 in 1994. It is well representing 90 percent of Argentine trade, are: Germany, known that the same fiscal deficit has different inflation Brazil, Chile, Spain, Holland, Italy,Japan, Uruguay, and the implications for different levels of income velocity of United States. money. 24 We are ignoring here changes in the demand for 19 There will also be changes in the price level if the dollar holdings by the population. non-traded scctor is large. 25 Therc is room for substantial expenditure reduc- 2U1 In some instances, some countries have issued tion according to a study conducted by FIEL, "La Admin- domestic debt to sterilize the effect of its foreign borrowing istracion Publica Nacional," 1996. Buenos Aires: C.E.A. on the domestic money supply. This has prevented the Also there is room for reduction in the expenditures of the exchange rate from appreciating at the cost of increasing the States. country's fscal burden due to high domestic interest rates. 21 This rediscount facility has a hioit equal to the stock of commercial banks' required reserves. BIBLIOGRAPHY Brock, Philip L. 1996. "High Interest Rates and Guarantor Diversification: Evidence from Emerging Financial Risk in an Open Economy: A Case Study of Chile, Markets." In Portfolio Investment in Developing Coun- 1975-83." Paper presented at the first meeting of the tries, 145-68.World Bank Discussion Paper 228, edited Central Bank Research Centers. Mexico, 1996. by Stijn Claessens and Sudarshan Gooptu.Washington, Buckberg, Elaine. 1993. "Emerging Stock Markets and D.C.:World Bank. International Asset Pricing." In Portfolio Investment in Dooley, Michael, Eduardo Fernandez-Arias, and Kenneth Developing Countries,World Bank Discussion Paper Kletzer. 1994. "Is the Debt Crisis History? Recent Pri- 228, edited by Stijn Claessens and Sudarshan Gooptu, vate Capital Inflows to Developing Countries." Policy 169-99. Washington, D.C.: World Bank. Research Working Paper 1327. World Bank, Interna- Calvo, Guillermo. 1994. "The Capital Inflows Problem: tional Economics Department,Washington, D.C. Concept and Issues." Contemporary Economic Policy Edwards, Sebastian. 1995. Crisis and Reform in Latin 12 (3): 54-66. America: From Despair to Hope. New York: Oxford Calvo, Guillermo, Leonardo Leiderman, and Carmen M University Press. Reinhart. March 1993. "Capital Inflows and Real El-rian, Mohamed. 1992. "Restoration of Access toVolun- Exchange Rate Appreciation in Latin America," tary Capital Market Financing: The Recent Latin Amer- 108-51. International Monetary Fund Staff Papers ican Experience" International Monetary Fund Staff 40-1. Paper 39 (1). Calvo, Guillermo A., and Mendoza Enrique G. "Petty Fernandez, Roque and Liliana Schumacher. 1997. "Narrow Crime and Cruel Punishment: Lessons from the Mexi- Banking." In Preventing Banking Sector Distress and can Debacle." The American Economic Review, Papers Crisis, proceedings of a World Bank conference, edited and Proceedings, January 1996. by S. Bery and V Garcia. Washington, D. C.: World Bank. Calvo Sara, and Carmen Reinhart. "Capital Flows to Latin Fernandez-Arias, Eduardo. 1994. "The New Wave of Pri- America: Is There Evidence of Contagion Effect?" In vate Capital Inflows: Push or Pull." Policy Research Capital Flows to Emerging Markes, edited by Morris Working Paper 1312.World Bank, International Eco- Goldstein.Washington Institute for Internaional Eco- nomics Department, Washington, D.C. nomics.Washington, D.C.: 1996 Friedman, Milton, 1959. "A Program for Monetary Stabili- Chuhan, Punam, Stijn Claessens, and Nlandu Mamingi. ty." Fordham University Press: New York. 1993. "Equity and Bond Flows to Asia and Latin Amer- Friedman, Milton, and Anna Schwartz. 1963.A Monetary ica:The Role of Global and Country Factors." Policy History of the United States, 1867-1960. Princeton: Research Working Paper 1160. World Bank, Washing- Princeton University Press. ton, D.C. Garcia,Valeriano. 1997. "Postscript." In Preventing Banking Corbo,Vittorio, and Leonardo Hernandez. 1993. "Macro- Sector Distress and Crises, proceedings of a World Bank economic Adjustment to Capital Inflows: Some Ratio- conference, edited by S. Bery andV. Garcia.Washington, nale and Some Recent Experiences." In Portfolio D.C.: World Bank. Investment in Developing Countries,World Bank Dis- Gavin, Michael, and Ricardo Houseman. June 1996.""The cussion Paper 228, edited by Stijn Claessens and Sudar- Roots of Banking Crises: The Macroeconomic Con- shan Gooptu, 353-82.Washington, D.C.:World Bank. text."Working Paper 318. Interamerican Development Corden,W Max. 1984. "Booming Sector and Dutch Dis- Bank, Washington, D.C. ease Economics: Survey and Consolidation." Oxford Gcrtler, Mark, and Andrew Rose. 1991. "Finance, Growth, Economic Papers 36 (3): 359-80. and Public Policy" Policy Research Working Paper 814. Corden,W Max, and J. Peter Neary. 1982. "Booming Sec- World Bank,Washington, D.C. tor and De-Industrialization in a Small Open Econo- Gil-Diaz, Francisco, and Carstens. "OneYear of Solitude: my" Economic Journal 92: 825-48. Some Pilgrim Tales About Mexico's 1994-1995 Crisis." de La Cuadra, Sergio, and SalvadorValdes. 1992. If Texas The American Economic Review, Papers and Proceed- Were Chile. San Francisco: Institute for Contemporary ings,January 1996. Studies. Giorgio, Luis Alberto. 1996. "Banking Distress and Crisis: De Santis, Giorgio. 1993. "Asset Pricing and Portfolio Lessons from Selected Experiences in T.ani America." 39 B I B L I O G R A P H Y u 40 Working Paper.World Bank,Washington, D.C. NewYork: M. E. Sharpe Publishing Co. Gooptu, Sudarshan. 1993. "Portfolio Investment Flows to Rodriguez, Carlos Alfredo. 1994. "Interest Rates in Latin Emerging Markets." In Portfolio Investment in Devel- America." Internal Discussion Paper.World Bank, Latin oping Countries, 45-77. World Bank Discussion Paper America and the Caribbean Region,Washington, D.C. 223, edited by Stijn Claessens and Sudarshan Gooptu. Rojas-Suarez, Liliana, and Steven R.Weisbrod. 1995. Washington D.C.:World Bank. "Banking Crises in Latin America: Experience and Gorton, Gary. 1998. "Banking Panics and Business Cycles," Issues." Prepared for the Interamerican Development Oxford Economnic Papers, 40: 751-81. Bank Conference on Banking Crises in Latin America, Harvey, Campbell R. 1993. "Portfolio Enhancement Using Washington, D.C., October 1995. Emerging Markets and Conditioning Information." In Sjaastad, Larry. 1994. "On Exchange Rates, Nominal and Portfolio Investment in Developing Countries, 110-44. Real."Working Paper. University of Chicago, Chicago, World Bank Discussion Paper 228, edited by Stijn Ill.Sjaastad, Larry. 1997. "Deposit Insurance." In Prevent- Claessens and Sudarshan Gooptu. Washington, D.C.: ing Banking Sector Distress and Crises, forthcoming World Bank. proceedings of a World Bank conference, edited by S. Kaminsky, Graciela L., and Carmen M. Reinhart. 1995. Bery and V. Garcia. World Bank, Washington, D.C. "The Twin Crises: The Causes of Banking and Balance Talvi, Ernesto. 1996. "Exchange Rate-Based Stabilization of Payment Problems." Paper prepared for the Inter- with Endogenous Fixed Response.?Workiing Paper 324. american Development Bank conference on Speculative Interamerican Development Bank,VWashington, D.C., Attacks in the Era of the Global Economy Washington, March 1996. D.C., 1995. Tesar, Linda, and Ingrid Warner. 1993. "U.S. Equity Invest- Mathieson, Donald J., and Liliana Rojas-Suarez. 1992. ment in Emerging Stock Markets." In Portfolio Invest- "Liberalization of the Capital Account: Experiences and ment in Developing Countries, 200-220. World Bank Issues." Occasional Paper 103. International Monetary Discussion Paper 228, edited by Stijn Claessens and Fund, Washington, D.C. Sudarshan Gooptu.Washington, D.C.:World Bank. Meltzer, Allan H. 1995. "Sustaining Safety and Soundness: Williamson,John. 1993."Issues Posed by Portfolio Invest- Supervision, Regulation, and Financial Reform."Work- ment in Developing Countries." In Portfolio Investment ing Paper.World Bank,Washington, D.C. in Developing Countries, 11-17.World Bank Discus- Mundell, Robert. 1971. Monetary Theory. Pacific Palisades: sion Paper 228, edited by Stijn Claessens and Sudarshan Goodyear Publishing Co. Gooptu.Washington, D.C.:World Bank. Phillips, Ronnie J. 1995.The Chicago Plan and New Deal World Bank, 1994.World Debt Tables: External Finance Banking Reform. for Developing Countries, vol. 1, Analysis and Sunrmary Tables. Washington, D.C.:World Bank. WORLD BANK LATIN AMERICAN AND CARIBBEAN STUDIES VIEWPOINTS SERIES Latin America after mexico: Quickening the Pace by Shahid Javed Burki and Sebastian Edwards Poverty, Inequality, and Human Capital Development in Latin America, 1950-2025 by Juan Luis Londofio available in English and Spanish Pobreza, desigualdad yformacion del capital humano enAme'rica Latina, 1950-2025 Juan Luis Londonio Dismantling the Populist State: the Unfinished Revolution in Latin America and the Caribbean by Shahid Javed Burki and Sebastian Edwards Decentralization in Latin America: Learning through Experience by George E. Peterson Urban Poverty and Violence inJamaica by Caroline Moser and Jeremy Holland La pobreza urbana y la violencia enJamaica Caroline Moser yJeremy Holland Prospects and Challengesfor the Caribbean by Steven B.Webb Black December: Banking Instability, the lMexican Crisis and Its Effect on Argentina byvValeriano E Garcia PROCEEDINGS SERIES Currency Boards and External Shocks: How Much Pain, How Much Gain? Edited by Guillermo E. Perry Annual World Bank Conference on Development in Latin America and the Caribbean: 1995 Edited by Shahid Javed Burki, Sebastian Edwards, and Sri-Ram Aiyer Annual World Bank Conference in Latin America and the Caribbean: 1996 Poverty and Inequality By Shahid Javed Burki, Sri-Ram Aiyer and Rudolf Hommes 4s, THE WORLD BANK 1818 H Street, N.V Washington, D.C. 20433 USA Telephone: 202-477-1234 Facsimile: 202-477-6391 Telex: MCI 64145 WORDI)BANK MCI 248423 WORLDBANK Cable Address: INTBAIFRAD WVASHINC;rONDC World Wide Web: http://ww.vworldbank.org/ E-mail: booksCa@worldbank.org 1 1 lj 3 9 6 ' 9 780821 339602 ISBN 0-8213-3960-5