WORLD BANK DISCUSSION PAPER NO. 360 WDP 360 Work in progress for public discussion M1a1rch Iqq7 Preventing Banking, Scctor Distress and Crises in Latin Amlerica P/ ,o(r,,'//i/4% ,/ (, (Io/6',,f ,/,, /,ldi I 1/Sh /'//(-to//, 1). (, , ./7/I 1-1, 1996 1w " 1~~~~~~~~~~~,, F. ("l' Recent World Bank Discussion Papers No. 292 Regutlated Deregulation of the Financial System in Korea. Ismail Dalla and Deena Khatkhate No 293 Design Issutes in Rutral Finance. Orlando J. Sacay and Bikki K Randhawa No 294 Financing Health Services Through User Fees and Insutrance: Case Studiesfrom Sub-Saharan Africa. R Paul Shaw and Martha Ainsworth No. 295 The Participation of Nongovernmental Organizations in Poverty Alleviation The Case Study of the Honduras Social investment Fuind Project. Anna Kathryn Vandever WA.ebb, Kye Woo Lee, and Anna Maria Sant'Anna No. 296 Reforming the Energy Sector in Transition Economies. Selected Experience and Lessons. Dale Gray No. 297 Assessing Sector Institutions. Lessons of Experiencefrom Zambia's Education Sector. Rogerio F Pinto and Angelous J Mrope No. 298 Uganda's AIDS Crisis: Its Implicationsfor Development. Jill Armstrong No 299 Towards a Payments System Lawfor Developing and Transition Economies. Raj Bhala No. 300 Africa Can Compete! Export Opportunities and Challengesfor Garments and Home Products in the Eluropean Market. Tyler Biggs, Margaret Miller, Caroline Otto, and Gerald Tyler No. 301 Review and Outlookfor the World Oil Market. Shane S. Streifel No. 302 The Broad Sector Approach to Investment Lending: Sector Investment Programs. Peter Harrold and Associates No 303 Institutional Adjustment and Adjutsting to Insttiutions. Robert Klitgaard No. 304 Putting Institutional Economics to Work: From Participation to Governance. Robert Picciotto No. 305 Pakistan's Public Agricultuiral Enterprises. Inefficiencies, Market Distortions, and Proposals for Reform. Rashid Faruqee, Ridwan Ali, and Yusuf Choudhry No. 306 Grameen Bank Performance and Stability. Shahidur R Khandker, Baqui Khalily, and Zahed Khan No 307 The Urtguay Routnd and the Developing Economies. Edited by Will Martin and L. Alan Winters No. 308 Bank Governance Contracts: Establishing Goals and Accountability in Bank Restructuring. Richard P. Roulier No. 309 Puiblic and Private Secondary Education in Developing Countries: A Comparative Study. Emmanuel Jimenez and Marlaine E. Lockheed with contributions by Donald Cox, Eduardo Luna, Vicente Paqueo, M L. de Vera, and Nongnuch Wattanawaha No. 310 Practical Lessonsfor Africafrom East Asia in Industrial and Trade Policies Peter Harrold, Malathi Jayawickrama, and Deepak Bhattasali No 311 The Impact of the Unrgutay Round on Africa. Peter Harrold No. 312 Procurement and Disbutrsement Manutalfor Projects with Community Participation. Gita Gopal No. 313 Harnessing Information for Development: A Proposalfor a World Bank Groutp Strategy. Eduardo Talero and Philip Gaudette No 314 Colombia's Pension Reform: Fiscal and Macroeconomic Effects. Klaus Schmidt-Hebbel No. 315 Land Quiality Indicators. Christian Pierl, Julian Dumanski, Ann Hamblin, and Anthony Young No 316 Sustainability of a Government Targeted Credit Program: Evidencefrom Bangladesh. Shahidur R. Khandker, Zahed Khan, and Baqui Khalily No. 317 Selected Social Safety Net Programs in the Philippines: Targeting, Cost-Effectiveness, and Optionsfor Reform. Kalanidhi Subbarao, Akhter U. Ahmed, and Tesfaye Teklu No. 318 Private Sector Development Dutring Transition: The Visegrad Countries. Michael S Borish and Michel Noel No. 319 Eduication Achievements and School Efficiency in Rutral Bangladesh Shahidur R. Khandker No 320 Household and Intrahousehold Impacts of the Grameen Bank and Similar Targeted Credit Programs in Bangladesh Mark M Pitt and Shahidur R. Khandker No. 321 Clearance and Settlement Systemsfor Secutrities. Critical Design Choices in Emerging Market Economies. Jeff Stehm No. 322 Selecting Development Projectsfor the World Bank Jean Baneth No 323 Evaluating Puiblic Spending: A Frameworkfor Public Expenditure Reviews. Sanjay Pradhan No 324 The Bangladesh Rural Advancement Committee's Credit Programs Performance and Suistainability. Shahidur R. Khand- ker and Baqui Khalily No. 325 Institutional and Entrepreneutrial Leadership In the Brazilian Science and Technology Sector:Setting a New Agenda . Edited by Lauritz Holm-Nielsen, Michael Crawford, and Alcyone Saliba No. 326 The East Asian Miracle and Information Technology: Strategic Management of Technological Learning. Nagy Hanna, Sandor Boyson, and Shakuntala Cunaratne (Continuted on the inside back cover) WORLD BANK DISCUSSION PAPER NO. 360 Preventing Banking Sector Distress and Crises in Latin America Proceedings of a Conference held in Washington, D.C., April 15-16, 1996 Edited by Suman K Bery Valeriano F Garcia The World Bank Washington, D.C. Copyright C 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 March 1997 Discussion Papers present results of country analysis or research that are circulated to encourage discussion and comment within the development community. To present these results with the least possible delay, the typescript of this paper has not been prepared in accordance with the procedures appropriate to formal printed texts, and the World Bank accepts no responsibility for errors. Some sources cited in this paper may be informal documents that are not readily available. 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 Directors or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility whatsoever for any consequence of their use. The boundaries, colors, denominations, and other inforrnation 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 nornally 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. Cover art by Beatrice Sito. ISSN: 0259-210X Suman K. Bery is chief of, and Valeriano F. Garcia is principal economist in, the Economic Adviser's Unit of the World Bank's Latin America and the Caribbean Technical Department. Library of Congress Cataloging-in-Publication Data Preventing banking sector distress and crises in Latin America: proceedings of a conference held in Washington, D.C., April 15-16, 1996 / edited by Suman K. Bery, Valeriano F. Garcia. p. cm. - (World Bank discussion papers ; 360) Includes bibliographical references. ISBN 0-8213-3893-5 1. Banks and banking-Latin America-Congresses. 2. Financial crises-Latin America-Prevention-Congresses. I. Bery, Suman K. II. Garcia, Valeriano F. III. World Bank. IV. Series. HG2710.5.A6P74 1997 332.1'098-dc2l 97-7156 CIP Contents Foreword v Abstract vii Acknowledgments ix The Conference Proceedings 1 Suman K. Bery and Valeriano F. Garcia Opening Remarks 15 Sri-Ram Aiyer Banking Crises-A Prologue 17 Alan Walters Does Argentina Provide a Case for Narrow Banking? 21 Roque Fernandez and Liliana Schumacher Comments Gerard Caprio Jr. 32 George J. Benston 35 Fernando de Santibanies 40 Floor Discussion 43 Deposit Insurance: Do We Really Need It? 47 Larry A. Sjaastad Comments Paul M. Horvitz 52 Roberto Junguito 55 Allan H. Meltzer 56 Floor Discussion 59 Market-Based Banking Regulation 63 Peter Nicholl Comments RolfJ. Luders 69 Jacques Trigo Loubiere 73 Paul L. Bydalek 75 Floor Discussion 79 Roundtable Discussion: Systemic Banking Crises 83 Robert Eisenbeis 83 Valeriano F. Garcia 84 Sergio Ghigliazza 86 Allan H. Meltzer 86 Jose Evenor Taboada 88 Roberto Zahler 89 Rodrigo Bolaflos Zamora 91 Floor Discussion 93 Postscript 97 Valeriano F. Garcia Participants and Attendees 101 iv Foreword O ver the past two decades countries in Latin America and the Caribbean have continually suffered the blows of financial market volatility, producing repeated and acute systemic banking crises. Despite sig- nificant support from the World Bank through structural adjustment and financial sector adjustment loans, many of the region's banking systems continue to be weak and subject to recurrent distress. In an attempt to solve this pressing problem, the Technical Department of the Bank's Latin America and the Caribbean Regional Office organized this conference to discuss topics relevant to the prevention of bank- ing crises in these countries. The main objective of the conference was to present in one place the varying prescriptions at hand and ultimately devise a common framework that would render the systems less vul- nerable to systemic shocks. It is hoped that the compendium of ideas and analyses in this volume will prove of value to future policy reform decisions. Sri-Ram Aiyer, Director Latin American and the Caribbean Technical Department v Abstract The following pages are a collection of discussions on banking crises in Latin America-their origins and reme- dies-as seen through the eyes of bankers, practitioners, and scholars. This cross-sharing of experience should prove instrumental to policymakers in designing future programs geared at preventing banking sector crises thoroughout the region. vii Acklowledgments The collaborative efforts of the following people were instrumental in the preparation of this conference and the publication of this volume: Valeriano Garcia for planning the conference, Suman Bery for his guidance, the Research Committee for its support, Elizabeth Minoso and Patricia Mendez for coordinating the conference, and Jorge Forgues for his timely support. Thanks must also be extended to the office of the Director of the Technical Department and the office of the Vice President of the Latin America and the Caribbean Regional Office, whose support was essential in the admninistration of this event. ix The Conference Proceedings Suman K. Bery and Valeriano F. Garcia T he conference on Preventing Banking Sector followed in establishing reserve requirements? Crises in Latin America, sponsored by the Deposit insurance is another vexing problem. Is Economic Adviser's Unit in the Technical insurance necessary to prevent liquidity problems at Department of the World Bank's Latin America and one or a few banks from turning into a panic? Or is the Caribbean Regional Office, took place in the price paid-in terms of the resource misalloca- Washington, D.C. on April 15-16, 1996. About 160 tion that results from the moral hazard created by attendees included heads of central banks, private such insurance-unacceptably high? Are there good bankers, policymakers, public officials, and acade- and bad types of deposit insurance? Prudential reg- mics. The three conference sessions focused on nar- ulation has been thoroughly discussed in other row banking, deposit insurance, and market-based forums, said Aiyer, but market-based regulation is a supervision and regulation. In addition, the confer- new weapon. Do such regulations substitute for or ence closed with a roundtable discussion on systemic complement conventional on-site supervision? If so, banking crises. in what way? And in countries that have market- based regulation (Chile, New Zealand), has the sys- Opening Remarks tem been tested? In welcoming the conference participants, Sri-Ram Prologue Aiyer said that it was well-known that Latin American countries have had long-standing and In his prologue Sir Alan Walters suggested that Latin stubborn problems with banking distress, often America's inflationary past made its search for mon- threatening entire banking systems. Although the etary credibility especially difficult, since economies management of banking crises has been extensively in transition from hyper- or high inflation are partic- analyzed in many forums-including work done by ularly susceptible to crises. In such countries an insti- the Bank's Latin America and the Caribbean tutionally fixed exchange rate (through some type of Technical Department-Aiyer had thought it was currency board) is a tempting means by which to time to discuss issues related to crisis prevention in gain credibility. But such a move also carries collat- commercial banking. Thus he had asked the eral difficulties-as Argentina's recent experience Economic Adviser's Unit in his department to bring had vividly illustrated. Since a currency board together academics, bankers, and practitioners to arrangement does not guarantee prudent banking discuss issues in three pertinent areas: reserve but does eliminate the lender of last resort, banking requirements and narrow banking, deposit insur- crises, if they occur, will threaten the peg. ance, and market-based regulation. Thus narrow banking may seem like an attractive In selecting these topics Aiyer noted that reserve solution for such countries, since it nips the problem requirements in the region vary from very high to of portfolio quality in the bud. Despite this appeal, zero-and opinions regarding how high they should however, Walters advised countries to avoid narrow be also vary from 100 percent on demand deposits banking. In his view narrow banking would neces- (narrow banking) to zero. What principles should be sarily drive business out of the banking system and Suman K. Bery is chief and Valeriano F. Garcia is principal economist of the Economic Adviser's Unit in the Latin America and the Caribbean Technical Department at the World Bank. 1 The Conference Proceedings into the equity market, the debenture market, and and because of the reluctance of political leaders and other credit-creating institutions. Although this ordinary citizens to use taxpayers' money to bail out move would shuffle credit risks out of the banking banks, Argentina should aim to evolve in the direc- system-thus insulating banks and the payment sys- tion of narrow banking. tem from the vagaries of the credit market-shifting The authors posed two main questions: Is risk to the nonbank capital market would transfer Argentina's banking system in a transition toward business to much less transparent institutions, with narrow banking? And is narrow banking a better greater insider trading. As a result the government banking framework for Argentina, given the con- would still be tempted to intervene, as it routinely straints imposed by its monetary constitution on the has in the banking sector. traditional role of the Central Bank as lender of last To avoid bank runs, Walters recommended strict resort? To answer these questions the authors first adherence to the capital requirements of the Basle reviewed the history of narrow banking proposals- agreements, good regulation, and tight supervision. the so-called Chicago plan of 1933 and its subsequent In his view the government should, above all, act to variations. Fernandez and Schumacher noted that preserve the payments system, even in the event of the main appeal of a 100 percent reserve requirement banking fraud. Only after the payments system is is its ability to restore control over the money stock secured should the government or central bank deal to the central bank while removing incentives for with fraud or any other issue. excessive government intervention in the asset deci- Walters insisted that one important element sug- sions of banks, and in asset markets more broadly. gesting that banks require special supervision, as dis- The intrusiveness of the monetary authorities in turn tinct from the supervision of nonbanks, is that banks is linked to two characteristics of fractional reserve are susceptible to large-scale fraud. The massive banking: deposit insurance and the dependence of losses run up by Banco de Intercambio Regional in the money stock on the independent portfolio deci- Argentina during the early 1980s and, more recently, sions of banks and the nonbank public, which force by Banco Economico in Brazil and Banco Latino in the authorities to pursue continuous offsetting (and Venezuela, give force to Walters's insistence on the potentially destabilizing) intervention in asset mar- importance of fraud. kets. Since the U.S. savings and loan debacle of the Walters concluded by stating that it is difficult to 1980s, the debate on narrow banking has shifted to a avoid deposit insurance either formally or infor- different concern: how to design an institutional mally. Still, he said, the best solution is to leave insur- framework that allows banks to develop in a com- ance to the private sector in order to avoid moral petitive environment. hazard as much as possible. After reviewing the institutional expression of a hypothetical "pure" narrow banking system, the Narrow Banking authors tumed to an assessment of the recent evolu- tion of the Argentine banking system. Characteristics The conference continued with the presentation and of the Argentine regulatory framework included: discussion of the background paper written by * A minimum ratio of risk assets to capital of 11.5 Roque Femandez, then-president of the Central percent, following Basle conventions. Bank of Argentina, and Liliana Schumacher, then- * Freedom to engage in both banking and securi- principal economnist at that bank. The topic of the ties activities, subject to conventional concen- paper was "narrow banking"-the proposal that tration limits on banking activities, and demand deposits be backed entirely by safe, highly capital-associated restrictions on equity hold- liquid assets. ings in nonfinancial corporations (which can- Fernandez and Schumacher favor narrow bank- not be funded with deposits). ing for Argentina. They claim that Argentina was * High reserve requirements-initially 43 per- able to weather the "tequila effect" that followed cent on checking accounts and 3 percent for Mexico's 1994-95 crisis because of the high reserve term deposits, although after the 1995 panic requirements (43 percent) imposed on demand these were replaced by a unified liquidity deposits. This requirement implied a very low requirement of 15 percent. deposit multiplier and gave the Central Bank excel- * No publicly provided deposit insurance; lent leverage to deal with the crises by reducing instead deposit insurance is fully funded by the reserve requirements. The authors asserted that banks themselves for small depositors. In addi- because of the convertibility plan, which precludes tion, there is legal recognition of the seniority of the Central Bank from acting as a lender of last resort, small depositors' claims in a liquidation. 2 Suman K. Bery and Valeniano F. Garcia The regulatory framework for commercial banks in On the second point, the authors concluded that a Argentina is closely related to the country's monetary further shift in bank asset holdings away from loans constitution, as embodied in the 1991 Convertibility and toward highly marketable securities would, in Law. Fernandez and Schumacher described this principle, be desirable-provided well-developed framework as a regime based on a fixed exchange rate capital markets existed to fund riskier investments in with full convertibility of the domestic currency into the economy, as well as to provide banks with liq- U.S. dollars, and bimonetarism. Under this regime 100 uidity (and derivative products) in order to reduce percent of the monetary base must be backed by inter- the market risk associated with such securities. The national reserves, although as much as 33 percent of authors recognized that, as nonbanks become more these reserves can be Argentine government obliga- important in risk financing, implicit govemment tions (assessed at market values) denominated in for- guarantees might also come to be provided to them eign currency. The authors emphasized two based on the too-big-to-fail principal. They feel that implications of this monetary constitution:' these risks could be managed, however. * The (domestic) money stock is highly endoge- Fernandez and Schumacher concluded that narrow nously determined by the behavior of the non- banking is the best approach for Argentina. Under the bank public; any attempt to issue money in alternative, fractional reserve banking, even if the excess of what is voluntarily held would money supply is equally endogenous, exogenous threaten the pegged rate. changes in capital flows will be "multiplied" because * As a result banks and the public recognize that of the fractional reserve feature, making the monetary the Central Bank cannot act as a lender of last system less stable. Fernandez and Schumacher resort based on its monetary expansion powers. acknowledged that under a narrow banking system Fernandez and Schumacher then explored two risky assets would be transferred to other markets. further questions: Are the Argentine monetary They believed that such a shift would be advanta- arrangements well-suited to deal with shocks, par- geous, however, because the payments system would ticularly those arising from capital movements, and be shielded from those risks. At present Argentina has would a stronger move toward narrow banking be only 15 percent reserve requirements on all checking desirable? and deposits for less than ninety days. Thus the sys- On the first point, Femandez and Schumacher tem still has a long way to go before it can be consid- argued that the main problem posed by volatile cap- ered even near to narrow banking. ital flows is the relative illiquidity of bank loans as an Gerard Caprio Jr. criticized the Basle capital adequacy asset. In their view Argentina has solved this prob- rules and commended Fernandez and Schumacher for lem by maintaining a high ratio of capital to risk developing alternative models. In Caprio's opinion the assets. This approach ensures that temporary surges recommended Basle framework-a ratio of capital to in bank deposits do not immediately translate into risk-adjusted assets of 8 percent-is fundamentally hard-to-reverse loan commitments. As proof that this flawed, probably for industrial countries and certainly mechanism works in practice, the authors pointed for developing countries given the greater concentra- out that in the Argentine panic of 1995 only 13 per- tions in the asset portfolios of their banks. Caprio also cent of the reduction in deposits needed to be highlighted the role of terms of trade volatility in sys- matched by a reduction in bank credit-representing temic banking crises in developing countries. In his a reduction of only 2.3 percent of credit outstanding view an alternative regulatory model is needed to sat- prior to the crisis. isfy certain characteristics. First, there should not be In Fernandez and Schumacher's view the combi- inordinate reliance on a "world class" supervisory sys- nation of high (risk-adjusted) capital requirements, tem. Second, bankers should have better incentives to bank-funded deposit insurance, mandated liquidity engage in prudent risk taking. requirements, and the absence of a lender of last In assessing narrow banking, Caprio argued that resort function on the part of the Central Bank makes the first-round effect would be to bid down the yield for a monetary and banking regime with relatively on riskless assets to the point that many depositors few means by which private banks can transfer risk would migrate to nonbanks. Concurring with to their depositors or onto public finances. To further Walters, Caprio was skeptical that the implicit gov- protect public finances (and the interests of deposi- eminent guarantee of banks could be prevented from tors) given the experience gained during the 1995 cri- migrating to nonbanks. And if it did migrate, most of sis, the Central Bank was given additional powers to the benefits of narrow banking would be lost. sell or partly liquidate a bank in trouble in order to Whether the guarantee can be limited in practice lies protect the interests of small depositors. in the domain of political economy. 3 The Conference Proceedings Rather than pursue narrow banking, Caprio rec- insurance can be solved by requiring banks to hold omrmended increasing banks' required capitalization sufficient capital or debt that is not guaranteed by the and owners' liabilities and demanding better finan- government to absorb any losses they might incur. cial information and disclosure. These altematives to The central bank's role as lender of last resort also the current OECD banking model have the advan- could eliminate concerns about systemic risk. There tage over narrow banking of preserving le-nding and is little reason to fear a financial panic or systemic deposit-taking activities. Caprio also noted that high failure as long as the central bank does not permit the reserve requirements have a perverse consequence: money supply to decline precipitously. they amplify the impact of changes in world interest Fernando de Santiba les was critical of narrow bank- rates on domestic loan rates. ing as a desirable objective for the Argentine system. George Benston emphatically disagreed with Santibanes also said it was wrong to categorize Fernandez and Schumacher on the superiority of Argentina's system in 1994 as close to narrow bank- narrow banking over traditional commercial bank- ing; there is a fundamental difference, he said, ing. Fernandez and Schumacher, he said, greatly between fractional reserve banking (even with high understated the benefits of traditional commercial reserve requirements) and narrow banking. The banking, and the putative solutions offered by nar- main feature of narrow banking is that it separates row banking are either illusory or can be achieved by lending from depository functions. Santibanies fur- other means. Benston challenged the authors' evi- ther asserted that, far from being a source of vulner- dence that the economies of scope in banking are low. ability, it was because Argentina had fractional Customers do benefit from using a single account for reserve banking that the authorities were able to help all their business (payments, investments, loans), banks cope with the 1995 panic. which reduces transactions costs and the costs of Santibafies said that even though narrow banking dealing with variability and uncertainty in cash might make the payments mechanism more stable, flows. In Benston's view the fact that no bank oper- under no circumstances would it increase the stabil- ating in the United States or elsewhere offered only ity of overall credit provision during a crisis. That depository services-despite being legally free to do would be true regardless of whether nonbanks were so-was a powerful indication that there are sub- funded by issues of debt liabilities or by issues of stantial efficiency gains from the joint offering of shares. Santibafies also claimed that nonbanks deposit and lending products. funded with liabilities would be subject to the same Benston was also dubious of the potential gains contagion effect as conventional banks. In his opin- from shuffling risky assets to nonbanks. Like Alan ion nonbanks could also incur maturity mismatches Walters, he was skeptical of the claim that, in the case that would lead to liquidity crises and to contagion. of failure, nondepository institutions would remain Santibanes alleged that the transition from con- free of state intervention. Benston cited the case of ventional to narrow banking would force a decline in Australia, where the authorities bailed out the credit because most depositors would be unwilling "depositors" in a large real estate investment com- to invest in volatile capital market products. pany. The state, he claimed, would not remain indif- Uninformed small depositors would have fewer ferent when facing the possibility of a stock market options under narrow banking because it is easier for crash. them to assess banks than nonbanks. Santibanies Regarding money supply control, Benston said shared Caprio's concern on the paucity of safe instru- that "broad" banking does not cause problems ments in the domestic markets of developing coun- because the central bank has the tools to control tries. The likely outcome, he said, was that narrow money through open market operatiorns. In fact, he banks would have to start investing abroad. Since it alleged, narrow banks would make it more difficult is unlikely that these funds would return to the coun- to control the money supply. Nonbank intermedi- try through international banks, there would be less aries in a narrow banking context would be tempted funding for domestic credit. Santibafies further to offer transactions balances (perhaps using elec- asserted that narrow banks would probably have to tronic transfers or check-like instruments) at a lower charge a fee for handling demand deposits, limiting cost and consequently would be managing some the benefits of extending the payments mechanism portion of the payments mechanism beyond the to a larger share of the population. He shared direct control of the central bank. Benston and Caprio's view that the joint provision of Narrow banking would prevent bank runs, deposit and credit services by banks is efficient, and acknowledged Benston, but so does deposit insur- that separating the services among specialized insti- ance. And the moral hazard problem of deposit tutions would be welfare reducing. 4 Suman K. Bery and Valeniano F. Garcia Deposit Insurance authorities pursue an exchange rate rule. In that case when a panic hits, as in Argentina in 1995, the central In his paper on deposit insurance Larry Sjaastad bank has only one instrument (international drew a direct link to the previous topic of narrow reserves) to serve two targets: maintaining the banking, defending the narrow banking idea and exchange rate rule and saving the commercial banks. saying that deposit insurance is needed only because Sjaastad offered several important clarifications narrow banking does not exist. Fractional reserve on deposit insurance. He pointed out that the term is banking, he asserted, is an irrational system that a misnomer. By pooling risks, conventional insur- came into being by historical accident. As long as ance (fire, auto, and so on) aims to protect the net fractional reserve banking persists, so will the risk of worth of an individual policyholder against specific bank runs, and with it, the need for a facility to risks but does not attempt, through the act of insur- reduce that risk. According to Sjaastad, a well- ance, to diminish the designated risk. By contrast, designed 100 percent reserve system would elimi- deposit insurance, although nominally an instru- nate the need for both deposit insurance and the ment for dealing with bank insolvency, is used to lender of last resort facility. reduce the risk of bank runs (that is, bank illiquidity) Fractional reserve requirements, said Sjaastad, by reassuring depositors that their interests will be introduce an externality into the banking system. In taken care of regardless of when they present their same way that a car entering a highway produces a withdrawal claims. As such, deposit insurance off- marginal delay for all other cars on the highway, the sets the externality associated with fractional reserve withdrawal of one depositor reduces liquidity for all banking referred to earlier. depositors, if only by a small amount. Even though But even though deposit insurance may reduce deposit insurance cannot increase the liquidity of the the probability of bank illiquidity, Sjaastad argued system-that requires a lender of last resort-it reas- that it does not in fact protect depositors from bank sures depositors that their money is safe even if they insolvency. Instead, it simply redistributes losses to are last in line to withdraw. other parties. As he put it, "from a social point of When banks are forbidden from paying interest view [such losses are] real and inevitable ... in this on demand deposits, fractional reserve requirements context deposit insurance merely redistributes losses force them to invest in nonprice competition. This sit- to other parties." Sjaastad stressed the potential for uation, Sjaastad contends, is particularly serious in deposit insurance to increase the probability of insol- an inflationary environment. In a high-inflation envi- vency through moral hazard, encouraging excessive ronment the banking system expands over the social risk taking by bank managers and reduced monitor- optimum because of the rents generated by infla- ing by bank depositors. Sjaastad further noted that tionary revenues (the inflation tax). These often take for deposit insurance to be effective at preventing the form of services (such as excessive branch bank- panics, it must so totally protect depositors from the ing, as seen in Brazil) that cost banks much more to risk of loss that they have no incentive to monitor the provide than their value to customers, at least at the portfolio or even the solvency of the deposit-taking margin. institution. Thus deposit insurance creates an indis- If required reserves on demand deposits were 100 pensable role for supervision and regulation, how- percent and if the central bank paid no interest on ever intrusive and unsuccessful they often may be. those reserves, the issue of prohibiting interest on If deposit insurance is an unavoidable require- demand deposits would never arise. In that case ment of fractional reserve banking, what are the least banks would have no choice but to charge for their damaging ways in which it might be provided? services, and the inefficiency argument would be Sjaastad suggested that the moral hazard conse- irrelevant. quences (particularly to the public treasury) of According to Sjaastad, even though deposit insur- deposit insurance are reduced if deposit insurance is ance and the lender of last resort facility complement private, risk adjusted, limited by a cap, and paid for one another, they are conceptually quite different. by the insured. Sjaastad also recommended substan- Deposit insurance is used to protect depositors tially higher capital ratios and reserve requirements against bank insolvency, while the lender of last along with multiple or even unlimited liability on the resort is a response to illiquidity problems. In prac- part of shareholders in deposit institutions. This tice, Sjaastad acknowledged, it may be difficult to move would shift monitoring from depositors (in the distinguish between the two. He also noted that the absence of deposit insurance) or regulators (where effectiveness of the lender of last resort may be there is deposit insurance) to shareholders. In this reduced or disappear altogether when the monetary respect Sjaastad was in complete agreement with a 5 The Conference Proceedings shift to self-discipline by imposing strict account- should be public or private, Horvitz said that work ability on directors and shareholders. Private insur- by Bert Ely and Panos Konstas had convinced him ance companies would be important in setting up that private insurance was feasible. Nevertheless, he eligibility requirements for depositors wanting to was not persuaded that private insurance had buy insurance. The banks where they elect to have important advantages over public insurance. their deposits insured would have to meet strict cri- Horvitz agreed with Sjaastad (and with Caprio) teria regarding capital, rating, and general risk that for publicly provided deposit insurance to work management. better, capital requirements would have to be sub- Paul Horvitz agreed with John Kareken's argu- stantially higher than the levels agreed under the ment against narrow banking, that if there were a Basle conventions. In this regard it would be crucial demand for a safe bank, the market would have pro- to use appropriate monitoring and market value vided it. Like Santibanies, he was also concerned accounting rather than historical cost accounting. about the trauma and costs implicit in the transition Good monitoring is also essential. With good moni- from broad to narrow banking. toring and effective early intervention the insurer is Horvitz agreed with Sjaastad that deposit insur- adequately protected and risk-adjusted premiums ance is not simply insurance; that its appeal is that it are not essential. For such a strategy to work it is can actually reduce the risk being insured against. He essential that regulators be able to take over a bank cited several legitimate public policy justifications well before its net worth is negative. Horvitz would for government provision of deposit insurance, prefer a higher obligatory intervention ratio than the including insurance's ability to avoid the macroeco- 2 percent of risk-adjusted capital mandated by the nomic policies needed to prevent a decline in the U.S. Federal Deposit Insurance Corporation, and money supply in the face of a banking panic and the referred approvingly to the remedies available to positive externalities associated with the spread of New Zealand authorities once capital drops below 8 the check payments system, which requires full con- percent. Insolvent banks have every incentive to take fidence in the safety of deposits. great risks, and in this case moral hazard could Horvitz agreed with Sjaastad that the moral haz- wreak havoc on the banking system. Horvitz dis- ard generated by deposit insurance increases the agreed, however, with Sjaastad's claim that liquida- potential cost of bank failure and that deposit insur- tion is almost always preferable to merger, though he ance's appeal-preventing failures by its mere exis- agreed that stockholders of a failed bank should be tence-might indeed be deceptive. He also agreed wiped out. Horvitz also agreed with Sjaastad that that it is difficult to limit the government's exposure there should be double liability for stockholders. when large banks are involved, but he thought that Horvitz favored rules over discretion when the benefits of deposit insurance-namely, reducing enforcing regulations. The FDICIA took the right the instability of the financial system-were too approach, he said, in relying on rules rather than on valuable to lose. the discretion of regulatory authorities. Under the act Horvitz said that recent changes in U.S. banking prompt action is required when net worth drops to 2 laws had incorporated lessons from the savings and percent of assets. Horvitz concluded that with appro- loan debacle of the 1980s. Structured early interven- priate safeguards, government deposit insurance tion (that is, mandatory intervention well before can be the best solution to dealing with the inherent shareholders' equity has been reduced to zero), as instability of fractional reserve banking. established in the Federal Deposit Insurance Roberto Junguito emphasized the need for coordi- Corporation Improvement Act (FDICIA), is a way to nation when central banks are independent of gov- limit the exposure created by publicly provided ernments, bank supervisors and regulators are deposit insurance. Horvitz also claimed that the independent of central banks (the norm in Latin recent legislated preference for depositors over other America), and deposit insurance is managed outside bank creditors also could reduce moral hazard, since the central bank. Coordination between these entities these other (presumably more sophisticated) credi- is important because it can sometimes be very diffi- tors now have more incentive to monitor banks. cult to distinguish between illiquidity and insol- Horvitz agreed that narrow banking would elim- vency. In Colombia, for example, the Central Bank is inate the need for deposit insurance but, echoing responsible for solving illiquidity problems (through Benston and Santibafies, said that no country with a a lender of last resort facility), while the government reasonably well-developed banking sector would oversees deposit insurance. This division of respon- choose to endure the painful transition to narrow sibilities has created conflict between the (indepen- banking. In terms of whether deposit insurance dent) Central Bank and the government, with each 6 Suman K. Bery and Valeriano F. Garcia trying to decide whether a bank failure (or more the system does not eliminate the possibility, under commonly, bank failures) should be handled by a fixed exchange rate regime, of a generalized run monetary or fiscal policy. frorn domestic to foreign currency, as happened in Regarding the too big to fail principle and its the Mexican crisis. Second, narrow banking does not implicit moral hazard, Junguito believed that during prevent gross claims from exceeding reserves. Third, the debt crisis many bailouts were the result of pres- small countries might have to choose between insuf- sures applied by foreign banks. He also noted that ficient diversification and foreign exchange risk. during periods of structural adjustment banks are in To protect the financial system, Meltzer proposed a particularly difficult position because of the open- a modem analog of Bagehot's (1873) proposal: the ing of the capital account and competition from for- central bank should announce the range of collateral eign banks. against which it will provide loans, at a penalty rate The potential for contagion and propagation of a higher than the market rate. At this rate borrowing financial crisis, Junguito affirmed, depends on a from the central bank is a right so long as borrowers country's macroeconomic framework. Moreover, he offer acceptable collateral. concluded, deposit insurance is not necessary when Meltzer emphasized that no system of regulation, macroeconomic fundamentals are in place. supervision, or market pricing can eliminate risk, Allan Meltzer thought it unlikely that narrow and that macroeconomic stability is a necessary con- banking would prevail, since depositors and banks dition for financial stability. He agreed with Sjaastad have a joint interest in leveraging a given stock of on the need to eliminate deposit insurance, with the reserves. He agreed with Sjaastad that eliminating caveat that there should be changes to get rid of time deposit insurance without institutional change inconsistency and to protect small depositors who would not be credible. Such a policy, he said, would cannot be expected to monitor banks. be seen as time inconsistent because it is tempting to In closing, Meltzer proposed a system that would spread costs over taxpayers rather than concentrate do away with deposit insurance in the United States: them on depositors. The pressure for bailouts is not lifting restrictions on the number and size of checks only domestic. In Uruguay and Chile in the early than can be written on money market funds that hold 1980s, Meltzer noted, foreign lenders pressured the only Treasury bills. In this way the United States government to underwrite bank losses-and the would have narrow banking as an option, with 100 government did. percent reserves and no deposit insurance. This sys- Thus the challenge is to improve existing systems tem would compete with banks that are now covered whose major defects are weak information and a by the FDICIA. The public could choose between powerful tendency toward regulatory forbearance. placing their demand deposits in money market Regulation, supervision, and examination are nei- funds with no insurance and 100 percent reserves or ther necessary nor sufficient to prevent banking pan- in conventional banks with no insurance to be regu- ics and losses. Supervision and regulation, usually lated by the FDICIA. This model, Meltzer claimed, enforced by audits and examinations, often fail to could be applied to most countries. identify problems in a timely fashion. And even if problems are identified, supervisors and regulators Market-Based Supervision and Regulation are likely to be pressured by politicians to engage in forbearance Although many countries have considerable experi- Meltzer claimed that better rules, coupled with ence with command regulation in the financial sec- market-based regulation, were the way to address tor, in many instances such regulation has failed to these regulatory issues. By imposing rules that prevent systemic crises. Thus there is growing inter- require banks to close before all capital has been lost est in market-imposed discipline (also cailled self-dis- (assuming that capital can be correctly accounted), cipline). Peter Nicholl, a former deputy governor of market-based regulation tries to maintain the bene- the Central Bank of New Zealand and currently an fits that come from widespread use of banking ser- executive director at the World Bank, prepared the vices while avoiding some of the social costs. Chile background paper on this issue. Nicholl was active and New Zealand, he said, have combined market in the discussions in New Zealand when the new pricing of risk with rules requiring public disclosure banking law was being designed. of prospective losses, and examiners' ratings are Nicholl said that policy interventions in the finan- published in the press. cial sector are justified only if they are likely to Meltzer highlighted three types of risk that would improve the soundness and efficiency of the banking remain even in a pure narrow banking system. First, system relative to a situation in which there is no pol- 7 The Conference Proceedings icy intrusion. Nicholl made clear, however, that in of the breach, and credit exposure is frozen if tier one New Zealand there is stil an important role for pub- capital falls below 3 percent. lic policy because of the following externalities: Nicholl explained that the system is in its infancy * the information problem depositors face in and has yet to be tested. Most academics, consumer assessing portfolio quality groups, anid politicians favor the new system. * the importance of protecting the payments Bankers'reactions have been somewhat mixed, with system some fearing that the new system will be even * the potential withdrawal of credit lines from all tougher than its predecessor. The two biggest criti- banks, not just those that have been mismanaged cisms of the new system, Nicholl said, are that direc- * the freezing of transaction balances tors may be unaware of the complexities involved in * the possibility of contagion. their bank, and that New Zealand is free riding on The current regulatory environment in New overseas supervisors because 90 percent of its banks Zealand is based on four general principles: some are foreign-owned. mandated, standard regulation; significant disclo- Nicholl agreed that there may be some validity to sure to the market of banks' financial position; the second point. In terms of the first issue, however, greater responsibility for directors and managers; he said that directors should not find themselves at a and automatic-rather than discretionary-inter- disadvantage in understanding the nature and state vention (also referred to as structured intervention) of their business. Under the new law directors have to breaches in capital requirements. In addition, all incentives to hire intemal supervisors, who will on-site inspection of banks has been eliminated. probably be much more helpful than the occasional The difference between New Zealand's current on-site supervisors of the past, who were not held market approach and that of most Latin American personally responsible for their eventual inaction. countries, said Nicholl, is a matter of degree. New Noting that moral hazard and asymmetric infor- Zealand has fewer mandated regulations, but sev- mation are more important in the financial sector than eral measures have been introduced to compensate in other businesses, Rolf Liders said that these pecu- for this, including greater disclosure and responsi- liarities have been the cause of many costly banking bility for shareholders and managers and automatic failures. According to Luders, macroeconomic weak- intervention. nesses were responsible for many financial crises in The new disclosure rules require banks to publish Latin America, including those in Argentina (1981), key information each quarter and to display a sum- Chile (1981-82), Colombia (1982), Uruguay (1982), mary of this information at all branches. Disclosure and Venezuela (1994). Luders claimed that these requirements are comprehensive and include crises have led to more regulation-despite the fact detailed information on exposures to related parties that bad regulation did not cause the crises. More and specific sectors, capital adequacy, asset quality recently, as deregulation has become popular, mar- and provisioning, interest rate risk, and so on. ket-based regulation has gained attention. Ratings are not mandatory, but banks with a credit According to Liiders, mandated regulation and rating applicable to long-term senior unsecured debt market regulation have similar aims: reducing sys- are required to disclose the rating prominently. temic risk and limiting bailout costs to government. Banks that do not have a rating are required to dis- Each approach has different costs, and consequently close that fact. different efficiency. Luders pointed out that the cost The new arrangements for director responsibilities of mandated regulation could be quite large, with mandate that false or misleading statements can lead some studies placing the cost to financial institutions to fines and ultimately to imprisonment. Moreover, if at 6-14 percent of noninterest operational costs. creditors are injured by false or misleading disclo- Luders alleged that government-funded deposit sure, directors can face unlimited personal liability. insurance exacerbates the moral hazard problem that Structured, automatic intervention is intended (as is endemic to the financial sector. This, in turn, in the United States) to reduce regulatory forbear- increases the complexity and cost of prudential reg- . ance. If a bank's tier one capital falls below 4 percent ulation. On the other hand deposit insurance gener- of risk-weighted exposure or total capital falls below ally has been effective at preventing isolated bank 8 percent, the bank is required to immediately sub- failures from becoming systemic. mit a restructuring plan. In addition, no profits can According to Luders, the high cost of prudential be distributed until capital requirements have been regulations has inspired the search for alternative restored, the bank's exposure to a related party can- mechanisms to achieve the same goals. Market- not increase from the level that prevailed at the time based regulation, an alternative that has the same 8 Suman K. Bery and Valeriano F. Garcia objectives as prudential regulation, is based on the should be sanctioned for improprieties. According to elimination of deposit insurance and the disclosure Bydalek, shareholders and managers are jointly of financial information that currently is only made responsible for distress at a bank. available to regulatory agencies. Unlike New Zealand's system, which has elimi- Luders claimed that market-based regulation nated on-site inspections, Brazil's regulatory envi- probably would cost less than standard prudential ronment is rigid and detailed. In addition to the regulation and, in principle, eliminate costly bailouts. Central Bank, Brazilian banks are subject to scrutiny It could also, in principle, eliminate the agency prob- from the Security Commission, the Insurance lem and-because of the elimination of deposit insur- Commission, the Pension Fund Commission, the tax ance-reduce moral hazard. It is still conceivable, authorities, and others. however, that a bank failure in a market-based system All banks in Brazil undergo semiannual full audits could "force" the government to provide an expen- that include auditor opinions, balance sheet data, an sive bailout despite the lack of deposit insurance. income statement, reconciliation of net worth, and a Luders explained that Chile's banking sector also source and application of funds statement, with has important elements of market-based regulation, notes and explanations. This information is then although they are mixed with a number of mandated published in local newspapers, typically within two regulations: minimum capital requirements (follow- to three months of the cut-off date. Unlike in New ing Basle guidelines), strict restrictions on entry, Zealand, there is no requirement to report asset con- restrictions on portfolio concentration, restrictions centrations, sector exposures, peak market risk expo- on foreign currency exposure, restrictions on asset sures during the period, or conflicts of interest by and liability term mismatches, and obligatory liq- board members. uidity reserves (linked to the size of the loan portfo- In Brazil, Bydalek said, directors do not attest to lio). Chile also has full deposit insurance on all sight the accuracy of disclosure or adequacy of risk man- deposits and coverage 'of up to $4,000 on time agement systems. He found worrisome the provision deposits; as a result Chile has maintained more man- for New Zealand that thrust the responsibility for dated regulation than has New Zealand. compliance and disclosure on bank directors and Public disclosure exists in both Chile and New managers. In a large, segmented institution, he said, Zealand but is more comprehensive in New Zealand. it is difficult or even impossible for all executives to In Chile banks must disclose their income and bal- be aware of the risks assumed throughout the bank. ance sheet information, the composition of the board Bydalek said that the main advantage of New of directors, and the quality of banking assets. In Zealand's system is the stream of quarterly data and addition, every month deposits are classified for risk performance interpretations. Early and full disclo- by independent risk classification firms. Bank direc- sure by Brazilian banks, as called for in New tors are responsible for providing correct informa- Zealand, would have attenuated Brazil's recent tion and cannot claim ignorance; penalties include banking problems. fines and possibly prison. Directors also can be held Jacques Trigo Loubiere affirmed that foreign banks personally responsible for depositor losses. play a much smaller role in Bolivia's financial sector Paul Bydalek noted the similarities and differences than in New Zealand's. In addition, Bolivia's between New Zealand's market-based regulatory national banking system has a much higher concen- system and that of Brazil. In New Zealand 90 percent tration of property confined to privileged economic of banking assets are held by foreign-owned banks. power centers. This situation, the superintendent Brazil retains 90 percent its banking assets in banks claimed, reflects distortions in financial market oper- owned by local investors. But despite obvious dif- ations and impedes adequate application of market- ferences in size and culture, Bydalek said, good busi- based regulation. ness principles transcend these distinctions. Trigo said that greater disclosure requirements, a In New Zealand shareholders do not share direct fundamental element of market-based regulation, responsibility for their banks; only managers and would pose a problem in countries like Bolivia, directors have that responsibility. In Brazil voting where illiteracy is high and many depositors would control over most banks resides in the hands of a few not be able to understand such information. families active in management. Thus shareholders Moreover, no institutions or public communication and managers are the same people, even for publicly channels exist to inform the public about the bank- listed banks with substantial minority participation. ing sector. In addition, willful or political misuse of Bydalek advised that responsibility for banking dis- information may distort public perceptions of finan- tress be extended to controlling shareholders, who cial services. 9 The Conference Proceedings According to Trigo, arm's length supervision can financial crises increases when government attempts, inhibit the detection of important deficiencies in the through direct ownership, to operate financial inter- banking system and so facilitate fraudulent behavior. mediaries. Thus Trigo encouraged the use of on-site inspection, Eisenbeis also asserted that regulatory forbear- which enables the detection of irregularities and ance and a lack of credible policies on the closure of their subsequent correction and ultimately has troubled institutions increase the potential for finan- improved the quality of disclosed information from cial crises. Moreover, the link between supervision financial entities and banks. Like other models of and intervention in individual institutions and the supervision, market-based regulation relies on exter- likelihood of financial crises should be emphasized. nal auditors. But in many countries, Trigo said, banks The goal of supervision and intervention should not have been closed or required government interven- be to prevent failures or reduce the attendant costs to tion despite having received a clean bill of health liability holders, he said, but rather to make failures from reputable auditing firms. independent events. Eisenbeis also said that a lack of According to Trigo, director and managerial transparency in the accounting of market values of responsibility is a crucial aspect in the supervisory on- and off-balance sheet activities can exacerbate process. Bolivia's Superintendency of Banks and the potential for financial crises, whether in private Financial Entities, for example, has prescribed sanc- or public entities. tions against administrative negligence following a Eisenbeis disagreed with the narrow banking pro- judicial process against directors and managers posal. In today's financial markets, he said, such pro- engaged in fraud. Like New Zealand, Bolivia has posals miss the point completely. There are clear established predetermined responses to infractions externalities to a well-functioning financial system: It or unfulfilled prudential requirements. But even enables the separation of real savings decisions from though the supervisory framework includes the real investment. It facilities intertemporal transfer of basic principles of the market-based philosophy (dis- consumption and investment. It enables all the types closure, director responsibility, predetermined of risk shifting that characterize modern financial responses), it also assigns an important role to on-site institutions and markets. Promotion of economic supervision-unlike in New Zealand, where on-site efficiency and specialization of labor are what dis- inspection does not exist. tinguish modem economies from barter economies, Trigo alleged that many countries are plagued by and achieving these benefits requires a well-func- problems that constrain the introduction of market- tioning financial system. Eisenbeis argued that based regulation and supervision: uninformed or threats to these positive externalities would trigger uneducated depositors, an absence of private insti- calls for governmental intervention during crises tutions capable of economic and financial analysis of even in the absence of direct involvement in the pay- the banking system, and a media that do not press ments system. Narrow banking solves no important bankers for adequate disclosure. In addition, most of problem in today's financial system, nor does it the judiciaries in Latin America and the Caribbean remove the possibility that taxpayers will be put on are too weak to impose fines or penalties on directors the spot should a crisis occur. In fact, Eisenbeis said, that provide false or misleading information. Hence, narrow banking may create an entire system of Trigo concluded, many such penalties would go implicit guarantees in which there is no monitoring unpunished. According to Trigo, the New Zealand to limit taxpayer exposure to risks. model can be applied with relative ease in stable and Valeriano Garcia asserted that macroeconomic fac- consolidated financial systems. Doing so would pre- tors are the main cause of financial distress and crises. sent greater challenges in systems undergoing In some cases bad macroeconomic policies are at the reform and adjustment, where stability is still frail root of the problem, but at other times going from bad and vulnerable to internal and external shocks. to good policies implies significant changes in relative prices and the disappearance of important banks' Roundtable Discussion on inflation-related profits-setting up the right environ- Systemic Banking Crises ment for crisis. Changes in relative prices-especially real exchange rates and real interest rates-can trans- Robert Eisenbeis said that financial crises are rooted in form sound bank portfolios into troubled, weak assets. monetary and fiscal policies that are not viewed as Both types of situations exist in Latin America, accord- credibly contributing to economic stability and low ing to Garcia. If a banking system has become distorted inflation. Moreover, credible monetary and fiscal because of inflation, it is crucial that it be adjusted to policies are harder to maintain and the potential for cope with a noninflationary environment. 10 Suman K. Bery and Valeriano F. Garcia Garcia affirmed that the banking sector could have tors are equally important and that market ineffi- features to make it more resilient to macroeconomic ciencies are responsible for many of the problems shocks. The three most important safeguards, he said, banking supervisors face. Moral hazard is also a big are regulation, supervision, and enforcement. In most problem for bank regulators. In many instances of Latin American countries regulation and supervision financial sector liberalization interest rates are freed, are relatively good. The problem, he asserted, is banks are allowed to freely allocate credit, and enforcement-particularly when it comes to enforc- reserve requirements are eliminated. Having made ing the rules that require intervention or liquidation. these moves, regulators assume that the market will Regulators try to hide the truth from the market in act wisely and allocate resources efficiently. But even order to prevent panics. But troubled banks require where markets are free, the conditions for efficiency early resolution, said Garcia. Early information made must be properly set. Ghigliazza also said that in available to the market could smooth out problems or countries where there is macroeconomic stability, prevent them from developing. economic growth, and no major uncertainty regard- Most banking crises, Garcia continued, involve a ing the exchange rate and other policies, financial run-out of the system (reduction in the real quantity of systems generally have developed and thrived- money) and a liquidity crisis (increase in the ratio of even when facing serious market imperfections. liquid to illiquid money). Narrow banking would pre- Allan Meltzer emphasized the need for rules in the vent liquidity crises and might prevent run-outs. In resolution of banking problems. He believed that early 1995 Argentina experienced both a liquidity cri- market enforcement of those rules is crucial so that sis and a run-out crisis, resulting in a 20 percent reduc- the opportunity for discretion is minimized. tion in the nominal supply of money. Argentina's The aim of regulation should not be to subsidize risk- system survived because it was buffered by reserve taking activities, Meltzer said, but it should also be requirements, and many large banks avoided great careful not to penalize them. Some of the suggestions suffering. made in the literature, such as unlimited liability for Deposit insurance, said Garcia, is a double-edged bank managers, would penalize risk excessively and sword. When countries have it, it is bad. When coun- hence could make the entire banking system totally tries do not have it, it is worse. For example, when risk averse. Such a move could end up moving risk- Argentina suffered a severe banking crisis in the taking activities to other parts of the system (such as early 1980s it had almost unrestricted deposit insur- informal markets). Thus there is a need to balance too ance. Deposit insurance was responsible for much of much risk and too little. More research is needed to the crisis because it had increased moral hazard, learn which risks are being removed and which ones which made banks behave irresponsibly. As a result are simply being shifted. deposit insurance was drastically reduced. According to Meltzer, many banking problems During the 1990s, with a currency board in place arise because banks and other financial institutions and a near-panic in the banking system, Argentina acquire assets that trade infrequently, if at all. It is the reinstated deposit insurance. Although the new bankers' function to make judgments about interest insurance incorporates features to reduce moral haz- rates, the strength of firms and sectors, and so on. ard, the problem remains because limited insurance Regulators cannot make these judgments. As a result cannot prevent a panic. Garcia agreed with Sjaastad the value of a portfolio tends to become politicized, and Fernandez and Schumacher that narrow bank- resulting in forbearance. ing is the solution. Narrow banking separates two Even though most of the participants at the con- products that are of a different nature-credit and ference had praised the Basle guidelines, in Meltzer's money. In doing so, it enhances the efficiency of cap- view they are seriously flawed. They do not take into ital markets, does away with deposit insurance and account modem portfolio theory. They assign risk to its associated moral hazard, improves monetary con- individual assets rather than to the portfolio. The risk trol in floating exchange rate regimes, dampens the structure of a given portfolio can change radically effect of exogenous capital flows in fixed exchange with the use of derivatives. Regulators can look at a rate regimes, prevents taxpayers' money from being bank's portfolio at one minute and apply the Basle used to bail out banks, and ensures more equitable criteria, but within twenty minutes the whole exer- outcomes, since the people who lose (or win) are the cise could be irrelevant. same people who took the risk. Meltzer insisted on the importance of avoiding sys- Sergio Ghigliazza disagreed with the emphasis on temic risk but not necessarily individual bank failure. macroeconomic factors as the main determinants of In this regard deposit insurance is not necessarily a banking crises. He alleged that microeconomic fac- bad thing, since it tends to make the financial system 11 The Conference Proceedings more readily acceptable and broader. In addition, stringent rules for evaluating assets, high capital Meltzer urged that banking systems have a lender of ratio requirements, early exit and intervention strate- last resort, contending that it is a public good that the gies for banks in deep distress, greater transparency central bank can supply at a low cost. Still, the lender to foster market discipline, direct intervention mech- of last resort should lend only at a penalty rate. The anisms (such as reserve requirements), and in some rediscount window should be open against a wide cases contingent and transparent government liabil- range of specific marketable assets, and the interest ity and personal accountability for directors and rate should always be higher than the market rate. managers. Given the need to maintain public confi- Moreover, the window should be open not just to dence-as well as the catastrophic effects banking banks but to nonfinancial enterprises as well. crises can have on an economy-Taboada proposed Jose Evenor Taboada asserted that there is a tight that central banks use the different mechanisms and relationship between banking system soundness and tools outlined at the conference. microeconomic policy, and that it is necessary to take Roberto Zahler based his comments on Chile's the condition of the banking system into account experience over the past twenty years. He said that when formulating economic policies, both as a key banking sector problems result from the asymmetry objective and as a constraint. of forces affecting the value of assets and the value of Although stabilization has been very successful in liabilities. The economic value of assets is essentially Nicaragua, he continued, there are severe difficulties variable and market-determined, while the eco- in the state-owned banking system. Until 1991 the nomic value of liabilities is rigid and fixed. banking system was entirely state-controlled, and When there are shocks that increase the difference under Nicaragua's old constitution all banking sys- between the value of assets and the value of liabili- tems and insurance were in government hands. Since ties, solvency problems and financial system failure then the new economic program has promoted the can result. To minimize the mismatch between assets liberalization of the financial sector. Interest rates and liabilities it is important to reduce the amplitude have been freed, legal reserve requirements have of economic cycles. In Zahler's opinion, fine-tuning been substantially reduced, and directed credit poli- monetary policy is the right approach. An example is cies have been eliminated. In addition, a to try to reduce the amplitude of economic cycles that Superintendency of Banks was created and private affect the values of assets and liabilities. banks were authorized. New banking and financial Zahler's second suggestion was that, given that regulations are in keeping with a modem, market- economic policy also has a major influence on key oriented financial system. prices in the economy-including interest rates, How did the financial sector respond to these new exchange rates, wages, and asset prices-whenever policies? According to Taboada, the reaction has been the trend in these prices is divorced from their fun- asymmetrical. Private banks have emerged as damentals, there should be a macroeconomic policy healthy structures within the free financial frame- response. In Zahler's opinion there is a need for pol- work. State-owned banks, on the other hand, have icy flexibility rather than rules, so that shocks to the been experiencing financial deterioration because system can be distributed among different markets, they are unable to cope without state subsidies, easy rather than putting the bulk of the pressure in a par- central bank credit, and a noncompetitive environ- ticular market. Thus policymakers should take heed ment. Nicaragua will try to avoid using radical whenever wide asset movements develop. He "surgery" to cope with the problems of the state believed that the timing and sequencing of financial banks, preferring to dramatically reduce their size reform are irrelevant, and supported his contention and transfer their nonperforming assets to a collec- with three examples from Chile. tion agency. Zahler recounted a conversation he had had with Taboada said that in the future state-owned banks a prominent economist working on financial liberal- as well as private banks will depend exclusively on ization in Chile. At that time real interest rates on their ability to operate effectively and to attract deposits were, for the second year in a row, on the deposits. Only time will tell whether Nicaragua's order of 45-50 percent. The economist had argued financial system will be able to survive the asymmetry that because the rates were "market determined," between state and private banking, or whether state they should be left alone. Zahler disagreed. Just banks are incompatible with free-market conditions. because the rates were market determined did not Taboada recommended a conservative approach mean they were right. It is always possible that the to banking: independent and effective supervision, market is not working properly; the problem could internal supervision, high capital requirements, lie with regulations, bank owners or directors, or 12 Suman K. Bery and Valeriano F. Garcia somewhere else in the system. Whatever the cause, it lack adequate legal and economic resources, it is must be corrected or the financial system is flawed. somewhat difficult to control the activities of off- Zahler mentioned a second example from the shore banks. period when Chile fixed its currency to the U.S. dol- On the issue of narrow banking, Bola-os said that lar. During that time domestic inflation was 35-40 Costa Rica is moving in the opposite direction. The percent a year and international inflation was 4-5 Central Bank has lowered reserve requirements to percent a year. The exchange rate policy and wage reduce costs and intermediation margins. It has been policy were clearly incompatible, yet many experts preoccupied with dealing with the problems created believed that Chile could quickly and smoothly by new financial instruments that are close substi- bring its domestic inflation in line with international tutes for sight deposits and with the dilemmas cre- inflation. There was no fiscal deficit and monetary ated by fiscal and electoral cycles, which tend to policy was working well, but domestic spending was increase the fiscal deficit during election periods. very high, causing a chronic current account deficit Costa Rica is considering a constitutional amend- that reached 14 percent of GDP in 1981. But since ment to limit the fiscal deficit (including the Central there was no fiscal deficit, many observers believed Bank deficit) to no more than 1 percent of GDP. that there was no problem. Soon thereafter, the entire In addition, the Central Bank is considering intro- system collapsed. ducing deposit insurance as a way to induce depos- Zahler concluded with a review of the challenges itors to shift toward domestic banks and away from facing Chile's efforts at consolidated supervision. parallel or offshore banks, though Bola-os was not Although Chile lacks consolidated supervision, this convinced that this was the best approach. In terms has not posed a major problem because most banking of banking supervision, he said that the state banks activities are domestic. A law being considered, how- needed the most help. Market-based supervision is ever, would allow domestic banks to invest in foreign an interesting way of improving the overall situa- banks. Zahler urged Chile to assess how foreign banks tion, Bolafios said, because more information should carry out supervision and suggested that domestic be available to the market, particularly relative to off- banks coordinate with their foreign counterparts. shore banking. Rodrigo Bolaflos Zamora began by identifying sev- In conclusion, Bolainos assured the audience that eral important weaknesses of Costa Rica's state Costa Rica is working to make its banking supervision banks: bad portfolios, insufficient capital, and comparable to international standards within the next spreads of up to 16 percent between average loan and decade. Moreover, he contended, this objective is deposit rates. Although the large share of deposits shared by the rest of the Central American countries these banks hold is declining, they are still highly as they try to reduce banking fraud, improve supervi- protected. Costa Rica, however, is starting to elimi- sion, and solve the problem of state banks. nate some of these state warranties (such as unre- stricted deposit insurance on all types of deposits). Note Until recently state banks held a monopoly on sight deposits and had exclusive access to the Central 1. This institutional arrangement, known as a currency Bank's discount window. These privileges ultimately board, is quite different from that of a standard pegged rate will be taken away. Moreover, over the past eight to (as in Mexico until December 1994) in the sense that the ten years Costa Rica has been strengthening its central bank has little leverage to exercise monetary policy. prudential supervision. One of the biggest problems, In particular, it cannot "sterilize" capital flows, as is done according to Bolafios, is offshore banks. Because the by other countries (for example, Brazil) with fixed capital account is relatively open and the authorities (pegged) rates. 13 Opening Remarks Sri-Ram Aiyer B anking crises have become common in many magnitude of the problem, most recently in the case parts of the world. They are costly. They are of a bank in Brazil, for example, it is shocking. Hence a threat to monetary and fiscal discipline. the need for prevention. They surely affect financial savings. More impor- Another reason for analyzing Latin American tant, they hurt economic growth through a slump countries is the continent's considerable experimen- in confidence, as we have seen in Mexico and tation in dealing with banking systems, sometimes Venezuela. Sometimes they even affect political with more boldness than in industrial countries. stability. Chile's experiment with market-based regulation is The origins of banking crises may be exogenous, one example. as when there are large changes in relative prices Prevention is always cheaper than a cure-and caused by factors such as devaluation or a lack of fis- surely it must be so for banking distress. In consid- cal and monetary discipline. Often they are also ering any set of measures to prevent crises, many endogenous: microeconomic weaknesses, inade- issues need to be considered. What should be the quate regulation, poor supervision, lack of enforce- guiding principles in establishing reserve require- ment, and fraud. The problem with traditional ments? How best to cope with the moral hazard supervision is that these failings invariably are dis- problem that deposit insurance introduces? Does covered only after the event. Thus the costs of resolv- market-based regulation substitute for or compil- ing these problems are usually very high. ment supervision, and in what way? Even in coun- This conference focuses on Latin America and the tries that have market-based regulation, one could Caribbean because many of the region's countries argue that it has not really been tested, and so what have had continuing problems with banking dis- is the proper combination? By asking the experts and tress-sometimes systemic, sometimes individual- analysts attending this conference to address these that often threaten the entire banking system. Almost issues, we hope to answer some of the pressing ques- every month of the past year has seen headlines tions and to derive a framework for preventing bank- about a new problem. And when one considers the ing crises. Sri-Ram Aiyer is director of the Latin America and the Caribbean Technical Department at the World Bank. 15 Banking Crises-A Prologue Alan Walters In the years since World War II industrial coun- Indeed, complete credibility is a theoretical concept tries have been notable for their absence of any and has little to do with the realities of Latin financial panic and collapse-certainly nothing American politics. like the massive bank failures that occurred in the A fixed exchange rate does not mean that a coun- United States and Europe in 1930-31. Nor have they try is immune to credit and banking crises. With free experienced any outbreak of hyperinflation. During trade and capital movements a fixed exchange rate the 1970s there was sufficient inflation to undermine guarantees only that tradable goods will have the the U.S. savings and loan institutions, with their bor- same prices and rates of inflation. It does not guar- rowing at high short and lending at fixed long. antee prudent credit policies by the banking and Although this situation was bad, it was manageable financial sectors. A currency board arrangement and did not cause a crash. Not so in the developing ensures only that notes are exchanged at a specified world, and certainly not in Latin America. parity and that there is no monetary financing of The pathology of Latin American crises and pan- deficits. It does not guarantee that speculative and ics is well known. Politicians' attempts to gain excessive credit pyramids will not be built on this power by looting the treasury and the fleeing tax base of high-powered money. But more important base give rise to large public sector deficits that are in this context is that, as seen most recently in financed by printing currency and by borrowing Argentina, a banking crisis always threatens the from domestic banks. Once expectations of high peg, reducing credibility and causing high real inflation are ingrained, it is difficult and costly to interest rates that induce slumps and erode political change them. Reform has had to focus on estab- credibility. A credit crisis and currency crisis feed on lishing credibility for low-inflation regimes. one another, threatening the payments system and Economies in transition from hyper or high infla- the currency fix. It takes courage and ingenuity to tion to financially stable regimes are particularly manage such a crisis. Argentina has held-so far- subject to crises. Governments and central banks and is likely to continue with the fix. In 1979-82 promise a lot, and it is natural that their credibility Chile succumbed and the peso went into a dirty should be questioned and discounted when the float, eventually to be rescued as a sort of target real pressures are on. exchange rate. Monetary and Exchange Rate Regimes The Chicago Plan A number of reforming authorities in Latin America This combination of credit crisis, bank runs, and col- have decided that the way to achieve credibility was lapse of credibility puts an economy on the rack. What to fix or peg their currencies to the U.S. dollar. An can be done to safeguard against such a devilish absolute fix implied a sort of monetary union. But a alliance? The natural response is to try to separate the true union, with the same nominal interest rates, credit process from the payments deposit system-the required complete credibility, which certainly was 100 percent solution or Chicago Plan put forth by not present in most Latin American economies. Henry Simon. Abolish fractional reserve banking. Alan Walters is vice chairman and director of the AIG Trading Group. 17 Banking Crises-A Prologue Institute narrow banking. Indeed, during the recent state or local government. Thus the state controls entry crisis in Argentina it was suggested that the problems into the banking system and can dictate its regulatory would not have arisen if the Central Bank had held framework. But it would be foolish to pretend that the reserves against all of its deposit obligations. But such government has unlimited license. If controls are too narrow banking solutions would necessarily drive a onerous, another country can supply financial services lot of loan business out of the banking system and into that a domestic government has made too expensive- the arms of equity markets, debenture markets, and thus Hong Kong is essentially the banking system for other credit-creating institutions. The credit risks southern China. It is, however, in everyone's interest to would be shuffled out of the banking system and so have a sound and efficient payments system. insulate the banking system and in particular the pay- The prime job of banking supervision and regula- ments system from the vagaries of the credit market. tion is to ensure that the payments system functions A 100 percent solution would give rise to an ultrasafe smoothly. This outcome is possible only if there is con- but much smaller banking sector. Would the underly- fidence in banks and their institutional structure. The ing risks, shuffled out of the banks to equity holders, golden rule is, when depositors are convinced that be better borne by the equity markets than by banks? they can get their money out they will leave it in; only Could the government stand by while stock markets if they fear that they will not be able to get it out will took a beating that had been hitherto contained in the they want it out. banking system? Many equity markets are far from In fractional reserve banking the central role of transparent, subject to manipulation and insider trad- supervisors is to see that the bank maintains adequate ing, and rarely carry substantial liquidity. For these capital reserves to meet its normal redemptions. The reasons it is unlikely that there will be any reform that multitiered capital requirements of the Basle agree- follows Simon's path. Among imperfect options, the ment are an important aspect of supervision. The gov- fractional reserve system is the best choice. einient or central bank usually comes to the rescue of any bank on which a run, threatening the payments Credibility and Banking system, has developed. This is generally true even if there is suspicion of fraud or criminality; the sanctity The main trouble with banks is that they borrow short of the payments system overrides such reservations. and lend long. Bank liabilities are predominantly Fraud must be dealt with later. demand deposits that are redeemable on demand, whereas bank assets are typically illiquid. If all depos- Deposit Insurance and Too Big to Fail itors demand redemption, then the system cannot pay-and the stability of any banking system depends The effects of deposit insurance are still debatable. critically on the confidence of depositors. When con- Friedman and Schwartz have credited deposit fidence sinks, there are likely to be runs on banks. And insurance with preventing runs and stabilizing the unless they are stenmmed by action by the central bank U.S. banking system. But there has been no rash of or another monetary authority, runs can lead to pan- runs in countries that lack such formal insurance. In ics and a breakdown of the payments system. This any case deposit insurance is a double-edged threat is obviously an externality of great importance, sword. With the U.S. savings and loan institutions Thus we must be wary of systemic effects that threaten it gave rise to extremely speculative investments the usually smooth process of clearing accounts. because owners were secure in the knowledge that if the investments fell flat the insurance would meet Supervision and Regulation the deposit liabilities, and if the speculation was successful the owners could pocket the winnings. Banks have a natural incentive to avoid runs and In any situation involving a government guaran- crises. But they are also driven by the need to make tee-explicit or implicit-regulators must closely profits or at least to cover costs. And the cost irnpli- supervise the assets of banks. The value of many cations of crises-and in particular, of breakdowns in assets is often extraordinarily difficult to assess. Many the payments mechanism-far exceed the private have no ready market, and regulators tend to value at costs of the banks. A break in the payments chain cost. In the United States in 1931 this weakness had the may spread like a plague. For these reasons alone it paradoxical effect of valuing loans to farmers secured is worth seeing if regulation can help avoid such con- on their property at cost whereas government bonds tingencies. were valued at their very depressed market value. Bank supervision is usually done through the Similar problems are occurring now with respect to awarding of a charter or similar authorization by a the collapse of property prices from their peaks in 18 Alan Walters 1990. The loans that Chilean banks made to grupos in fraud. Bank operations in which large flows of 1978-80 are another example. money are speeding around can make fraud both In the past most bank assets have been loans to inviting to fraudsters and challenging for supervi- industry rather than marketable assets. This pat- sors. I suspect that only a small fraction of frauds are tern is gradually changing with the onset of securi- actually prosecuted. Political coverups and the costs tization. Although not all securities have suitably of bringing cases to trial are powerful deterrents. The deep secondary markets, securitization increases various solutions (closer supervision of day-to-day the liquidity of assets and provides banks with a transactions, requirements to report open positions, basis for valuation. To the extent that these securi- and so on) are not very attractive in a liberal society. ties are sold outside the banking sector, the risks are spread. Securitization, however, invites off- Deregulate and Be Damned? balance sheet passthrough and all the problems that the valuation of partial liability holds. Regulation is often rightly pilloried for being the cause of bank crises, even panics, and there is no Nationalized Banks and Intervention doubt that it has much to answer for. In particular, regulated interest rates-as in the infamous regula- One might think that government ownership, or per- tion Q-have caused untold damage to capital mar- haps management, of banks would ensure effective kets and shifted many offshore. Specification and supervision. But this is not so. As the provincial limitations on the loan book, often paraded as a pru- banks in Argentina showed, government ownership dential regulation, have caused many banks to fail. often creates pressures to make loans to politically The geographical restrictions in the United States are important people or other entities, almost regardless a common source of weakness. of their viability. Even in France the great national- Why can we not set the banks free and ensure that ized banks, such as Credit Lyonnais, have been sub- if they make a mistake they will pay for it? At first ject to political malleability and disastrous this seems a rash recommendation. But protagonists management. point out that when Scottish banking was free (until It is important to distinguish between the perma- 1844) and there was no restriction on the issuance of nent politicization of state-owned banks and the currency, competitive banking appeared to work transitory intervention by authorities when a bank well and, according to scholars such as White and needs winding up, takeover by a stronger bank, or Selgin, there was no evidence of a crisis-let alone a restructuring and capitalization. Restructuring and panic. capitalization are forms of bank doctoring. The U.S. Of course, that was before the development of Resolution Trust Corporation has been a remarkably checking and the system of transfers that we have successful example of such doctoring. At the cost of today. But there are other examples of deregulated hundreds of millions of dollars, it averted a panic, banking systems. The best known is London's disposed of or wrote off dubious assets, and capital- Eurocurrency market. Since its inception in the 1960s ized savings and loan banks in just two years. The and following the convertibility of the major curren- pall of uncertainty was soon dissipated and the sav- cies in 1973, the Euromarkets have been subject to ings and loan banks-at least those that were left- only the lightest regulation. There are no reserve could start functioning again. requirements and no restrictions on the types of busi- In the long run, after intervention and recapital- ness that can be pursued. Broadly speaking, the Bank ization, the general solution has been to privatize of England's attitude has been that the authorities do banks and, it is hoped, eliminate the politicization not mind what transpires so long as it is not in ster- that so distorts their behavior. This has been the ling. The Euromarkets have flourished, growing main solution in Argentina. But banks must be truly from $100 billion in 1969 to several trillion dollars privatized, unlike, for example, the French banks today. Yet there has not been a single case of sub- that were nominally privately owned but were stantial failure causing systemic problems. under close government control. Privatization must No doubt the Euromarket is special in many be substantive, not superficial. respects. Above all, it is wholesale, not retail, bank- ing. The Eurobanks are mostly branches or sub- Supervision and Fraud sidiaries of foreign banks; thus lender of last resort facilities go through the parent to the appropriate One argument for special supervision for banks is central bank. Many similar considerations show that that they are particularly susceptible to large-scale the Euromarket is a rather special variation of the 19 Banking Crises-A Prologue standard fractional reserve banking model. Still, the departure from the traditional systems of bank regu- success of the Eurobanks suggests that so-called lation, it uses techniques that are familiar to bank "command" regulation is not the way forward: the managers. regulatory system somehow must be developed to complement banks' natural proclivity to maximize Conclusion profits. Regulation should be light and, as far as pos- sible, flow from banks' normal risk management From all quarters there have been calls for closer operations. supervision of banking systems to prevent the sort of collapses recently seen in Latin America. U.S. deputy Incentive-Based Regulation secretary of the Treasury Lawrence Summers recently called for the International Monetary Fund One promising approach to regulation, known as (IMF) to adopt a "more assertive role in monitoring the "internal models" approach, has been explored banking systems to help avoid such destabilizing by the Basle committee and endorsed by the G-10 banking crises as those experienced in Latin America govemors. Each bank would use its internal risk over the last two years" (26 March 1996, Financial assessment arrangements to calculate the maxi- Times). I have my doubts about this proposal, but: mum loss it might sustain, with 99 percent confi- * There is a need for greater transparency- dence, over the next ten days. Regulators would banks should tell the truth and tell it quickly. I base their capital charge partly on this maximum have little sympathy for the need for commer- loss and partly on the bank's historical performance cial secrecy, so beloved because it is thought to in containing these risks. The basic argument for fool depositors. internal models is that banks know far more than * Some regulation is needed to ensure a mini- regulators about the risks they are running and the mum capital requirement. Mandatory reserves subtle distinctions between one asset and another. are more dubious and appear to be unneces- But, notwithstanding the advantages of this sary, as in the United Kingdom and approach, the command features of regulation are Switzerland. The best solution is to harness still present. bank's own incentives to control risk. A bank's An alternative is a "pre-commitment" system. As choice of a pre-commitment based on its assess- in the internal models approach, the bank uses it own ment of market risk, with suitable penalties if it models to assess risks in the ten days ahead. It cal- loses more than the sum pre-committed, seems culates its maximum loss, choosing its own confi- to be the best approach. dence level. Regulators fix the capital charge for * It is difficult to avoid deposit insurance, formal market risk (ignoring fraud and so on) at this maxi- or informal. The best solution is to leave the job mum loss. If the bank exceeds this maximum loss the to the private sector to avoid as much moral regulator exacts a suitably high penalty in the form hazard as possible. Thus the authorities would of a high fine or restrictions on trading and dividend require banks to insure (small) deposits but distributions, or both. they would not supply the cover. The pre-commitment approach is also analogous * Lender of last resort facilities can be associ- to the decisions banks make when weighing the pos- ated with currency boards and a temporary sibility of having to borrow at penal rates to meet liq- departure from convertibility provided there uidity requirements. Banks run their cash position is a widespread belief that the authorities will such that the expected cost of being caught without return to the same parity once the crisis ends. cash and having to borrow at penal rates is balanced * Finally, wise macroeconomic policies (moder- by the advantages of having marginal, nonliquid ate monetary expansion, manageable budget assets. and current deficits, low tax rates, and so on) In other words, the basic decisionmaking process will not prevent a credit and banking crisis, in the pre-commitment approach is well known in but they will make it less severe and more banking circles, and although it seems to be a radical manageable. 20 Does Argentina Provide a Case for Narrow Banking? Roque Femandez and Liliana Schumacher B roadly speaking, narrow banking requires that the federal government immediately take that demand deposits be backed entirely by over actual ownership and management of the safe short-term assets. On two occasions nar- Federal Reserve Banks; that the Federal row banking has been part of recommendations for Reserve Banks should guarantee the deposits radical changes in the regulation of U.S. banks-in of all Member Banks ... that the Federal the 1930s with the Chicago Plan for Monetary Reserve Banks be instructed ... to dispose of all Reform and in the 1980s as part of the debate that fol- assets of the Member Banks, to pay off deposit lowed the savings and loan crisis. liabilities and ... declare the corporations dis- Argentine banking has some features that raise solved; that banking legislation be enacted pro- the question of whether it is in a transition toward viding for incorporation of a new kind of narrow banking. Although it is far from having car- institution which a) alone shall be entitled to ried out such a transformation, is narrow banking a accept funds subject to check or to payment on better banking framework for Argentina, given the demand; b) which shall be required to maintain constraints imposed by the country's monetary pol- reserves of 100 percent in lawful money and/or icy on the Central Bank's traditional role as lender of deposits with the Reserve Bank; c) which shall last resort? serve exclusively as institutions for deposits and transfer of funds; ... that additional legis- Narrow Banking Proposals lation be enacted providing incorporation of a distinct class of institutions in the general form Proponents of narrow banking have several objec- of investment trusts which ... shall perform the tives: functions of existing banks with respect to sav- * To achieve full control over the monetary supply. ings deposits. (cited in Phillips 1995, pp. * To allow deposit insurance to prevent runs 193-98) without creating moral hazard and thus to allow financial institutions to break the wall After World War II financial problems seemed to between classic intermediation and securities be relics of the past. But Chicago economists never business. abandoned the "100 percent" bank reserve proposal. * To eliminate runs in a regime with no deposit Some were in favor of equity over bank debt as a pre- insurance, since liquid (narrow) banks are riot ferred financial arrangement. For example, Simons susceptible to runs. (1948) stated that: The original proposal for narrow banking in the United States was a response by members of the a second long-term issue concerns the future of University of Chicago faculty to the banking crisis of banks as sources of capital funds for private the early 1930s. In a memorandum sent to the secre- business. A 100 percent reserves requirement tary of agriculture on March 16, 1933, immediately would leave banks free to provide such funds after the Emergency Banking Act was passed, they out of their own capital ... If banks as lender- proposed: investors were dissociated from banks as Roque Fernandez, president of the Central Bank of Argentina when this paper was written, is minister of finance in Argentina. Liliana Schumacher is principal economist at the Central Bank of Argentina. 21 Does Argentina Provide a Casefor Narrow Banking? depositary-clearing agencies, the lender- With respect to the bank failures in the Great investor enterprises might then focus upon a Depression, Friedman and Schwartz (1963) noted: vital and essential function of providing long- term capital and, at best, of providing it in an As a fraction of total wealth the losses produced equity form. by bank failures were minor. .. . If the bank fail- ures deserve special attention, it is clearly The 100 percent bank reserve was seen as a system because they were the mechanism through that could restore to the monetary authority control which the drastic decline in the stock of money over the effective quantity of money while at the was produced and because the stock of money same time removing reasons for excessive govern- plays an important role in economic develop- ment intervention. Friedman (1959) also became a ments. The bank failures were important not strong supporter. His point of view was that "the primarily in their own right but because of their central problem is not to construct a highly sensitive indirect effect. instrument that can continuously offset instability introduced by other factors, but rather to prevent Friedman departed in one respect from the original monetary arrangements from themselves becoming Chicago plan: he proposed that interest be paid on a primary source of instability" (p. 23). He further the 100 percent reserves. explained that: During the 1980s narrow banking was brought again into policy debates after deregulation and com- our present fractional reserve banking system petition ended in the massive failure of savings and has two major defects. First, it involves exten- loan institutions and the reduction of banks' interest sive governmental intervention into lending margins (Phillips 1995; Bryant 1988). But the motiva- and investing activities that should preferably tions here were different. There were no monetary be left to the market. Second, decisions by hold- considerations; instead the debate focused on how to ers of money about the form in which they want design an institutional framework that would allow to hold money and by banks about the structure banks to develop in a competitive environment. In of their assets tend to affect the amount avail- particular, what were the legal conditions under able to be held. This has often been referred to which financial product diversification could best as the inherent instability of a fractional reserve proceed? system. (p. 66) One of the earliest proposals was made by Litan (1987). His approach called for the creation of When referring to government intervention, "'financial holding companies'(FHCs), which would Friedman had in mind deposit insurance, among be free to engage through separate subsidiaries in other arrangements. "Concem about [deposits] has any activity, financial or nonfinancial, subject to the led to an ever widening degree of control over the following restrictions: operations of commercial banks, of which the most * The 'banks' in FHCs would be required to oper- recent and extensive is perhaps associated with the ate as (insured) money market mutual funds, federal insurance of deposits" (p. 67). accepting deposits and investing only in highly Thus Friedman proposed a system of 100 percent liquid safe securities. reserves that would require banks to split into two * FHC could extend loans but only through sepa- separate institutions. rately incorporated lending subsidiaries wholly funded by uninsured liabilities ... and equity, One would be a pure depositary institution, a issued directly or by the parent FHC itself" literal warehouse for money. ... The other insti- (p. 165). tution that would be formned would be an This approach focuses on two problems: moral investment trust or brokerage firm. It would hazard and the optimal way to achieve financial acquire capital by selling shares or debentures diversification. Prevention of runs is not a concern and would use the capital to make loans or because Litan assumes at all times that there is acquire investments. Since it would have no deposit insurance. Thus his proposal is a way to power to create or destroy money, monetary reduce moral hazard behavior in (proposed) fully considerations would not demand any special diversified financial institutions, that is, those that control over these activities. Hence, it need be can take part in both lending and securities activi- subject to no more governmental supervision ties. Full control of money supply is not part of this than other financial institutions. (pp. 69-70) proposal either. In fact, the proposal's effect on the 22 Roque Fernandez and Liliana Schumacher money supply is the same as with a conventional Institutions that provide all types offinancial services bank because Litan's narrow banks are not (except those relating to the payments system) required to hold 100 percent of deposits as interest- bearing reserves with the central bank, but to These institutions basically seek funds for long-term invest in safe assets, such as treasury securities: projects through uninsured liabilities: commercial paper, bonds, or other types of securities. Divorcing deposit-taking from risk-bearing activities performed by banks ... would elim- Criticisms inate the concern that in a fully deregulated environment depositors' funds either would Two main criticisms have been leveled at narrow be used to bail out 'risky' nonbank activities banking proposals: or unfairly channeled to customers of non- * Because narrow banks are totally liquid, they bank affiliates ... The proposal would also destroy the optimal risk sharing that is address concerns that highly diversified achieved when banks provide contracts that financial organizations would accumulate or allow agents to insure liquidity risk (Wallace control excessive amounts of economic 1988,1996) resources. Fear of both conflict of interest and * Narrow banking is inefficient because it sepa- undue concentration primarily stems from rates bank deposit and loan services. Thus it the fact that deposit insurance allows banks to gives up economies of scope (Ely 1991). gather large pools of funds and thereby to With respect to the first criticism, the optimality of exercise significant control over the allocation illiquid banking has an important limitation since, in of credit. If banks in highly diversified orga- the models cited above, bank portfolios are illiquid nizations could not fund their loans with but riskless. With riskless portfolios, bad equilibria insured deposits, these fears would not be can be avoided with government deposit insurance guaranteed." (Litan 1987, p. 6) because insurance in such a setting cannot create moral hazard. Some supporters of narrow banking have also Another interesting point is that in these models focused on structured securitized credit as a new banks' raison d'etre is to insure agents against liq- technology that "can help make the transition to a uidity risk since private information about the true fundamentally different and better financial sys- liquidity needs of the agents makes the optimal tem" (Bryant 1988). Summing up, these are the insurance contract unavailable in the market. But main contents and foundations of narrow banking under the assumption that the proportion of agents proposals. with liquidity needs is known at the time contracts are written (even if information about individual Risk-free depository institutions and payments system agents' state is not public information), the optimal allocation could be achieved with a liquid or narrow Narrow banking allows banks to issue demand bank plus an insurance contract (see appendix). Such deposits (and eventually savings accounts or short- a contract would provide the insurer with access to a term deposits in small denominations) if their pro- kind of suspension mechanism that allows him to ceeds are invested in assets with low credit risk pay up to the known proportion of agents that will (mainly treasury instruments) and low interest rate tum out to be in need of liquidity. risk (short-term or floating-rate instruments). These With respect to the second criticism, it has been institutions can have insurance, although it is not argued that economies of scope would result from necessary for liquidity considerations since narrow the spreading of fixed costs over an expanded prod- banks are not susceptible to runs. Insurance removes uct mix. They could also result from cost comple- the remaining risk in banks' liabilities due to fraud, mentarities among product categories when account for example. Moral hazard behavior is precluded by and credit information (used to develop deposits the limitations on investment and by the separation products) is used to reduce the credit information of the two types of institutions. One consequence of and monitoring requirements for loan products for these features is that the interest paid on term the same customers (Ely 1991). deposits is low. But these "banks" require limited But empirical studies have found that large U.S. capital because risk is low, too. Thus retums on banks have not seen significant cost complementari- banks' equity would be enough to make this alter- ties between deposit and loan products, although native profitable. there may be significant benefits from sharing fixed 23 Does Argentina Provide a Case for Narrow Banking? costs-maybe 4-5 percent between these two classes play an active role as the lender of last resort and, in of products. But it is also argued that, when com- case of trouble, the Central Bank cannot bail out pared with the cost reduction from the virtual elimi- banks by printing money. nation of deposit insurance premiums that would be Given the constraints imposed by the monetary possible when insured deposits are backed with safe regime, the Central Bank established high reserve assets, the net cost of narrow banking is low (Pulley requirements to preserve a mass of liquidity that could and Humphrey 1993). be released in case of a systemic run. Since those reserves were a tax on the financial system, the optimal Argentina's Banking Regime structure was to levy a higher rate on the most inelas- tic deposits. Hence the rates were set at 43 percent for Argentina's banking regime already has some fea- checking accounts and 3 percent for term deposits. tures that make it close to narrow banking-in par- After the panic these rates were made uniform for all ticular, high reserve requirements and high capital liabilities-15 percent. Even so, this is still high and requirements. The association between high capital could provide a significant supply of liquidity to the requirements and narrow banking stems from the system if necessary. fact that capital-asset ratio regulations indirectly Procedures for closure and liquidation of institu- determine reserves. tions in trouble were established in 1991 by the Argentine Congress in an amendment to the Financial Mainfeatures Institutions Act, although some changes were intro- duced later. According to that amendment the Central Argentine banks are free to engage in both classic Bank could suspend a troubled bank for thirty days banking activities and securities activities but are lim- and request a capitalization proposal. After receiving ited in their equity holdings in nonfinancial corpora- the proposal the Central Bank could do one of two tions (up to 15 percent of bank capital, which cannot things: reopen the bank or remove its charter. If the be funded with deposits). There are also restrictions charter was removed, the bank became subject to the on lending to affiliates (up to 5 percent of bank capital laws goveming corporations and it was up to the to each affiliate with a maximum total of 20 percent for courts to deal with bankruptcy, to liquidate assets, and all affiliates) and client diversification (lending limits to pay depositors and other claimants. In the years up to 15 percent of bank capital per client without col- before the amendment there had been a combination lateral and 25 percent per client with collateral). of deposit insurance and Central Bank intervention Securities activities basically refer to trading, mutual that, in practice, insured all bank liabilities. This strat- fund management, and some (not too well-devel- egy was funded with money creation and made a sub- oped) underwriting activities. Except for these restric- stantial contribution to the hyperinflation of the late tions there are no limitations on the amount of risk that 1980s (Femandez 1990). As a result deposit insurance banks can undertake. There is, however, a required and bank bailouts with public resources were firmly capital-asset ratio of 11.5 percent, with assets adjusted rejected by both citizens and politicians. by risk based on the suggestions of the Basle When the Central Bank faced the consequences of Committee for Bank Supervision and Regulation. the 1991 amendment for the first time, after the Deposit insurance was eliminated in 1991. Mexican devaluation, it realized that an early court Argentina also has strong limitations on the intervention probably would not minimize the costs Central Bank's ability to act as a lender of last resort. involved in liquidation, in particular depositor To stop hyperinflation, in 1991 the government losses. Thus in April 1995 Congress passed another adopted a monetary regime based on a fixed amendment that extended to ninety days the period exchange rate with full convertibility of the domes- during which an institution could be suspended and tic currency into U.S. dollars and bimonetarism. augmented the Central Bank's ability to liquidate an Under this regime 100 percent of the monetary base institution without committing fiscal resources. The has to be backed by international reserves, 33 percent Central Bank can now arrange the sale of any bank of which can be Argentine public bonds denomi- in trouble. If the bank cannot be sold, the Central nated in foreign currency. Thus the money supply in Bank can split the banks' assets before court inter- Argentina is highly endogenously determined, and vention and sell them to pay depositors and other any attempt to issue currency above this amount senior debt (salaries, Central Bank loans). The would cause a loss of reserves and threaten the con- remaining assets and liabilities follow the usual vertibility of the domestic currency. This has two bankruptcy procedure. The Central Bank used this implications for the banks: the Central Bank cannot power five times during the recent crisis. 24 Roque Fernandez and Liliana Schumacher The rejection of deposit insurance and bank determination of the money multiplier can be bailouts with fiscal resources does not mean that worked out from monetary aggregates to monetary Argentine laws do not protect small, uninformed base to deal with the exogeneity of foreign investors' depositors. The 1991 amendment made deposits mood toward emerging economies. senior to all other bank liabilities (up to $3,000) and From the definitions of money M and monetary the 1995 amendment raised that limit (to $5,000) and base B in terms of currency C, deposits D, and reserve created deposit insurance fully funded by the banks requirements R, a linear system can be determined that covers deposits up to $10,000 or $20,000, with five unknowns in two equations: depending on maturity. Summing up, how did Argentina address the main issues regarding banking regulation? With full infor- (1) M = C + D mation and no deposit insurance, moral hazard should not have been an issue. Because information is asym- (2) B = C + R. metric, however, restrictions were established with respect to lending to affiliates. The April 1995 deposit The closed economy multiplier results from solving insurance law took moral hazard explicitly into con- the system under the assumption that B is exoge- sideration and established specific limitations in cov- nous and adding the equations C = cD and R = rD erage to avoid it. where c is a constant and r is a binding legal reserve With respect to diversification, banks were requirement. The solution is represented by M = allowed to carry out both classic banking and secu- mB, where rities activities. Limits were set on ownership of non- financial affiliates, however. (3) m = [(1 + c)/(c + r) ]. There is no explicit device to prevent panics, but Argentina's regime, with limited or no deposit insur- A tequila effect for an emerging open economy ance, is expected to create sufficient depositor disci- producing capital outflows implies that M is exoge- pline to set a price to banks' portfolio risk in normal nously determined by foreign investor confidence times and punish them selectively if there is a panic. and the multiplier determines the endogenous value High reserve requirements are also seen as a way to of the monetary base (B). reduce financial instability caused by capital out- To illustrate the financial vulnerability that arises flows (or highly volatile deposits in general). To from a tequila effect, subtract equation 2 from equa- make this point clear, the next section develops a sim- tion 1 to obtain pie framework. It also shows how Argentine regula- tions could produce results similar to a pure narrow (4) M - B = D - R = L, banking strategy. where L represents the loanable capacity generated Reserve requirements and capitalflows by deposits. If m>1, a change in foreign investors' mood producing capital outflows and reducing M The problem of capital flows reflects the closed implies a credit contraction. That is, the comparative economy conventional wisdom of the money mul- static of the system gives tiplier and banking multiplier. A fractional reserve banking system would never have enough liquid (5) dL/dM = (m - 1)/m > 0. reserves to deal with massive capital outflows (or bank runs). And a fixed exchange rate or convert- Because loans cannot be immediately recalled, the ibility commitment (by limiting the lender of last capital outflows caused by reducing M produces a resort capability of the Central Bank) would con- banking crisis. Thus high reserve requirements can vert a balance of payments problem into a banking help reduce financial vulnerability because, when problem. Powerful and simple, the money and banking (6) r - 1, m -* 1 and dL/dM - 0. multiplier is perhaps the most widely used instru- ment for monetary targeting and financial program- Simple and powerful as it is, the multiplier can be ming (under the assumption that the Central Bank misleading if used to argue that capital outflows controls the monetary base). In an open economy generally will produce a crisis. As emphasized in simple multiplier effects must be carefully analyzed Fernandez and Guidotti (1995), reserve require- to understand the role of capital flows. A backward ments and capital requirements jointly determine 25 Does Argentina Provide a Casefor Narrow Banking? the structure of the financial system and, under also shows that a set of variables representing some conditions explicitly considered in standard risk and efficiency measured as of November prudential banking regulation, capital flows could 1994 defines the (future) classes of failing, be irrelevant to the stability of the banking system. merging, and surviving banks. Failing banks Capital requirements establish a minimum were among the worst (riskiest) banks, and amount of capital as a proportion k of risky assets merging banks were high-risk, inefficient A. (This is a Basle constraint of the form K2kA.) The banks with possible diseconomies of scope and balance sheet of the consolidated commercial bank- scale. This finding suggests that there was ing system is depositor discipline during the panic-that is, depositors punished more heavily institutions (7) A + R = D + K. that were seen as being less able to survive the confidence shock (Schumacher 1996). Taking K as a predetermined variable (given that * Some $9.4 billion in liquid resources (about 20 commercial banks are not free to change it at will percent of total deposits) invested in deposits at without the consent of the Central Bank) when legal the Central Bank. This mass of liquidity was reserve requirements are zero, banks are free to helpful in two ways. It helped control credit determine the technical level for R and, assuming growth in the pre-panic period, when deposits that the Basle constraint holds as strict equality, then grew 450 percent between April 1991 and December 1994. It was also used to compensate (8) D - R = K[ (1- k)/k ]. for the fall in total deposits during the panic and thus minimized the total fall in credit and This means that the loanable capacity out of deposits bank losses. is independent of M (that is, dL/dM = 0), implying By mid-May 1995 the total loss in deposits that capital outflows cannot produce a banking crisis was $8 billion, or about 18 percent of total if the Basle constraint (and not legal reserve require- deposits (table 1). Of this, $3.4 billion was com- ments) dominates the expansion and contraction pensated by releasing reserve requirements, $2.3 dynamics of banks. This ability to smooth the impact billion in repos and Central Bank loans to banks of capital flows comes from the fact that excess sup- and $1 billion in credit reduction-that is, 41 per- ply of intemational short-run capital cannot be cent of the total fall was compensated with immediately converted into deposits because the reserve requirements and only 13 percent with a commercial bank capital will first have to increase. credit cut. This cut represented 2.3 percent of This takes time, thus insulating the domestic system credit outstanding on December 21, 1994. from extemally generated financial volatility. Thus Argentina's banking regime-with no deposit insurance and a limited role for the Central How Argentina's Banking Regime Helped Bank as lender of last resort-appears to have cre- Minimize the Social Costs of the Panic ated sufficient depositor discipline without exposing the system to high social costs. The social cost of panics is measured mainly by the fall in the amount of outstanding credit, especially if Changes Introduced after the Panic healthy institutions fail. When the shock hit the Argentine economy, banks had: Although the crisis was managed successfully, in its * An average nominal capital-asset ratio of 13.4 wake attention was given to a more efficient alloca- percent (18.2 percent when assets are adjusted tion of resources that could also preserve the neces- by risk). That served as a buffer against the sary amount of liquidity. Thus important changes in losses suffered by banks, and there was enough regulations were introduced, in particular: depositor discipline to ensure that depositors * Reserve requirements were replaced by liquid- punished more heavily those institutions that ity requirements, and uniform rates were were perceived ex ante as being weaker and established for checking accounts and term thus less likely to survive. deposits. A study of banks that merged or failed * Liquidity requirements were extended to other because of the panic shows that these banks liabilities. suffered higher withdrawals than surviving * New rules on information disclosure. banks in general and than surviving banks with * More power to the Central Bank to impose similar characteristics in particular. The study penalties. 26 Roque Fernandez and Liliana Schumacher Table 1 Changes in the Argentine banking system, December 1994-July 1995 (millions of pesos) New liquidity New Central Credit to created by the Central Bank Bank liquidityl private sectorl change in change in Total Credit to Loans Reserve deposits deposits Period deposits private sectora Repos for banks requirementsb (percent) (percent) Stock, December 21 45,367 48,221 0 58 9,416 December 21-February 28 -3242 304 369 256 2,400 93.31 -9.38 March 1-March 31 -4170 -1424 436 842 1,000 54.63 34.15 April 1-May 14 -948 9 15 439 0 47.89 -0.95 May 15-July 31 3,357 110 -431 16 -200 -18.32 -3.28 December 21-May 14 -8,360 -1,111 820 1,537 3,400 68.86 13.29 December 21-July 31 -5,003 -1,001 389 1,553 3,200 102.78 20.01 Note: New liquidity created by the Central Bank and change m credit to the private sector do not equal total deposits because of other sources of liquidity (such as external credit lines). a. Unofficial estimate based on banks' balance sheet data and daily survey conducted by the Central Bank. b. Includes reserves in Central Bank accounts, reserves for dollar deposits, and cash in vault as of December 21, 1994. Source: Central Bank of Argentina 1995. Liquidity requirements. The panic showed that together with the interest rates offered for different reserve requirements gave the Central Bank the tools types of deposits and included in certificates and it needed to fight the crisis. But reserve requirements checking account statements to clients. This regula- also affected the cost of credit. Thus the Central Bank tion goes into effect in September 1997. decided to preserve the amount of liquidity in the sys- temwhile minimizing its cost, and banks were allowed More power to the Central Bank to impose penalties. to invest their reserves in low-risk assets such as bonds The 1995 amendment to the Financial Institutions issued by OECD countries or repo transactions with Act extended Central Bank power to impose penal- the Central Bank. More important, banks can now ties on bankers who violate regulations. The invest reserves in bonds issued by the Argentine gov- Central Bank can impose monetary penalties and enmnent or mortgage-backed securities when their ban people (temporarily or permanently) from prices are protected by a put option. Rates were set at becoming managers, board members, or share- 15 percent for checking accounts and deposits up to holders in another financial institution. A law 90 days, 10 percent for deposits between 90 and 179 recently passed by Congress extended the Central days, and 5 percent for deposits between 180 days Bank's authority to impose the same type of penal- and a year. ties on risk agencies and external auditors when they break Central Bank or professional regula- Extension of liquidity requirements to nondeposit lia- tions. The Central Bank also has the power to bilities. Banks for which deposits were a lower share remove their licenses. Such punishments increase of total liabilities (that is, wholesale banks oriented to dramatically the cost of irresponsible or criminal trading activities) suffered higher withdrawals. This action and thus act as a powerful deterrent to such suggested that the contribution of each bank to sys- action. Increasing personal accountability also temic liquidity should depend on all liabilities, not increases the transparency of decisionmaking just deposits. Consequently, banks were made to within the banks and the relationship between an invest a share of their liabilities (5 to 15 percent, institution and its clients. depending on maturity) in risk-free assets. Is Narrow Banking the Answer for Argentina? Information disclosure. The panic showed that depositors used the information available about Narrow banking has been proposed at various times banks' portfolio quality. To improve the assessment and places with different goals, in particular to of risk made by depositors, in September 1995 the restore to monetary authorities control over money Central Bank required that all financial institutions supply and to prevent panics or, if there is deposit be rated by one or two agencies (depending on the insurance, to prevent it from inducing moral hazard bank's size). Moreover, after being approved by the behavior. Does narrow banking offer a better alter- Superintendency, the rating must be advertised native to the current Argentine regime? 27 Does Argentina Provide a Case for Narrow Banking? Given a fixed exchange rate, narrow banking alone financial contract that pays a sure retum. At the would be insufficient to achieve the first goal, since same time large, better-informed, or institu- under a fixed exchange rate regime the mioney sup- tional depositors will choose to contract in cap- ply is endogenously determined. It could, however, ital markets that offer higher retums and make a contribution because narrow banks, by dis- greater volatility. By allowing economic agents couraging highly volatile investors with their lower to get into a contract according to their prefer- returns, will transfer to the capital market most of the ences and abilities to monitor counterparties, instability that is borne by the banking sector and the narrow banking improves social welfare. In payments system. terms of political motivations, narrow banking With a constant real demand for domestic money, provides a natural protection to small savers the main source of instability in a fixed exchange rate and passes volatility on to those who know regime is from external capital flows. Many propos- how to price it. als have been made to exercise some control over These features seem to be particularly attractive those flows (Calvo, Leiderman, and Reinhart 1993, for Argentina given its restrictions on providing liq- 1994). Narrow banking seems to be a better alterna- uidity, the limitations of insurance funded by banks, tive, since it can achieve the same goal without gov- and the reluctance of politicians and citizens to estab- ernment intervention. Thus narrow banking could lish deposit insurance or bail out banks with fiscal make a substantial contribution to monetary stabil- resources. Narrow banking could be a path to recon- ity without introducing distortions. These points are cile the sure return demanded by small depositors especially relevant to the dynamics of the Argentine with the discipline imposed by financial decisions, panic. Between November 1994 and March 1995, and the stability of the payments system with a phi- 3,490 large depositors (those holding more than losophy of no distorting government intervention. $100,000) left Argentine banks. This exodus caused But would the failure of a large nonbank induce bank deposits to drop by $3 billion-representing 90 the authorities to bail it out in anyway? In principle, percent of the fall in term deposits and 64 percent of the commercial paper market should provide strong the drop in total deposits. Thus demand deposits and incentives for nonbanks to avoid excessive risk tak- term deposits in small denominations provided core ing. But even if a nonbank risked failure, the Central deposits that together with reserve requirements Bank would not be a lender of last resort or bail it out were helpful in keeping the system going. because there is no reason to fear contagion effects. The second issue is whether narrow banking The failure of El Hogar Obrero in 1991 provides evi- would be more effective in controlling panics than a dence of this; there were no spillover effects to the system based on liquidity requirements, the Central banking sector. Thus there seems to be good reasons Bank's (limited) ability to provide liquidity, deposi- to believe that Argentina could benefit from narrow tor discipline, and information disclosure. From this banking. point of view narrow banking is also superior not But an important point to remember is that nar- only to the current Argentine regime but also to a row banking relies on active capital markets. The regime based on deposit insurance. There are several problem is that if demand deposits, savings reasons: accounts, and short-term deposits are all invested in * Narrow banking removes almost all the asym- safe assets, there may be nothing left to fund long- metry of information because bank assets are term (risky) projects. Thus the pros and cons of each not only made risk free but they are also mar- regime can be seriously addressed only after a capi- ketable. Thus the chance of a contagious run is tal market develops. much lower, even without deposit insurance. Argentina has been making big strides in this One limitation of the Argentine regime is that regard. Part of private investment is being funded with no matter how much information is available to private bonds. These were unknown in Argentina depositors, there will always be some remain- before the Convertibility Plan, partly because of high ing asymmetry of information between bank inflation and partly because their tax treatment made owners and managers and bank depositors, competition with public bonds impossible. In January because bank assets are information intensive 1995 a law was passed that provides the legal frame- by nature. work for closed-end funds and consequently for the * Narrow banking introduces a separating equi- development of asset-backed securities. In October librium, given that risk-averse depositors and 1995 the Central Bank approved a standardized mort- small depositors that cannot afford the cost of gage loan contract that will make easier the securitiza- acquiring information will have access to a tion of residential mortgages. Banco Hipotecario 28 Roque Fernandez and Liliana Schumacher Nacional, a federally owned bank, will be originator of subject to: the loans and also will sponsor mortgage-backed secu- rities by other financial institutions. (2) 2 [R(1=-(c)l Securitization reduces capital requirements (high 2 1 (- a)j in Argentina) by removing assets on balance sheets and increases retums by generating fees. Securitization also provides companies with direct where ca is the proportion of type 1 individuals and cl access to the capital market and to broader and is the consumption of the individual type i at time j. cheaper funding sources. In this sense, it could be the vehicle for a natural transition to narrow banking in Additionally, U' > 0; U" < 0; risk aversion implies: Argentina. (3) U(L) Appendix pU(R) The point will be made in this appendix that, under Under this choice of preferences, the optimal con- certain assumptions, a narrow bank plus an insur- sumption bundle satisfies: ance fund can achieve the optimal consumption bun- dle derived in Diamond and Dybvig (1983). The first cl= c2= 0 section summarizes the main features of the model presented by Diamond and Dybvig. The second sec- U'(cl) - R tion shows how a combination of a narrow bank and pU'(c2,) an insurance fund can achieve optimal allocations. The third section elaborates on the assumptions and made in the second section. Diamond and Dybvig's model (4) 1* R(1; a)ci] 1 Diamond and Dybvig present an economy with three periods and a single homogenous good in Equations 3 and 4 taken together imply that which there is a riskless technology that can be rep- resented in the following way: cl*> 1; c2 < R T =O T =1 T =2 This means, as shown in figure A.1, that the optimal -1 1 R pair is located to the right of the consumption allowed by the technology. That is, a unit of good invested at t = 0 allows to con- sume one unit of good if the production process is It is important to note that interrupted at t = 1 and R if the process is allowed to continue until the next period, such that R > 1. cl*= f(a T, U) Interruption is costly and any amount of physical good that is taken out of the process cannot be reused c'= g(2a, T, U) as an input. There are two types of individuals. Types depend That is, optimal consumption in each period is a on their consumption needs at t = 1 and t = 2. function of the parameters of the model: technology, Preferences are such that individuals consume only preferences, and the proportion of people that turn in one period depending on their type: type 1 will out to be type 1 or type 2. only consume at t = 1 and type 2 will only consume In addition, individual types are private informa- at t = 2. At t = 0 there is uncertainty about how many tion. If a contract is contingent on an event, then it individuals will be type 1 or type 2, and types once must be known whether or not the event occurred. revealed at t = 1 are private information. A social Thus Diamond and Dybvig conclude that regular planner maximizes a social utility function: insutrance contracts cannot be written, but the fact that the optimal pairs satisfy the self-selection (1) M ax { aU(cl) + (1- a) pU(c2) constraint allows a bank to write contracts that offer {cl, c 2) 1 2 rl at t = 1 such that rl > 1. If rl happens to be equal 29 Does Argentina Provide a Casefor Narrow Banking? Figure A.1 How a narrow bank and an insurer banking contract is optimal and only a good Nash achieve the optimal solution equilibrium is possible. The point that will be made Date 2 here is that under these same conditions there is no need for an illiquid bank. Following Wallace (1996), a narrow bank in a con- text of Diamond and Dybvig model can be identified with a bank that offers the following returns: clf= 1; c2'= R That is, a narrow bank is a liquid bank. Assume now the following contract written at t 0 O between the agents, the bank, and an insurer: At t = 1 individuals R { that turn out to be type 1 will get 1 from the narrow bank and z from the insurer (see figure A.1); the pay- ment by the insurer will be funded with the insur- ance fees collected from type 2 individuals and equal C2 - --------------t---------s to x/R. The narrow bank pays 1 at t = 1 and R at t = 2 , | \ ~~~~~~~~and is also in charge of transferring the fees to the z \ insurer. Since the contract satisfies the self-selection l. l/ constraint, no individual 2 will claim to be type 1. We 1 Cl 1/a should remember here that it is costly to interrupt the Date 1 technological process; thus type 2 individuals cannot claim to be type 1 and use the proceeds of their with- to cl, then a bank contract is optimal, although two drawals to get extra consumption in the next period. Nash equilibriums can arise. In the good equilibrium Thus individuals do not have incentives to misrep- individuals assume that everybody behaves accord- resent themselves. The advantages of these solutions ing to their types. In the bad equilibrium-a sunspot are that funds in the insurance pool are not used to phenomenon-everybody realizes that the bank is fund production but are available for the extra con- illiquid (since it is promising rl > 1) and runs the sumption needed at t = 1, and the bank is liquid and banks. not subject to runs. If a is known, then a suspension mechanism can If the insurer is subject to a sequential service con- be implemented and this assures that only the good straint, as the bank in the Diamond and Dybvig Nash equilibrium achieves. If a is random, Diamond model, the same bad Nash equilibrium can arise; and Dybvig conclude that only deposit insurance then the insurer will need some mechanism that provided by the government can prevent runs. But allows him to pay only up to an amount compatible what Diamond and Dybvig call deposit insurance is with the known a. But if the sequential service con- a payment made by the government to those who straint is removed with respect to the insurer, then at claim to be type 1, funded with an optimal tax. An the end of the day the bank can transfer the exact optimal tax is described as one that is levied after amount of fees compatible with the number of indi- withdrawals take place on those who withdrew viduals that claimed to be type 1. The insurer will early. If necessary, taxes collected are invested back then distribute that amount among all the type 1 in the bank. Since individuals are concerned with the individuals. That is, at the end of the day type 1 indi- after-tax value of the proceeds from their with- viduals will get one unit from the bank plus the pay- drawals, only people who really have liquidity needs ment by the insurer, or: will withdraw at t = 1, thus allowing this bank con- tract with deposit insurance to achieve the optimal l -1 +(1- a)Nx/R 1* consumption. I I Na I Achieving optimal allocations which by definition is the optimal amount. In this case there is no need for a suspension-type Diamond and Dybvig rule out insurance because mechanism. Furthermore, once the sequential ser- individual types are private information. But if the vice constraint is removed, the optimal consumption number of individuals in each type is known, then a can be achieved even if it is random. 30 Roque Fernandez and Liliana Schumacher Are the assumptions realistic? . 1994. "The Capital Inflows Problem: Concepts and Issues" Contemporary Economic Policy Uuly)12: 54-66. If the sequential service constraint is maintained, then Central Bank of Argentina. 1995. "Managing a Liquidity the crucial assumption in the last section is that a, the Shock." Research Department, Buenos Aires. fraction of people turning out to be illiquid, is known. Diamond, Douglas W., and Philip H. Dybvig. 1983. "Bank How realistic is this assumption? Wallace (1986, 1996) Runs, Deposit Insurance, and Liquidity." Journal of and Diamond and Dybvig conclude that, since sus- Political Economy 91 (une): 401-19. pension did not work in the historical experience, a Ely, B. 1991. "The Narrow Bank: A Flawed Response to the should be random and many episodes that were Failing Deposit Insurance." Regulation 14 (spring): believed to be bubbles were actually real liquidity 44-52. shocks, or high realization of the random ax. Although Fernandez, Roque. 1990. "Comentarios sobre el Proyecto it is true that suspension was not very effective, a ran- Oficial de Reforma de la Carta Organica del Banco dom cc might or might not be the answer. When think- Central de la Republica Argentina" Annual Meeting ing of an empirical counterpart for a, the best of the Association of Argentine Banks (ADEBA), candidate seems to be the aggregate demand for August. money for transactions purposes. Many authors have Fernandez, Roque, and Pablo Guidotti. 1995. "Regulating estimated the aggregate demand for money for differ- the Banking Industry in Transition Economies." Central ent countries, and it is possible to predict an expected Bank of Argentina, Buenos Aires. value. The question is whether an error term can be Friedman, Milton. 1959. A Program for Monetary Stability. held responsible for many episodes of past panics. New York: Fordham University Press. In any case, according to equation 2 above, if the Friedman, Milton, and Anna Schwartz. 1962. "A Monetary number of individuals that turns out to be illiquid is History of the United States, 1857/1960." National random, optimal consumption should be random too. Bureau of Economic Research, Cambridge, Mass. Wallace (1996) reaches a similar solution, and con- Litan, Robert. 1987. "What Should Banks Do?" Brookings cludes that "equilibrium arrangements involve some Institution, Washington, D.C. dependence of returns on the order people withdraw"; Phillips, Ronnie. 1995. The Chicago Plan &New Deal Banking that is, for certain economies debt is not an optimal Reform. Armonk, N.Y.: Sharpe. contract any more. This takes him closer to recent nar- Pulley, Lawrence, and David Humphrey. 1993. "The Role row banking proposals like Litan (1987), which sug- of Fixed Costs and Cost Complementarities in gested that narrow banks should be required to Determining Scope Economies and the Cost of operate as money market mutual funds, or earlier nar- Narrow Banking Proposals." Journal of Business 66 row banking proposals like Simons (1948), which sug- (31): 437-62. gested that equity could be a preferred arrangement. Schumacher, Liliana. 1996. "Bubble or Market Nevertheless, it should be taken into account that in Discipline: A Study of Failures and Mergers over the Diamond and Dybvig there are no asymmetries of Argentine Panic." Central Bank of Argentina, Buenos information with respect to the agent's actions or out- Aires. put and thus some equity-like forms of risk-sharing Simons, Henry. 1948. Economic Policy For a Free Society. arrangements might turn out to be first best. Chicago: University of Chicago Press. Wallace, Neil. 1988. "Another Attempt to Explain an References Illiquid Banking System: The Diamond and Dybvig Model with Sequential Service Constraint Taken Bryant, Lowell. 1988. "Breaking Up The Bank: Rethinking Seriously." Federal Reserve Bank of Minneapolis. an Industry under Siege." Dow Jones Irwin. Quarterly Review (fall). Calvo, G., L. Leiderman, and C. Reinhart. 1993. "Capital - . 1996. "Narrow Banking Meets The Diamond- Inflows to Latin America: The Role of External Factors" Dybvig Model." Federal Reserve Bank of Minneapolis. IMF Staff Papers (March) 10: 108-51. Quarterly Review (winter). 31 Comment Gerard Caprio Jr. R oque Fernandez and Liliana Schumacher invest only in safe assets, and other institutions deserve credit for examining alternatives to (called nonbanks) will provide risky short- and the OECD model of banking, which stipulates long-term finance using loans, equity, and deben- that countries should adopt an 8 percent capital ade- - tures. Some assessments of narrow banking argue quacy rule and work on improving bank supervi- that it is not practical where there is an insufficient sion. Such an approach is fundamentally flawed. Not supply of riskless or low-risk paper. This claim is only is the agreed capital level low by historical stan- useful to explore because it points the way to dards, it also ignores noncredit risks, as well as the understanding possible adjustments if a narrow likelihood that private agents will respond when reg- bank model were adopted. ulated. It is hard to believe that even if the OECD Suppose that a developing country has only $1 model were appropriate for industrial countries- million of government debt outstanding in a sys- and a string of banking crises from Japan to tem with initial demand deposits of $100 million. If Scandinavia suggests otherwise-that the same stan- banks are required to hold government debt to dard would be sensible for smaller, more concen- back their deposits, their first reaction might be to trated developing countries. Developing countries bid up the price of this debt and correspondingly that have experienced systemic episodes of bank lower the rate of return both on this paper and on insolvency saw far greater terms of trade volatility what they can afford to pay on demand deposits. If than countries where banking problems were not citizens really want the security of having their systemic (Caprio and Klingebiel 1996). Thus it is funds in banks, then they might have to pay banks about time that greater efforts be made to examine to hold their funds-much like depositors of specie alternative models to ensure safe and sound risk-tak- used to do for the early banks, which were nothing ing in the financial sector. more than lock boxes until those tending them saw In searching for a replacement, one requirement that they could lend the idle funds and make a nice should be that the proposed framework reduces the return. In other words, citizens that insist on hav- need for a "world class" supervisory system. Such ing security can pay for the privilege, and their systems are extremely difficult to attain in rich coun- banks might then have to pay for the ability to hold tries; achieving them in poor countries seems almost govemment debt. Thus government debt would impossible. A second requirement is that incentives trade above par and the interest rate would be for participants in financial sector activities should negative. reward prudent risk-taking. This is not meant to This outcome is unlikely. Instead, as the rate of deter all high-risk ventures, but rather to ensure that return on demand deposits declines, depositors financial intermediaries fund a diverse bundle of likely will diversify away from these deposits in activities but do not gamble excessively. Narrow favor of riskier but potentially more remunerative banking is attractive because it appears to meet this investments-as U.S. depositors did in 1994 when key requirement. What are its pros and cons? riskless interest rates plummeted to their lowest Fernandez and Schumacher remind us that nar- level in more than thirty years. This likely adjust- row banking will mean that narrow banks will ment should be the focus of the debate on narrow Gerard Caprio Jr. is lead economist m the Policy Research Departrnent at the World Bank. 32 Gerard Caprio Jr. banking. If a sizable chunk of funds migrates from look at other possible ways to align incentives, such (narrow) banks to nonbanks, whether there is any as by ensuring mutual liability among banks, raising gain from adopting narrow banking tums on capital ratios, raising liability limits on bank owners, whether government guarantees migrate along or increasing franchise values (making greater bank with the funds. If governments guarantee the large profitability the carrot that induces greater pru- holdings of nonbanks, clearly there is no gain. All dence). All of these approaches would need to be that has happened is a relabeling of the banking coupled with better information and disclosure, so system. that outsiders could serve as effective monitors of Would governments guarantee nonbanks? This banks. Also, a second Chicago Plan, namely a question lies in the domain of political economy. requirement that banks issue large amounts of unin- Argentine authorities may well be able to contain sured subordinated debt, merits debate (Caprio the spread of guarantees. In the United States the 1994). These alternatives have the advantage of pre- adoption of narrow banking would lead to a huge serving the synergies between lending and deposit migration of funds in favor of nonbanks, substan- taking, the loss of which has been offered as an argu- tially increasing the size of companies like ment against narrow banking. American Express, Merrill Lynch, and Fidelity Fernandez and Schumacher also are concemed Investment. If these organizations offered not only about how investment will be funded in a narrow equity instruments but also loans and debentures, banking model. Although I believe that the market they would be susceptible to failure and contagion. will take care of this problem, if the authors want As Caprio and others (1996, p. 22) put it: to follow the evolutionary path they appear to sponsor then they could note that it would be The fundamental problem is that the commu- important to make sure that deposit insurance in nity wishes to transform some current output the current banking system is not underpriced. The into a capital stock; the capital stock is itself problem is that it is not possible to determine the illiquid. If any claims that ultimately derive price of an open-ended guarantee, and bankers can their value from the capital stock have fixed adjust to the existence of a safety net by taking nominal values, there is a risk that someone more risks. Still, if some control of risk-taking is must bear. possible, it should carry high insurance premiums to drive banking into money market mutual funds, Femandez and Schumacher say that narrow thereby hastening narrow banking. This approach banking removes almost all asymmetry of infor- would only make sense if the previous considera- mation, but it does not change this problem. So tions about government guarantees for nonbanks with this risk there is the possibility that the gov- were addressed satisfactorily. emment would ensure it and, in the U.S. case, that Femandez and Schumacher also argue that high large nonbanks also would be too big to fail. capital levels and high reserve requirements can Although the authorities were willing to allow shelter the domestic financial system. Caprio and Drexel to fail in 1987, it appeared that the Federal others (1996) argue the opposite with respect to Reserve was propping up nonbanks in the face of a reserve requirements, namely, that in a relatively large systemic shock. Thus, in any attempt to adopt simple model it is easy to show that the higher is narrow banking, governments must plan how they the reserve requirement the greater is the change in will limit the spread of guarantees on larger and the loan rate in response to a change in world inter- more politically powerful nonbanks and be pre- est rates. So with high reserve requirements, a pos- pared for the difficulty of making such plans cred- itive shock would increase capital inflows beyond ible. If the guarantee moves with the funds, the what they would otherwise be and a negative danger is that a substantial portion of the financial shock would lower them. Reserve requirements system would be without much supervision. As in only help as a tool against capital flows if the gov- banking in many developing countries in the 1980s emment is willing to raise and lower them in the and 1990s, govemments would quickly see that, if face of shocks; an endogenous reserve requirement they are going to provide a safety net, they need to that increases reserve ratios as world rates fall insu- be able to supervise. And if supervision is required, lates the domestic loan market from the extemal then one of the main attractions of narrow banking shock, not because it discourages international cap- disappears. ital mobility but because it discourages domestic In considering narrow banking, it may be helpful financial intermediation. The growth effects of to compare it not just to the OECD model, but also to such a policy need to be explored. 33 Comment on "Does Argentina Provide a Casefor Narrow Banking?" References Paper prepared for the World Bank's Annual Bank Conference on Development Economnics, April 25-26, Caprio, Gerard Jr. 1994. "Bank Regulation: The Case of the Washington, D.C. Missing Model." Paper presented at KPMG-Brookings Caprio, Gerard Jr., Michael Dooley, Danny Leipziger, and Institution Conference on Sequencing Financial Sector Carl Walsh. 1996. "The Lender of Last Resort Function Reforrns, October 3-4, Washington, D.C. under a Currency Board: The Case of Argentina." Policy Caprio, Gerard Jr., and Daniela Klingebiel. 1996. "Bank Research Working Paper 1648. World Bank, Insolvency: Bad Luck, Bad Policy, or Bad Banking?" Washington, D.C. 34 Comment George J. Benston R q oque Fernandez and Liliana Schumacher iden- Pulley and Humphrey (1988) obtained their find- tify two major possible advantages of narrow ings from a study of 205 large U.S. banks (all with banking-better control over the money sup- assets over $1 billion in 1988) in 1978-88 and 265 ply and increased bank efficiency-and three possible large banks in 1989-90. Assuming that their esti- disadvantages-banking and credit inefficiency and mates were valid,1 they point out an important limi- the failure of large nonbanks. The authors conclude tation of their study: that the possible disadvantages are not present, will be overcome by the development of a capital market, or while there can be joint production of deposits are not important. Thus it appears that narrow bank- and loans within a state, this is not yet permit- ing offers some benefits and few, if any, disadvantages. ted [in the United States] interstate. Thus sepa- ls this an accurate assessment? rate loan-production offices of large banks operate in major cities outside of a bank's home Presumed Disadvantages of Narrow Banking state and cannot provide deposit services. Had interstate branching been permitted, our scope Fernandez and Schumacher identify three potential economies may have been larger. (p. 458) disadvantages of narrow banking: the loss of economies of scope, the limited capacity to fund risky In addition, and of greater importance, their projects, and the need for governments to bail out large study suffers from four other limitations. First, the nonbanks. However, the study on which they base benefits to customers of obtaining deposit and their condusion-that narrow banking would not be lending services from the same organization are inefficient-is seriously flawed. Although the second not considered. These benefits include savings in and third disadvantages largely could be overcome, the cost of providing information to and obtaining the development of alternative institutions and the information from lenders. Second, if overdraft restructuring of commercial banks into narrow banks banking and interest payments on deposits were is likely to be costly. permitted, customers could benefit from using a single account for payments, investments, and Economies of scope loans, which would reduce transactions costs and the cost of dealing with variation and uncertainty Femandez and Schumacher dismiss the concern that in cash flows. Third, all banks in the United States narrow banking would result in banking inefficiency are permitted to offer all of the deposit and loan by citing Pulley and Humphrey (1993), who find little products used in the analysis. They are not evidence of economies of scope from cost complemen- required to offer loans. That none of the banks stud- tarities between deposits and loans, although they do ied offers only deposit services indicates that they find savings in fixed costs of 4-5 percent of total oper- find joint production of deposit and lending prod- ating expenses. These savings, Pulley and Humphrey ucts to be beneficial. Indeed, few (if any) banks in claim, would be offset by savings in deposit insurance the United States or any other country offer only costs, keeping the cost of narrow banking small. deposit services. Fourth, because the banks studied George J. Benston is John H. Harland Professor of Finance, Accounting, and Economics in the Goizueta Business School at Emory University. 35 Comment on "Does Argentina Provide a Casefor Narrow Banking?" are not legally constrained from providing deposit Bailout of large nonbanks and loan services in amounts that they find prof- itable (unlike the Glass-Steagall Act's constraint on Femandez and Schumacher give the example of a securities services), it is not conceptually possible large Argentinian nonbank, El Hogar Obrero, whose for researchers to measure economies of scope. failure in 1991 did not spill over into the banking sec- Thus the Pulley and Humphrey study does not and tor. Considering the paucity of evidence about con- cannot provide useful evidence showing that there tagious failures in banking, this outcome is not would be little loss if banks were not permitted to surprising. What is surprising is that the authorities offer both deposits and loans. did not intervene. In other countries where large banks and nondepository financial institutions Fundingfor risky projects failed, the authorities have stepped in to protect depositors and debtholders (although usually not As Fernandez and Schurnacher suggest, nondeposi- stockholders). This loss of support is an important tory financial institutions could indeed provide cost of narrow banking. funds for risky projects. Companies in the United Although narrow banks would almost never fail, States, such as General Electric Credit Corporation, institutions that are not called banks but that offer Merrill Lynch, and Household Finance, make a large bank-like services probably will spring up or volume of business and personal loans. In a way they expand, and some will fail. The authorities, then, are have a comparative advantage over banks in that likely to be under considerable public pressure to they are almost unregulated. They can open offices bail out "depositors," who will claim that they were almost anywhere, are not examined or supervised, not aware that they were not protected. For example, and can structure loans as they and their customers Australian authorities bailed out the "depositors" in please, subject only to state usury statutes and fed- a large real estate investment company (see Kane and eral fair lending laws. Kaufman 1993). Furthermore, other customers and As Fernandez and Schumacher point out, bonds employees of these institutions will assert that the and closed-end equity funds can provide a means for closure of the firm will damage them and their com- businesses to obtain financing. These sources of munities. In the past such pressures have been suc- funds, however, often are inferior to loans. Loans cessful in getting the authorities to intervene. offer borrowers flexibility in amounts and repay- ment schedules. By contrast, bonds usually are Evaluation of Presumed Advantages of issued in relatively large amounts with fixed maturi- Narrow Banking ties and are subject to agency risk. Equities subject issuers to possible loss of control or subject buyers to It is likely that traditional banking offers both pro- the risks of minority ownership status. Loans can be ducers and customers of banking services the advan- made, however, by companies that are funded with tage of economies of scope. These would be lost if bonds and equity rather than with deposits. These narrow banking were mandated. The other two pre- companies also could be funded with savings. sumed disadvantages could be largely overcome. Assuming that demand deposits (transactions However, both the development of alternative insti- accounts) would be held only or predominantly by tutions and the restructuring of commercial banks narrow banks, these banks would have to hold a into narrow banks are likely to be costly. large proportion of the national stock of short-term Consequently, I turn now to an analysis of the pre- treasury obligations and commercial paper and sumed advantages of narrow banking. other similar short-term and virtually credit risk-free corporate paper. As Fernandez and Schumacher Money supply control acknowledge, these resources would not be available for longer-term corporate debt. Narrow banks, how- Fractional reserve banking. Fractional reserve banking ever, might be permitted to hold diversified portfo- has long been a source of changes in the supply of lios of marketable longer-term corporate debt and deposit money. Femandez and Schumacher incor- equity if they held sufficient capital to absorb fully rectly point out that this source of instability would the interest rate and credit losses that might be be eliminated if banks were required to hold 100 per- incurred. In this situation longer-term financing cent of their deposits in short-term treasury bills. would not be constrained. But then the "narrow" Treasury bills, like bank loans, are earning assets but banks would look very much like well-capitalized carry no credit risk and a small interest rate risk. commercial banks. Banks would still hold fractional reserves in cash in 36 George J. Benston order to meet depositors' demands. If depositors ran Banking efficiencies to currency (say, because they feared that their bank's assets were being dissipated by fraud), the money Fernandez and Schumacher identify five efficiencies supply would change as the narrow banks sold their that they believe could be gained from narrow bank- treasury bills to obtain the currency. Thus there ing-prevention of bank runs, low-cost deposit would be a multiple contraction of the money supply. insurance, low interest rates on deposits and low bank capital, reduced systematic risk, and facilitation Fixed exchange rates and volatile investors. The only of securitized credits for loans. depositors who would be attracted to narrow banks would be people holding funds for transactions. The Bank runs. Narrow banking would indeed prevent highly volatile investors about whom Femandez and bank runs. But bank runs are not a problem if there Schumacher are concerned would put their funds in is deposit insurance or if the central bank does its job. nonbanks. But these funds would still be highly Moreover, if narrow banking were adopted, the wall volatile, which presumably means that investors between classic intermediation and the securities would move their funds in and out of the country business would be strengthened, not broken. The quickly. Thus even though the money supply was U.S. Glass-Steagall Act separated commercial and not being disrupted (if it were held entirely at narrow investment banking not because the combination banks, which is doubtful), capital markets would be made commercial banks more risky, but because at subjected to instability (as Femandez and that time (1933) the separation benefited the institu- Schumacher point out). tions involved, though at the expense of consumers (Benston 1990). Narrow banking would prevent the Money supply control generally. In a traditional recombination, except perhaps through bank hold- banking regime the central bank can control the ing companies, and this would obviate most benefits money supply with open market operations. from economies of scope and diversification of assets Reserves against deposits are not required so long as and operations. the central bank can predict banks' desired reserves. When deposits are reliably insured, there is no incen- Deposit insurance. Deposit insurance would be less tive for protected depositors to run to currency; expensive under narrow banking. But an alternative hence there is no contagion effect when one or more scheme that would permit banks to operate largely banks fail. Uninsured depositors are unlikely to hold as they have for centuries would have much the same withdrawn deposits in currency, since this would effect. subject these assets to theft and complicate transac- tions. Rather, they redeposit withdrawn funds in Interest rates paid on deposits. Interest paid on other banks. If all banks have the same ratio of deposits (demand or term) would be low under nar- deposits to reserves, the money supply will not be row banking, and narrow banks would require low affected. If ratios differ and even if there is no deposit capital. These banks' earnings also would be low, insurance and depositors run to currency, the central however-so low that substantial fees likely would bank can offset the decline in high-powered money have to be charged to demand depositors. and hold the money supply (or its rate of change) constant. Thus narrow banks are neither necessary Systemic risk. Narrow banking would eliminate risk nor desirable for money supply control. in the depository banking and payments systems. The Moreover, narrow banks are likely to make central other institutions that would provide the services now bank control of the money supply more difficult. provided by commercial banks would not be free of Because narrow banks must hold assets that have risk, however. They are likely to provide payments ser- low yields, these banks are unlikely to earn sufficient vices, possibly through narrow banks, much as brokers amounts to pay the cost of providing transactions (such as Merrill Lynch) now provide checking services services (for example, checking). By contrast, non- through commercial banks. These other institutions banks can earn higher returns from loans and have can and probably will fail. (And the authorities are incentives to offer transactions services (perhaps likely to bail them out.) Concerns about systemic risk, through electronic fund transfers or with check-like however, should be restricted to concerns about the instruments) at a lower cost. These transfer instru- central bank. As documented in Benston and Kaufman ments serve as money that is not reported to the cen- (1995), there is little reason to fear a financial panic or tral bank and, thus, is more difficult for the central systemic failure as long as the central bank does not bank to control. permnit the money supply to decline precipitously. 37 Comment on "Does Argentina Provide a Casefor Narrow Banking?" Securitized credits. Narrow banking might help from two big problems, one past and one present. substitute structured securitized credits for bank The past problem is fractional reserve banking when loans, but this is doubtful and not necessarily bene- base money could change exogenously, generally as ficial. Securitized credit primarily offers lenders the a result of gold outflows and inflows and runs to cur- advantage of being able to make and possibly service rency. A well-managed central bank can and in most loans for which they have a comparative advantage, countries has eliminated this problem. while not having to hold these loans if doing so The current problem is de jure or de facto deposit would force them to invest in an undiversified port- insurance. As Femandez and Schumacher note, pro- folio of assets. Securitization has been particularly tecting depositors from losses results in moral haz- beneficial to geographically restricted U.S. thrifts ard behavior. This problem can be solved by and banks with respect to home mortgages and, to a requiring banks to hold sufficient capital or debt that lesser extent, consumer credit. These loans are par- is not guaranteed by the government and is subordi- ticularly well suited for securitization because they nated to depositors who would absorb most of the are standard (conforming) contracts, the interest rate losses the banks might incur. Thus, as for other cor- and credit risks of which can be reasonably well esti- porations, equity holders would have no incentive to mated. Other loans, however, particularly business take excessive risks because they would bear the cost loans, are subject to moral hazard concems and oper- if things turned out badly. Equity holders also would ational problems resulting from idiosyncratic con- have to pay debt holders for the risk that the equity tracts and demands by borrowers for nonstandard might be depleted. repayment schedules (Benston 1992). In fact, an There is an alternative and operationally viable important advantage of traditional commercial bank capital requirement scheme-structured early inter- loans is that such borrowers can borrow the amounts vention and resolution-that was essentially adopted they wish when they wish and make repayments in in the United States as part of the Federal Deposit amounts and at times that best suit them. This flexi- Insurance Corporation Improvement Act of 1991 (see bility is not offered by securitized credits. Benston and Kaufman 1988, 1994). Institutions with government-insured deposits should be required to Narrow Banking in Argentina hold capital equivalent to about 10 percent of assets. If capital ratios fall below this level, the authorities Fernandez and Schumacher emphasize two aspects of first may and then must take actions to restrict the Argentine banking that might affect or be affected by offending banks' activities and get them to restore narrow banking: fixed exchange rates and high their capital. If capital falls below a positive percent- reserve requirements. High reserve requirements (43 age of assets-say, 2 or 3 percent-the banks must be percent against checking accounts) have brought taken over by the authorities. This system gives the Argentine banks halfway to narrow banking. banks a powerful incentive to avoid excessive risks Fernandez and Schumacher do not say whether banks and, if they do get into trouble, to resolve the situa- were paid a market rate of return on these reserves (if tion through additional capital infusions or voluntary held at the Central Bank) or whether the reserves merger or liquidation. Deposit insurance would be could be held in market instruments. Because they required only to protect depositors from the cost of characterize the reserve requirements as a tax on the massive fraud and very rapid asset depletion. financial system, presumably interest is not paid. In An important advantage of this scheme is that the this event demand depositors would be paid little, if banking system does not have to be restructured, as anything, on their balances and would attempt to would be required if narrow banking were adopted. keep as little on deposit as is feasible. A consequence All that is required is a substitution of explicitly unin- is that volatility would be higher both from technol- sured debt or equity for de jure or de facto insured ogy-supported float management and from substitute deposits and sufficient government regulation to means of effecting payments. This could cause sub- ensure that banks are maintaining their equity and stantial changes in the money stock. If this is the case, uninsured debt at the required level. narrow banking would likely exacerbate the situation, transferring investment volatility to nonbanks. Note An Alternative Approach 1. Pulley and Humphrey do not specify what is included in costs and do not consider the extent to which Traditional commercial banking is much better than annual accounting costs accurately measure economic narrow banking. Traditional banking has suffered costs. They do not appear to have included the number 38 George J. Benston of offices operated by banks as independent variables, Taguchi, eds.,Financial Stabilityina Changing Environment. nor do they mention converting the data to constant dol- New York: St. Martin's Press. lars, despite considerable inflation during the period Benston, George J., and George G. Kaufman. 1988. Risk and studied. Solvency Regulation of Depository Institutions: Past Policies and Current Options. Monograph Series in Finance and References Economics. New York University, Graduate School of Business Administration, New York. Benston, George J. 1990. The Seperation of Commercial and - . 1995. "Is the Banking and Payments System Investment Banking: The Glass-Steagall Act Revisited and Fragile?" Journal of Financial Services Research 9: 209-40. Reconsidered. London: Macmillan and New York: Kane, Edward. J., and George. G. Kaufman. 1993. "Incentive Oxford University Press. Conflict in Deposit-Institution Regulation: Evidence -. 1992. "The Future of Asset Securitization: The from Australia." Pacific-Basin Finance Journal 1: 13-29. Benefits and Costs of Breaking Up the Bank." Journal of Pulley, Lawrence, and David Humphrey. 1993. "The Role Applied Corporate Finance 5 (Spring): 71-82. of Fixed Costs and Cost Complementarities in - . 1995. "Safety Nets and Moral Hazard in Banking." Determining Scope Economies and the Cost of Narrow In Kuniho Sawamoto, Zenata Nakajima, and Hiroo Banking Proposals." Journal of Business (31): 437-62. 39 Comment Fernando de Santibafies A fter defining narrow banking as a banking reserve requirements to preserve a mass of liq- regime in which deposit-taking activities are uidity that could be released in case of a sys- ..,ft separated from lending activities-and temic run. ... The rates were set at 43 percent identifying features in the Argentine banking regime for checking accounts and 3 percent for term that resemble narrow banking-Roque Fernandez deposits. and Liliana Schumacher raise interesting questions about whether narrow banking would be a better Actually, after the financial crisis that followed the banking framework for Argentina, given the con- Mexican devaluation, reserve requirements were straints imposed by monetary policy, and what con- replaced by liquidity requirements, and these were ditions would be required for its successful set at 15 percent for checking accounts and deposits imnplementation. Their main conclusion is that nar- up to 90 days, 10 percent for deposits between 90 and row banking could improve the Argentine system by 179 days, and 5 percent for deposits between 180 days making it less prone to contagion, while reconciling and a year, as well as extended to liabilities other than the sure return demanded by small depositors with deposits. Thus Fernandez and Schumacher assign to the discipline imposed by financial policymaking the Argentine system some features of narrow bank- and ensuring the stability of the payments system ing because of its comparatively high reserve or liq- while avoiding distortionary government interven- uidity requirements. tion. There are two problems with this argument. First, it As the authors note, however, narrow banking completely blurs the difference between narrow bank- requires that alternative sources of funding be avail- ing and fractional reserve banking. But there is a qual- able for long-term projects; hence its implementation itative difference between the two, one completely depends on the development of capital markets. independent of the level of reserves: deposit and lend- Thus before it can adopt narrow banking, Argentina ing activities are separated in narrow banking, while must develop these markets. As these points are clar- banks grant credit in fractional reserve banking. And ified and extended, however, some of paper's obser- when reserve requirements are less than 100 percent, vations and remarks weaken and contradict the main no matter how small or large, all the problems the arguments. Here I address these breakdowns in authors identify with fractional reserve banking logic. (which I disagree with) reappear immediately. Consider this. When a fractional banking system Problems with Narrow Banking is subject to an increase (decrease) in legal reserve requirements, can it be asserted that it has become Fernandez and Schumacher contend, first, that the more (less) narrow? Are banking systems with very Argentine banking regime has features that resemble low reserve requirements slightly narrow? The anal- narrow banking. The main argument is that ogy made by the authors is trivial and misleading. Second, the authors do not recognize that the cov- given the constraints imposed by the monetary erage provided by high reserve requirements in a regime, the Central Bank established high systemic run-which allowed Argentina to prevent Fernando de Santibafles is president of the Argentine Bank of Credit. 40 Fernando de Santibarnes a 41 percent drop in deposits from translating into a The authors are wrong to claim that narrow corresponding drop in credit-is only possible banking requires the availability of altemative under a fractional reserve system. Under narrow funding for longer-term projects. This is a miscon- banking, where banks do not lend, the reduction in ception of the nature of credit and capital markets. credit generated by a financial crisis cannot be com- At all times there is a demand for credit from indi- pensated for by a reduction in reserve requirements. viduals and corporations for all maturities and var- Moreover, the high reserve requirements that the ied financial conditions; these will be provided by Argentine banking system had before the financial other financial institutions, since narrow banks are crisis (and that leads the authors incorrectly to inhibited. Thus segments of the stock of credit are speak of quasi narrow banking) were an inheritance always maturing. of the ample base that the inflation tax had in the If credit institutions are funded by issuing liabili- past, before the success of the convertibility-based ties, then during a financial crisis, as those liabilities stabilization plan. mature, there will be a withdrawal of funds by To support their main proposition-that narrow investors and credit will deteriorate. If, alternatively, banking is a superior design for a banking system- credit institutions are funded by issuing shares or the authors argue that: equivalent instruments, the prices of these instru- * It can make the banking system less prone to ments will collapse during a financial crisis. As a contagious runs. result there will be no new flow of funds to these * It reconciles the sure return demanded by small companies and they will move toward precautionary depositors with the discipline imposed by liquidity, drastically reducing credit to the private financial policies. sector. Ex ante, the demand for the liabilities of the a It reconciles the stability of the payments sys- first type of financial company will be higher than tem while avoiding distortionary government the demand for the shares of the second type of finan- intervention. cial company. The authors imply that, because the deposits The authors' propositions in favor of narrow received by narrow banks are invested in liquid and banking imply wrongly that the special financial cor- low-risk assets, regulators need not monitor banks' porations would be funded with shares or equiva- investments. Moreover, they claim that deposit lents, because otherwise the contagion effects and insurance (if it were needed) would have no signifi- systemic liquidity crisis of a run would be similar to cant fiscal cost. those caused by fractional reserve banking. But apart This point raises some ambiguity about the from the fact that the transformation of a demand for description of narrow banking and its benefits. When time and savings deposits into a demand for shares the authors refer to stability of the payments system would only be partial, the negative wealth effect of a or cite the narrow banking proposal of the 1930s in sharp fall in the price of those shares during a crisis the United States, they seem to limit narrow banking would seriously disrupt aggregate demand, activity, to demand deposits; in other parts of the paper they and employment. seem to include savings and time deposits. At some The authors confuse capital markets with matu- points they concentrate on narrow banking's ability rity matching between assets and liabilities. Credit to stabilize the payments system or the money sup- institutions also can incur mismatches that may lead ply; elsewhere they favor it for preventing the credit to a liquidity crisis. And in any case, any ability these contraction that follows a financial crisis. institutions have to match maturities by issuing Let us first dispose of the motivation cited by the long-term bonds would also be available for frac- authors for narrow banking: full control of the tional reserve credit-granting banks. money supply. Little argument is needed to appreci- The paper devotes only one paragraph to whether ate that, under full convertibility of the peso, the the failure of a large nonbank would have contagion money supply is endogenously determined. effects on other nonbanks, discarding the possibility In a system of narrow banking, with a separation without any serious argument or piece of evidence. between deposit and lending activities (narrow The formidable issues of transition from a fractional banks receive deposits that they invest in reserves or reserve system to a narrow banking system are not low-interest, low-risk assets, while credit flows addressed at all, nor is there any treatment of the through other financial institutions), the credit desta- comparative statics of the exercise. Moreover, no bilization effect of a financial crisis cannot be comparison is made between the two systems under avoided, although its dynamics depend on the normal (noncrisis) conditions-despite the excep- nature of other financial institutions. tional character of a crisis. 41 Comment on "Does Argentina Provide a Casefor Narrow Banking?" Additional Considerations channels into separate distribution networks for deposit raising, credit granting, securities and cash In the transition to narrow banking, part of the management services, and so on and distribution demand for time and savings deposits will disap- channel expenditures are the most expensive type pear (with consequences for credit supply) because of financial activity. This cost increase would come narrow banks will only be able to pay compara- at a time that frontier financial technology is mov- tively low interest rates, while many depositors will ing toward integration of financial activities, not not be prepared to invest funds in volatile capital fragmentation. In addition, the need for stronger market products. supervision of credit financial institutions will bal- Contrary to what the authors say, narrow banking ance out any cost reduction that may arise from would not allow economic agents to enter into con- downsizing a narrow bank supervision agency. tracts according to their preferences and abilities to Finally, even with further development of capital monitor counterparties. Rather, under narrow bank- markets in Argentina, it is difficult to conceive of an ing small uniformed investors will see their options intense period of commercial paper or bond issuing restricted, because they will be less able than banks by small and medium-size corporations that will see to judge the risk of a special financial corporation. their credit opportunities curtailed. An analogous Narrow banks would have to invest their deposits argument can be developed for the different credit abroad, because investing them locally would imply products for individuals. In this way narrow bank- taking excess risk if there were a financial crisis. ing will also limit the sources of credit available to Hence they would be able to pay intemational inter- non-prime corporations and individuals. est rates (minus administrative costs)-with the result that part of the local savings would flow Conclusion abroad to international banks or offshore banks of local financial corporations. This chain of events is Given these facts, it is difficult to share the authors' the natural outcome of exposing banks to conditions views on the lack of consensus of political and social under which they cannot compete with the intema- support for partially funded deposit insurance tional financial system. schemes and lender of last resort functions and a cor- Despite notable improvements in recent years, responding support for narrow banking. Argentina remains a country of high risk, and it can- Although the paper is an interesting (although at not be expected that local savings sent abroad will many points mistaken) examination of unproved revert to local corporations as credit granted by inter- theories, its views on optimal financial industry national banks. These banks are subject to the risk design and social welfare are impractical and should regulations of countries that provide their financial be discarded. systems with different mixtures of deposit insurance, Moreover, although the development of capital central bank lender of last resort function, and so on. markets in Argentina is a task of paramount impor- A second point is that, as the economies of joint tance, Argentina's biggest financial challenge is to production are loosened (with a considerably promote saving, financial saving, and credit in all reduced volume of deposits and with low returns for forms. The natural development of capital markets assets), narrow banks will have to charge customers (in which banks have and will play a major role) higher fees for their services-at a time when fee will define, through the market choices of investors, reductions are crucial to deepening bank penetration individuals, and corporations, the complementary into more segments of the population. functions between commercial banks and other Moreover, the increase in fees will be com- capital market products and institutions. It would pounded by the development of both narrow banks be irresponsible to bureaucratically destroy what and credit financial institutions. The reason is that has been achieved and to immerse financial activity narrow banking implies splitting existing banking into a vacuum. 42 Floor Discussion Question: In considering the assets of narrow bank- demand deposit or the savings account, the bank ing, all the speakers said there were two types: short- can exercise the option and return the money to the term government paper and short-term commercial depositor. Of course, those options have to be paid paper or other forms of short-term liability. But the for out of the depositor's money. The maturity trans- only true riskless asset is government paper, not formation is not free. But the good thing is that commercial paper. And since the commercial paper banks pay and tell the truth about what they are market is not well developed in many countries, nar- doing. That is the idea. row banks would end up holding mostly short-term government paper. Would this not create some pref- Question: Listening to Roque Fernandez and erential financial arrangements for governments? In George Benston, I wonder if the primary utility of Italy, for example, demand deposits are 40 percent of narrow banking comes not from thinking of it as a GDP and the treasury bills held by banks are 20 per- full-fledged alternative but rather as a sort of intel- cent of GDP. If narrow banking were introduced, it lectual punching bag with which to refine some of would force banks to buy 20 percent of GDP in trea- the details of conventional banking. My sense is that sury bills because Italy does not have a commercial most people would conclude that narrow banking is paper market. Although this would make the Italian an unrealistic approach. Consider the numbers for government happy, it might not be the best Argentina: There is about $50 billion of deposits, $80 approach. billion of bank assets, and a capital market with the capacity to generate perhaps $5 billion a year in trad- Roque Fernandez: I think we must be flexible in able securities. Even if Argentina reaches a stage the way we think about narrow banking. Although where it can process a lot more commercial paper- Argentina has a fractional reserve requirement sys- despite a fiscal situation that permits the government tem akin to traditional commercial banking, we think to issue a lot more treasury securities-the banking in terms of narrow banking. There is no short-term system would not be standing still. By then the bank- commercial paper available, but we are developing a ing system would most likely be two or three times mortgage market that we believe international banks its current size. There is no easy way of squaring will trade in. We already have some contracts to sell these numbers. Or look at the United States. What these mortgages, for example, to the Deutsche Bank. amount of additional treasury bills or commercial Options and derivatives allow you to have, for papers would have to be issued for most sight example, a put option on the mortgage. Thus banks deposits to be invested in them? Not to mention that have the option, if there is a run, to sell the mortgage the yields on the safe instrument would drop close to to the Deutsche Bank. So a local domestic financial zero, if not turn negative. institution in Argentina can have a mortgage Another aspect that has not been mentioned is financed with short-term liabilities, and if it has a that for non-narrow banks to perform the financial put option on Deutsche Bank it can replicate a sort intermediation that conventional banks perform of narrow banking situation. That way, if there is a today, they are going to have to raise trillions of dol- run or the depositor does not want to renew the lars in long-term bond markets. When both of those 43 Floor Discussion of "Does Argentina Provide a Casefor Narrow Banking?" effects-a drop in short rates and a significant spike Second, although early, structured intervention in long rates-are taken into account, it basically may work in the United States, without widespread equals what current (imperfect) systems pay in terms reform of macroeconomic policy it is difficult to of the periodic large costs of bank bailouts. Given imagine how reform of banking policies would work this, is narrow banking even desirable? in most developing countries. Given the history of And even if it were desirable, is it possible for a many countries in Latin America, are there people system like this to come about? Since there are no who would bear the risk of structured early inter- countries in which to run this experiment from vention when, for example, a country such as Chile scratch, we need to engineer a transition from an has changes in the real exchange rate that are larger existing banking system. If that is what we are deal- than anything seen in industrial countries? To what ing with, what does it amount to other than asking extent are risks being subsidized by the present sys- whether the banking capital adequacy ratio should tem? If they are not being subsidized, could you get be 15 percent, not 8 percent; whether the results anyone to bear them through structured interven- should be 20 percent, not 10 percent; and whether tion, which asks people to insure banks or their regulators should have all of the responsibility for depositors against government policies without any auditing banks, or whether a portion should be guarantee on how those policies would be restricted? shared? In other words, is narrow banking an alter- native or something that allows us to tinker with the Roque Femandez: In Argentina the risk is in the current system? banking system. There is no warranty; there is no lender of last resort. Thus we do not have to transfer Roque Femandez: I am not in favor of revolution, any risk. That is why we must improve the technol- I am in favor of evolution. Narrow banking has its ogy for managing risk. Still, it is impossible to elimi- merits. The market for a given country can provide nate economic risk. The question is, does it make enough liquidity and can replicate the behavior of sense for the government to try to cover the risk of traditional commercial banks. That mechanism can economic activity? be used not to banish commercial banks, but to com- There are also macroeconomic risks that can result plement them. In Argentina and other parts of the from a devaluation brought about by, say, a wrong world people are developing derivatives in the mar- economic policy that introduces a problem in the ket and transforming these instruments. Thus if we banking system. Government attempts to cover that say explicitly that there is no subsidy for commercial risk are misguided. Sometimes the government will operations, there is no lender of last resort, and there use macroeconomic instability as a source of rev- is no free deposit insurance, then the incentives are enue, using the financial system to levy and collect a going to favor an evolution toward narrow banking. tax to finance. So, that is equivalent to saying that the In a sense narrow banking is a superior technol- government will return taxes to the people, because ogy for managing risk and liquidity. I am not propos- they were using the macroeconomic policy to tax the ing, however, that every country produce new people. legislation proposing 100 percent reserves for every- The problem is whether something can be done one and allow the government to issue short-term about a liquidity crisis that has nothing to do with paper to cover the 100 percent reserve. Some pro- economic risk. I think that if there is a derivatives posals in the United States-introduced when there market, the narrow banking proposal can make a were serious crises in the system-have suggested contribution. Although this is not a revolutionary that this be done. Now that those crises are over, contribution, it is an evolutionary one. however, we can think in terms of evolution in the financial system and not react in a hurry to solve Fernando de Santibanies: Allow me to take us financial crises. back to the real world. During the most recent crisis in Argentina deposits went down almost 20 percent Allan Meltzer: Two issues do not seem to have ($8 billion or $9 billion) and there was a lender of last been resolved. First, which banking risks are social resort. Although there are restrictions, the Central and which are private? It may well be that we have Bank can act as a lender of last resort. Although nar- the kind of banking structure we have because we row banking may not be imposed in Argentina, it underwrite the risks through various systems, such looks as if there is an evolution toward it. If that is the as too big to fail and explicit or implicit deposit insur- case, I am afraid that narrow banking will not pro- ance. Throughout history, few banks have failed. So vide the best tools should a similar crisis occur in the who actually bears the risks? future. 44 Floor Discussion of "Does Argentina Provide a Casefor Narrow Banking?" Comment: To build on that comment, consider the banking does not mean you have no risk, it means issue of reserve requirements as a shock absorber. you have minimum risk. There has been some revisionist thinking about reserve requirements, from the early view that they Question: Has Roque Fernandez given any were attacks on financial intermediation to the feel- thought to the idea that narrow banking may be the ing that they could be used as something of a buffer. miracle solution in, say, Paraguay, which is still in the midst of a serious crisis with no solution in sight? Question: What would narrow banking mean to Argentina, and how does the market see risk in Roque Fernandez: I do not believe in miracles. Argentina? The old narrow banking models are Moving from the academic world to the real world, I based on a dollar economy, where assets and liabili- believe in making adjustments to systems that need ties are denominated in the same currency. But in them. In Argentina our system has survived. We do Argentina the market prices one-year peso assets at not know if it is the best system, but we were able to a 5 percent differential to one-year dollar assets. That, manage the crisis. in turn, has created a situation of moral hazard inside I would not argue for radical modifications to any the banking system, with people who have peso system that works reasonably well. Even in assets borrowing in dollars. As a result 70 percent of Paraguay it is important to think in terms of the inno- the banking system has become dollarized. vations in financial markets and to see how they can So if there is to be a narrow banking system in be used to create a more stable financial system. Argentina, might not there be a situation where the What is happening everywhere is that narrow bank- risk is concentrated in narrow banks? During the ing and fractional reserve banking have become tequila crisis money actually ended up in the econ- politicized, so we are almost discussing ideology omy through the capital market rather than through instead of economics. This is what worries me. the banking system. Question: It seems that narrow banking is aimed Roque Fernandez: One of the measures taken at making the system safer and improving confi- during the tequila crisis was to dollarize the reserve dence. In doing so, lower rates will be paid to depos- requirements of banks. Thus any time commercial itors. That might encourage the development of banks bring pesos to the banks for the reserve other institutions not governed by regulators. This requirement, they are converted to U.S. dollars. We has happened in many countries, both developing did not know what was going to happen after the and industrial, and ultimately can undermine confi- tequila devaluation, but we were not going to dence. As a regulator, how would you stop that from devalue. That was a measure of strong commitment happening? If you believe that you should not regu- to the exchange rate policy by government. Of late, how would you educate the public so that they course, other central bank regulations limit the expo- understand the difference between a proper financial sure of commercial banks to exchange rate risks. institution, a proper mutual fund, and one that was Those are the same in Argentina as everywhere else. not properly regulated? So if narrow banking takes away some of the risk in foreign exchange, banks will have to have enough Roque Fernandez: Argentina is just one case capital to cover the risk. Remember, narrow banking where we did not intervene to rescue financial insti- does not mean that there will not be any risk. The tutions. In most other countries a lot of intervention idea is to have enough capital to cover the risk. It prevents financial institutions from failing. could be credit risk, exchange risk, or interest rate Given that it is tough on the regulatory authority, risk. cotntries must be very careful with banking super- In Argentina, for example, there is a different reg- vision and with the development of nonbank insti- ulation for interest rate risk. Suppose that you have tutions. In Argentina we control nonbanks. They are a ten-year mortgage on the asset side and you are not allowed to grow in a disorderly fashion, and they funding it with a thirty-day deposit. If the interest are under supervision of the Central Bank. For exam- rate on the mortgage is a fixed rate, and the rate for ple, General Motors was not granted a license in deposits is a floating rate, then you have an interest Argentina. When it came to Argentina, General rate risk. Commercial banks in Argentina are forced Motors wanted a different kind of license than it had to compute that risk and to have capital to cover it. in the United States. We concluded that this could The same applies for other risks. Hence narrow pose a risk, and turned them down. 45 I Deposit Insurance: Do We Really Need It? Larry A. Sjaastad B anking panics have been part of the financial one deposit reduces the liquidity of all remaining landscape for centuries, but deposit insurance deposits (much as an extra vehicle on a crowded free- as a means to prevent those panics is a more way reduces the speed of all other cars). When a per- recent innovation. In the United States state-run son withdraws money from his or her account, it deposit insurance schemes, often voluntary, were imposes potential costs on all remaining depositors. common by 1920, but federal deposit insurance for The increase in risk to those depositors due to the commercial banks was introduced only in 1933, in reduction in liquidity may be slight, but the cost of response to the catastrophic failure of more than avoiding that risk is also slight: simply withdraw the 9,000 banks. Insurance for savings and loan deposits deposit. But each additional withdrawal carries with did not come until a year later. it the samne (or an even greater) extemality; in the Both deposit insurance and the lender of last absence of a lender of last resort, illiquidity on the resort facility offered by central banks have evolved part of a single bank can spiral into an uncontrolled as devices to stabilize an inherently unstable insti- (and uncontrollable) panic. tution: fractional reserve banking. Fractional reserve Deposit insurance cannot increase the liquidity of banking is an irrational system that came into being the banking system; that is possible only through the by historical accident. In the early days of gold and lender of last resort facility. But deposit insurance silver coinage, money holders required safe places does reduce the tendency toward bank runs because to store their gold and silver coins, so they tumed to depositors know that their money is safe even if they goldsmiths and their secure storage facilities. are last in line to withdraw. Thus deposit insurance Money holders deposited their gold and silver coins does not internalize the externality but it does partly with a goldsmith and received a warehouse receipt. neutralize its effect. Economic theory, however, indi- Over time goldsmiths learned that only a few depos- cates that the first best way of dealing with an exter- itors came to retrieve their coins on any given day, nality is to exorcise it. Papering it over is only second and hence they began lending, at interest, a sizable best. Deposit insurance and the lender of last resort part of their deposits. Thus fractional reserve bank- facility are ways of mitigating the effects of an inher- ing was invented. That it has survived with little ent flaw in fractional reserve banking, but they fail to change in its fundamentals over the centuries has deal with the flaw itself. rendered the system neither more rational nor inher- Fractional reserve banking is not only irrational ently more stable. So long as fractional reserve bank- but is economically inefficient. That is because the ing persists, so will the ever present risk of runs on benefit to a bank of attracting a deposit is propor- banks and with it the need for some facility to reduce tional to the size of the deposit, but the cost of servic- if not eliminate that risk. ing that deposit depends on the volume of transactions associated with it. This source of ineffi- Source of Banking Instability ciency is unavoidable if banks are prohibited from paying interest on deposits, since competition will The instability of fractional reserve banking arises force banks to offer transaction services to depositors from a well-known externality: the withdrawal of free of charge. Complete elimination of fractional Larry A. Sjaastad is professor of economics at the University of Chicago and adjunct professor at the University of Western Australia. 47 Deposit Insurance: Do We Really Need It? reserve banking would result in a more efficient reduction in fires. Rather, it is intended to reduce the financial system and, if properly constituted, a 100 variance of their income or wealth-or both. The percent reserve system could eliminate the need for economic function of insurance is to reduce the lia- both deposit insurance and a lender of last resort. bility of the insured, not the risk of an adverse out- come (or liability). Indeed, because of moral hazard Deposit Insurance and Lender of Last Resort the risk of an adverse outcome is commonly thought to be substantially increased by deposit insurance. Although deposit insurance and the lender of last resort facility of central banks are related and com- The Moral Hazard Problem plementary, they are different. Deposit insurance protects (partly, if there is an effective ceiling) depos- If deposit insurance or guarantees are to be effective, itors from losses resulting from the insolvency of a they must so totally protect depositors from the risk financial institution; indeed, it is only in the case of of loss (perhaps only up to some cap) that depositors insolvency that the insurer has any liability. The have no incentive to monitor the quality of the port- lender of last resort facility, on the other hand, is folio-or even the solvency-of the deposit-taking intended as an antidote for illiquidity rather than institution. This moral hazard associated with insolvency. The effectiveness of the facility, however, deposit insurance is particularly serious because is drastically reduced-and may disappear alto- coverage is usually 100 percent (at least up to a ceil- gether-when the monetary authorities pursue an ing) and, in contrast to other insurable risks (such as exchange rate rule, particularly if the illiquidity is fire, flood, accident, and illness), insolvency of the systemic owing to a flight from the currency (as hap- deposit institution may not inflict the slightest pened in Argentina in early 1995). pain-or even inconvenience-on depositors. There are, of course, various ways of avoiding or The Argentine crisis of 1980 was an instance in stopping runs on banks. During the nineteenth cen- which the design of deposit insurance led to an extra- tury Bank of England tellers were instructed to use ordinarily high degree of moral hazard. In early 1980 small bills and count slowly if there was a run on the Argentine Central Bank was already phasing out deposits. In this century banking "holidays" have deposit insurance when, unexpectedly and for unre- been declared frequently in an attempt to stem runs lated reasons, three major banks faced imnminent col- on banking systems. Yet another measure is higher lapse. To deal with the crisis, deposit insurance was capital and reserve requirements for deposit-taking quickly reinstated and reinforced to cover not only institutions. 100 percent of the principal but all accrued interest. Moreover, there were no delays in collecting Does Deposit Insurance Really Insure? deposits. When an institution (almost always a financiera) failed, which occurred with alarming fre- The main social benefit of deposit insurance is not quency, the Central Bank had the checks prepared that it protects individuals from loss of deposits; from immediately. As a result depositors actively sought a social point of view, that loss is real and inevitable, out institutions that were likely to fail, since they and occurs because of insolvency. In the United States paid the highest interest rates. Something similar during the 1980s failed savings and loan institutions occurred in the United States during the recent sav- recovered their deposits but that in no way reduced ings and loan crisis, when hopelessly insolvent insti- the real social losses due to bad investments. In this tutions advertised rates that more prudent and context deposit insurance merely redistributes losses solvent competitors could not (and did not) meet. to other parties-in the case of the savings and loan The moral hazard problem, then, is inherent to debacle, to U.S. taxpayers. deposit insurance or guarantees, which means that The most frequent cited benefit of deposit insur- close supervision or regulation of the financial indus- ance is that it reduces the risk of a run on banks if one try is essential. The only feasible way to avoid the or a few banks are rumored to be insolvent or even problem is to avoid deposit insurance, but perhaps lacking liquidity. Accordingly, deposit insurance is a even that is futile. The key issue is credibility. Is it misnomer because rather than pooling risks it possible for a government to make a policy of no reduces the risk of bank panics. A better term would insurance or guarantees credible, or will the public be deposit guarantees. perceive that there is an implicit guarantee, at least When, for example, a farmer buys crop insurance for larger institutions? There is, after all, such a thing or a homeowner buys fire insurance, it is not because as the "too big to fail" phenomenon. When a major they expect an improvement in the weather or a institution is on the verge of failing, there is a 48 Larry A. Sjaastad widespread expectation that extraordinary steps will deposit insurance. That would shift the monitoring be taken by the authorities to prevent its collapse for task from depositors (in the absence of deposit fear of a banking panic. That expectation was com- insurance) or regulators (when there is deposit pletely fulfilled, for example, in the 1984 failure of the insurance) to the shareholders and would negate Continental Bank of Chicago. the mnain disadvantage of deposit insurance-moral Chile's experience is also illustrative. After finan- hazard. cial reforms in the second half of the 1970s the num- A third and obvious option is privatization of ber of financial institutions in Chile expanded so deposit insurance. There is no reason insurance com- rapidly that it was quite beyond the ability of the panies should not be able to offer deposit insurance. Superintendencia de los Bancos to adequately super- Privatization should be particularly attractive to vise them. As an altemative the govermment estab- developing countries, where the human and other lished a "no guarantee" policy, one that was capital required to administer state-run deposit enunciated frequently and widely by govermment insurance schemes is often lacking. officials. When the first bank that found itself in trou- Many observers have argued that private insur- ble was bailed out by the govemment, the public ance is not feasible. These arguments, however, are quickly understood that there was an implicit guar- reminiscent of those raised against proposals to pri- antee, and hence not to worry. The intense moral haz- vatize another sacred cow, the postal service. Even ard in Chile during the late 1970s and 1980, together so, private companies in some U.S. states insure with grossly inadequate supervision, played an deposits in state-chartered institutions. That some of important role in the collapse of the financial sector these companies have failed (largely in Ohio and in the economic crisis of the early 1980s. Maryland) has to be viewed in light of the dismal While the too big to fail phenomenon is not nec- monitoring of the Federal Savings and Loan essarily an argument in favor of deposit insurance or Insurance Corporation. The relevant issue is not guarantees, it convincingly supports close supervi- whether private insurers perform perfectly. Rather, it sion of financial institutions. Banking appears to be is whether they perform worse than government one sector where even the most ardent free market agencies. economist might find regulation to be inevitable. A fourth option is to shift the purchase of deposit insurance from the institution to the depositor, which Alternative Ways of Delivering Deposit Insurance would clearly reduce moral hazard. One would expect that firms specializing in portfolio rating In the United States deposit insurance premiums are would immediately spring up. Premiums for deposit paid by the deposit institutions, and those premiums insurance could be expected to be inversely corre- typically are not adjusted for risk. Moreover, again in lated with the quality of the portfolio of the institu- the United States, the moral hazard associated with tion. Hence the extra cost of insurance on deposits in deposit insurance is exacerbated because the insur- high-risk institutions would tend to offset the higher ance is essentially unlimited-despite the $100,000 interest rates that such institutions offered. nominal ceiling-because insolvent institutions are rarely liquidated. Rather, the Federal Deposit Essential Characteristics of Deposit Insurance Insurance Corporation and the Federal Savings and Loan Insurance Corporation absorb the negative net A few key elements of a deposit insurance system, worth and then arrange a merger with a healthy public or private, are essential if it is to reduce the institution. Some moral hazard could be eliminated moral hazard problem. if the ceiling were made effective, either by liquidat- * Deposit insurance should never be available on ing all insolvent institutions or by confiscating all demand unless it is being purchased by deposi- deposit amounts above the ceiling. tors, and then only if certain other conditions are Because deposit insurance is intended to at least met. There should be strict eligibility require- partly eliminate the risk of loss of deposits caused ments for the insured institution, ranging from by insolvency, any measure that reduces the proba- capital and reserve requirements to more mun- bility of that insolvency is a clear substitute. One dane issues, such as the minimum qualifications such measure consists of substantially higher capi- of directors and composition of the board. tal and reserve requirements for deposit-taking * Premiums should be risk rated, and deposit institutions. Another approach is multiple or even insurance should not be withheld if an institu- unlimited liability for shareholders in deposit insti- tion opts for a risky portfolio. The risk of the tutions, which would make a weaker case for portfolio, however, should be reflected in the 49 Deposit Insurance: Do We Really Need It? deposit insurance premium; otherwise the reserves often amount to only a fraction of the money moral hazard problem is exacerbated. supply, when a currency crisis occurs (as in Argentina * Deposit insurance should not be unlimited in the early months of 1995) the central bank faces the because effective ceilings also reduce moral agonizing choice between saving the commercial hazard. Moreover, ceilings can substantially banks and saving itself. If the central bank chooses to lower the cost of deposit insurance without sig- save the commnercial banks, it runs the risk of reserves nificantly reducing effective monitoring by falling below the critical level at which the exchange depositors. rate rule loses its credibility and invites a speculative * If deposit insurance is provided by a govem- attack. If it chooses to save itself, wholesale bankruptcy ment agency, steps must be taken to ensure is not unlikely. What is needed, then, is a second instru- effective regulation and supervision of insured ment that pennits the luxury of a second target. institutions. In this context it would not be Central banks can, of course, hold sufficient inter- unreasonable to impose substantial personal national reserves to defend both an exchange rate rule liability on the managers of the agencies and exercise the lender of last resort function. An inter- charged with supervision. esting example is Chile during the 1995 tequila crisis * Insolvent institutions should be liquidated, that infected some South American countries. For rea- never merged. Liquidation provides a visible sons that have precious little to do with the normally and desirable lesson for would-be imitators. accepted functions of a central bank operating under This, of course, applies only to situations where an exchange rate rule, the Central Bank of Chile had only a few institutions require liquidation. In a accumulated a portfolio of foreign assets somewhat in massive cleanup operation, such as the recent excess of the M2 measure of money supply, thereby U.S. savings and loan debacle, mergers may be eliminating any possibility of a speculative attack on the only practical solution, since liquidation the Chilean peso. This high level of intemational becomes too time consuming. reserves had come into being in response to a massive * To reducemoralhazard, multiple (or evenunlim- issue of bonds by the Central Bank of Chile; since the ited) liability should be imposed on at least the Chilean exchange rate regime is basically an exchange major shareholders in insured institutions, with rate rule, it follows directly from the monetary adequate safeguards to prevent evasion of that approach to the balance of payments that the large but liability. Limited liability is a convention, not dic- negative level of Central Bank domestic assets had to tated by any law of nature or of economics. be balanced by even more foreign assets. Moreover, there are ample precedents for multi- Although this arrangement unwittingly served pie or even unlimited liability. Indeed, Lloyd's of Chile extremely well in the tequila crisis, the annual London is an insurance institution that imposes real resource cost of that insurance was hundreds of unlimited liability on its members. millions of dollars. The bonds issued by the Central e To further reduce moral hazard, clear contin- Bank of Chile carried real yields of up to 9 percent, gency rules should be written into insurance con- while the foreign assets consisted largely of short- tracts that partly or totally void the liability of the term U.S. Treasury paper whose nominal yield was insurer if the insolvent institution is found, even 6-7 percent. As a general defense against currency after the fact, to have knowingly and deliberately substitution in developing countries, this approach engaged in certain proscribed activities (such as is not likely to be endorsed by anyone other than the falsifying information or withholding informa- secretary of treasury of the United States. tion from the monitoring agency). A more efficient second instrument is the combi- nation of a currency board with the so-called Chicago Design for a Panic-Proof System Plan for banking reform. Under that arrangement the central bank is replaced by a currency board that For obvious reasons, most developing countries prefer issues currency in return for foreign exchange at a exchange rate rule over a money supply rule. An fixed exchange rate. Because the currency board is exchange rate rule severely limits the lender of last legally required to hold 100 percent intemational resort facility, particularly under the very circum- reserves to back its issue of domestic currency, a stances when that facility is most needed. In this con- wave of currency substitution of any magnitude text the central bank has but a single instrument could be met with no difficulty and without risk of (international reserves) but two paramount targets: abandoning the fixed exchange rate. maintaining the exchange rate rule and the liquidity of A currency board has a second advantage, at least the financial institutions. Because intemational for some countries prone to inflation-the nominal 50 Larry A. Sjaastad money supply is endogenous and hence inflationary investment banks that, in effect, would issue only finance of government spending becomes impossi- equity. And while the volume of credit provided by ble. The board, however, would be free to define the the financial system might not be seriously affected fixed exchange rate against a single currency or one way or another, commercial banks would no against a basket of currencies (such as the ECU or the longer be a source of credit. This gap, however, SDR), and to choose the composition of the basket to might be filled by an increase in the number of mitigate the undesirable effects of fluctuations in the investment banks, both large and small, and by exchange rates among major currencies. institutions (such as money-market mutual funds) Currency boards, however, have some disadvan- that hold short-term, highly liquid assets and per- tages. In the first place, they involve a loss of seignior- mit equity owners to write checks against their age from currency creation. That, however, is likely to equity. Just how these new institutions would be be a minor consequence because currency holdings structured and function under the Chicago Plan is typically are small relative to gross domestic product. difficult if not impossible to imagine, but the U.S. Moreover, the foreign assets backing the currency experience over the past fifteen years has demon- could be interest bearing. Second, a currency board strated quite convincingly that the financial sector implies a total denial of discretionary monetary policy. is remarkable not for its inability to innovate, but It is less than obvious, however, that the past exercise for its all but unlimited capacity to do so. of monetary policy in developing countries in general, and in Latin America in particular, has yielded positive References net benefits for their constituents. Because the currency board can hold only foreign Baer, Herbert. 1985. "Private Prices, Public Insurance: The assets, it cannot act as the lender of last resort. Financial Pricing of Federal Deposit Insurance." Economic institutions would be strictly on their own. The Perspectives 9(5): 45-57. Federal Reserve Bank of Chicago Plan, however, solves that problem by divid- Chicago, Chicago. ing those institutions into two parts. The first, deposit Calomiris, Charles W. 1992. "Do 'Vulnerable' Economies banks, accepts only sight deposits and, as in the case of Need Deposit Insurance? Lessons From U.S. the currency board, is required to hold 100 percent Agriculture in the 1920s." In Philip L. Brock, ed., If Texas reserves against those deposits. This arrangement not Were Chile: A Primer On Banking Reform. San Francisco, only provides a high degree of stability for the bank- Calif.: Institute for Contemporary Studies. ing system but also negates the need for both deposit Diamond, Douglas W., and Philip H. Dybvig. 1983. "Bank insurance and the intense supervision needed to con- Runs, Deposit Insurance, and Liquidity." Journal of tain the moral hazard created by deposit insurance. Political Economy 91(4): 401-19. Because the Chicago Plan has never been formally England, Catherine. 1989. "A Market Approach to the adopted, the rules concerning the nature of reserves of Savings and Loan Crisis." In Edward H. Crane, and the deposit banks are hypothetical. At one extreme the David Boaz, eds., An American Vision: Policiesfor the '90s. reserves could take the form of currency, in which case Washington, D.C.: Cato Institute. no interest would be earned. At the other extreme the Grossman, Richard S. 1992. "Deposit Insurance, deposit banks might operate like mutual funds, invest- Regulation, and Moral Hazard in the Thrift Industry: ing in highly marketable short-term paper. In an inter- Evidence From the 1930s." American Economic Review 82 mediate case reserves might be invested in short-term (4): 800-21. treasury paper. Although the 100 percent reserve Hart, Albert G. 1935. "The 'Chicago Plan' of Banking requirement on sight deposits would be unlikely to Reform." Review of Economic Studies 2: 104-16. change total wealth holdings, it may alter their com- Scheinkman, Jose A. 1989. "Deposit Guarantee and position. Since the reserves of deposit banks could be Deregulation." Bulletin Mensuel 123 (December). Centre held in the form of short-term interest-bearing assets, D'Information Sur L'Epargne et le Credit, Paris. however, those banks could pay interest on deposits, Simons, Henry. 1934. A Positive Program For Laissez Faire. and hence the substitution away from deposits in favor Chicago: University of Chicago Press. of currency would be minimal. Sjaastad, Larry. 1988. "Exchange Rate Rules for Small The Chicago Plan is not without certain disad- Countries" University of Chicago, Department of vantages that are difficult to assess. Since deposit Economics, Chicago, III. banks could not make loans, the so-called "trans- Smith, Fred L. 1988. "Capping Taxpayer Liability." formation service" function of commercial banks Competitive Enterprise Institute, Washington, D.C. would be eliminated. There would, however, be a White, Lawrence J. 1989. "The Reform of Federal Deposit second set of financial institutions consisting of Insurance." Journal of Economic Perspective 3 (4): 11-29. 51 Comment Paul M. Horvitz G iven the difficulties and limitations of deposit Need for Deposit Insurance insurance, why would any country that does not already have government deposit insur- Even if most depositors were willing to bear the (pre- ance want it? Deposit insurance as a government sumed small) risk of bank failure, there are valid rea- responsibility was once unique to the United States. sons for governments to try to avoid the potentially It has become more popular around the world in disastrous consequences of a systemic wave of bank recent years, and is required for participants in the runs and failures. First is the macroeconomic policy European Union. Some countries have adopted it need to prevent the decline in the money supply that a without a full understanding of its implications. wave of bank failures would cause. Based on the U.S. Even so, with appropriate safeguards, deposit insur- experience, there is little basis for fear of contagion ance can make a positive contribution to the stability effects in banking, but the fear exists. True, appropri- of financial systems. ate action by the central bank can provide needed liq- I cannot accept Larry Sjaastad's assertion that frac- uidity, avoid the need for banks to dump assets at tional reserve banking is an irrational system. It may fire-sale prices in response to runs, and prevent a well be that the "system ... came into being by histor- decline in the money supply. But as Sjaastad points out, ical accident," but it has survived four centuries and deposit insurance is not simply insurance. Its appeal is many financial crises. If the system was totally irra- that it can actually reduce the risk being insured tional, at some point in the past 400 years a better sys- against. tem would have been implemented. There have been Another legitimate reason for governments to many attempts to explain why fractional reserve provide assurances of the safety of banking lies in an banking with demandable debt is logical and efficient. extemality that works in the opposite direction of the Although none is completely convincing, if there were one Sjaastad notes with respect to liquidity. There are a demand for perfectly safe banks, the market would external benefits associated with widespread use of have generated them. In fact, the National Bank Act of the check payments system-for example, there are 1863 provided a sort of narrow banking or "Chicago additional benefits to every user of the payments sys- Plan" by requiring that bank notes be backed 100 per- tem when new users join. But getting full participa- cent by government bonds. At the time bank notes tion in the payments system requires full confidence were the principal bank liability and were perfectly in the safety of deposits. safe. Over time, deposits became more important, and An important attraction of deposit insurance is market forces did not prevent the movement of that, if it prevents bank failure, it appears to be cost- national banks away from narrow banking. less. Similarly, government loan guarantee pro- This evidence from the market leads me to be wary grams appear to provide benefits to borrowers at no of the presumed advantages of narrow banking. The cost to the govemment (if there are no defaults). But benefits of narrow banking might be plausible for a as we have learned in the United States, appearances primitive economy without any financial system. But can be deceiving. In fact, again as stressed by the trauma and difficulty of transition argues against Sjaastad, moral hazard increases the potential cost of such a system for any Latin American country. bank failure. Paul M. Horvitz is Judge James Elkins Professor of Banking and Finance at the University of Houston. 52 Paul M. Horvitz The moral hazard problem is so difficult that crucial is that the regulator be able to take over the Sjaastad concludes that deposit insurance is abad idea, bank before its net worth drops to zero. The FDICIA He notes the difficulty of limiting govermnent expo- sets the level for mandatory intervention at 2 percent. sure when big banks get into difficulty. I agree. But the This is better than zero, but a higher level would be benefits of deposit insurance in reducing financial sys- preferable. Peter Nicholl's paper on New Zealand tem instability are too valuable to lose. More impor- (elsewhere in this volume) describes severe sanctions tant, measures can be taken to minimize the moral if risk-based capital drops below 8 percent. It is when hazard and "too big to fail" problems. The savings and net worth drops to low or negative values that the loan and banking problems of the past fifteen years moral hazard problem of bank management is most may represent one of the few times in the economic his- severe. Insolvent banks have every incentive to take tory of the United States when lessons have been great risks. They gain the upside of such gambles, learned from mistakes. The 1991 Federal Deposit while the downside belongs entirely to the insurer. Insurance Corporation Improvement Act (FDICIA) Incidentally, Sjaastad makes a good point in recom- really was an improvement. Another development is mending double liability for bank stockholders. This the recently legislated preference for depositors over was part of the national banking system in the United other bank creditors. This provides additional protec- States before federal deposit insurance, and acted both tion for depositors (and deposit insurers) but it also as a cushion to protect depositors and a discipline to helps with the moral hazard problem because creditors discourage excessive risk-taking. other than depositors are clearly at risk in case of fail- Sjaastad lacks confidence in the judgment of reg- ure. Thus they have every incentive to be careful in ulators, particularly in dealing with big banks in their decisions to provide credit to banks. trouble. The FDICIA takes the right approach in com- There are probably advantages to fractional ing down strongly on the side of relying on rules reserve banking, but, whether there are or not, no rather than on the discretion of regulators. Although country with a reasonably well-developed banking it is still possible for the U.S. government to bail out sector is going to accept the pain of transition to 100 a failing big bank, there are daunting obstacles in the percent reserve banking because of theoretical way of regulators who want to take such action. advantages put forward sixty years ago. There are I do not see the logic of Sjaastad's recommendation serious problems with government deposit insur- that insolvent institutions always should be liquidated ance as implemented historically in the United and never merged. What I think he is driving at is that States. That experience, as well as theoretical consid- stockholders should be wiped out, and that the man- erations, lead Sjaastad to recommend that govern- agement responsible for bringing down a bank should ment deposit insurance be abandoned. If deposit be removed. But this does not require liquidation. The insurance is to be retained, he would prefer private bank can be taken over and continue in operation, rather than government-sponsored insurance. avoiding disruption to depositors and borrowers, but Private deposit insurance is feasible (Ely 1990; without protecting management or owners. Konstas 1992), but I am not persuaded that it has These considerations also suggest that risk-based important advantages over public deposit insurance. insurance premiums are not essential. In any case the relevant risk is risk of loss to the insurer, not volatil- Improving Deposit Insurance ity of assets or probability of failure. Feasibility of monitoring is more important than other measures of This brings Sjaastad and me to common ground. risk. If asset values can be monitored, then fraud What controls and limitations are necessary to make becomes a more important source of loss than asset government deposit insurance work? The key ele- volatility. But there is no way to assess the probabil- ment is a high capital requirement. If stockholders' ity of fraud. A serious loss to the insurance agency is funds are at risk, and not depositors' or the insurers', not possible with good monitoring and in the then bank management has an incentive to be cau- absence of fraud unless the agency fails to take tious. Whatever the level of capital, it is important prompt corrective action (that is, while net worth is that regulators be able to monitor it with accuracy. still positive). I do not see how insurance premiums Sjaastad correctly emphasizes the importance of can be related to the risk that the insurance agency such monitoring. Effective monitoring requires valu- will fail to meet its responsibilities. As long as capi- ing assets and liabilities at their fair market value tal requirements are too low, however, risk-based rather than at historical cost. premiums make sense. If capital is substantial and can be monitored, then I am intrigued with Sjaastad's suggestion that the the riskiness of the portfolio is less important. What is managers of government insurance agencies should 53 Comment on "Deposit Insurance: Do We Really Need It?" face substantial personal liability for their perfor- complete confidence in, however, is that it will not mance. Kane has written extensively about the prob- be too long until a bank or group of banks does lem that the incentives faced by such managers may something that brings on the crisis that provides be in serious conflict with good public policy. the real test of the reformed insurance system. Personal liability is an approach that merits serious exploration. References Conclusion Calomaris, Charles W., and Charles M. Kahn. 1991. "The Role of Demandable Debt in Structuring Optimal Thus, with appropriate safeguards, government Banking Arrangements." American Economic Review deposit insurance can be the best way to deal with 81(3): 497-513. the inherent fragility of a fractional reserve banking Campbell, Tim S., and William A. Kracaw. 1980. system. The reforms enacted in the FDICIA provide "Information Production, Market Signaling, and the the sort of safeguards that are necessary, though it is Theory of Financial Intermediation." Journal of Finance worth noting that the regulatory agencies have not 35(4): 863-82. carried through on the suggestions in the legislation Diamond, Douglas, W., and Philip H. Dybvig. 1986. for market-value accounting and capital require- "Banking Theory, Deposit Insurance, and Bank ments that consider interest rate risk. Regulation." Journal of Business 59(1): 55-68. Nobody can be sure that the structure in place is Ely, Bert. 1990. "FIRREA: Implications for the U.S. a sound one, because that structure has not yet been Financial System." Paper presented at the annual con- put to a real test. The banking environment since ference on bank structure and competition, Federal the FDICIA has been very favorable (although Reserve Bank of Chicago, Chicago. improved banking stability might be, at least in Kareken, John. 1986. "Federal Bank Regulatory Policy: A part, a result of the FDICIA). Until a banking crisis Description and Some Observations." Journal of arises that puts the current deposit insurance sys- Business 59 (1): 49-54. tem to the test, we cannot have complete confi- Konstas, Panos. 1992. "How to DIG Ourselves out of the dence in what has been created. One thing I do have Insurance Mess." Journal of Retail Banking 14(1): 16-24. 54 Comment Roberto Junguito C olombia established a deposit insurance powers, the questions are what the role of the central scheme in 1985 after a banking crisis that coin- bank should be in dealing with bank failures when the cided with-and was affected by-the Mexican supervisory agency is independent and what degree debade of 1982. Moreover, the policy issues raised by and form of coordination the two agencies should take deposit insurance are also linked to the institutional in dealing with illiquidity or insolvency (or both). structure of Colombia-that is, an independent central Sjaastad refers to the "too big to fail" issue, which is bank (at least since 1991) with a mandate to control the rational expectation that the government will bail inflation and a role as lender of last resort, but where out problem banks. On occasion such bailouts are the financial sector supervision, as well as management of result of pressure from foreign banks, as happened in the insurance scheme, are government responsibilities. Latin America during the debt crisis. Another point Larry A. Sjaastad argues that the roles of providing worth noting is the potential effect of the foreign deposit insurance and acting as lender of last resort exchange regime on the financial perfonnance of local typical of a central bank are complementary. Whereas banks. Opening the capital account, which is part of deposit insurance protects against bank insolvency, the structural adjustment, subjects a country to inflows lender of lastresort function helps overcome bankilliq- and may have a negative impact on banks because of uidity. The complementary character of the two exchange rate risks and direct competition from for- becomes blurred, however, when the lender of last eign banks. To the extent that foreign exchange regimes resort function is a responsibility of the central bank are managed by central banks, this could be another and the deposit insurance scheme is in the hands of the reason to coordinate to prevent conflicts between the government. central bank and the supervisory agency. The central bank must establish clear rules to decide An issue not dealt with by Sjaastad is the appropri- if a bank's problem is illiquidity or insolvency. The ate portfolio choice for the insurance deposit fund. issue becomes one of conflicting interests between the How are depositors' resources best protected: inside or centralbankand the government, eachtrying to decide outside the country? If the decision is made to invest whether a bank's failure (or, more commonly, multiple outside (as done by Argentina), there is an exchange bank failures) is to be covered by monetary or by fiscal rate risk. But the fund's net worth is vulnerable if it is policy. In Colombia the fiscal dimension of bank insol- placed in the domestic financial sector, and could have vency is even more important because many of the an undesirable monetary impact if it is invested in cen- more financially fragile banks are official and have a tral bank paper. Investing in treasury paper seems a capital guarantee extended by the government. good choice, but it depends on the depth of the market. Moreover, illiquidity can lead to insolvency, an issue Pinally, fear of contagion and financial crisis that acquires special significance in countries where depends, above all, on a country's macroeconomic sta- the central bank is independent. bility. In this sense one could venture to answer the Another one of Sjaastad's points is that with deposit question posed in the title of Sjaastad's paper by say- insurance, there is a need for greater supervision and ing that deposit insurance is less necessary when regulation of the financial sector. Under separation of macroeconomic fundamentals are in place. Roberto Junguito is director of the Banco de la Republica in Colombia. 55 Comment Allan H. Meltzer L arry A. Sjaastad has done something that is hard moral hazard-but with a different interpretation. to do. He has written a provocative paper on a The paper argues, correctly, that eliminating deposit subject that has been thoroughly researched insurance is not credible. Such a policy would be and discussed. The paperstarts with abang.Fractional seen as time-inconsistent for either of two reasons. reserve banking, he tells us, has changed little over the When losses occur, pressure from domestic deposi- centuries, is inherently unstable, and is totally irra- tors encourages the government to spread the losses tional. Despite these flaws govermnents have devel- over taxpayers instead of concentrating losses on oped deposit insurance and the lender of last resort depositors. In Uruguay and Chile in the early 1980s function to preserve fractional reserve banking. foreign lenders pressured governments to under- Sjaastad believes that a better solution would be to write banks losses, and the governments did. eliminate fractional reserve banking by requiring 100 Taxpayers may object to paying the bill, as in Japan percent reserves. This change, he claims, would recently, but there are few protests when the govern- remove the system's inherent instability. ment announces that depositors will be paid in full. I am not sure why Sjaastad thinks the system has Sjaastad concludes that even the most ardent free changed little or is irrational. Moreover, proposals market economist might find regulation to be for 100 percent reserves systems have a long history. inevitable. This is correct but misleading. A more One reason they have not been adopted is that complete statement would be that regulation, super- bankers and depositors have incentives to share the vision, and examination are neither necessary nor private benefits of producing more loans and sufficient to prevent banking panics and losses. One deposits from a given amount of reserves. Society reason is that supervision and regulation, typically can, of course, increase reserves at zero cost provided enforced by audits and examinations, often fail to the country is on a fiduciary standard. This could be detect problems in a timely fashion. Moreover, when done by having the central bank issue reserves, either problems are identified, supervisors and regulators as a pure transfer or by buying bonds from banks can be pressured by politicians to engage in for- until all deposits have a 100 percent reserve. For bearance. That is why closures and failures have countries with modem financial systems and low been less common than taxpayer bailouts in many legal reserve requirement ratios, this approach countries. would require a substantial change in the ownership Recognizing the pressures for forbearance, several of debt. The case for debt neutrality, however, simply countries (Chile, New Zealand, the United States) is not strong enough to support such a move. have sought to reform supervision by moving toward market-based regulation. Some countries Changing Approaches to Supervision and have followed the proposals made by Benston and Regulation Kaufman (1988) to intemalize the cost of bank fail- ures by using capital or loan markets to price the risk. Before turning to three issues that are often neglected There is not yet enough experience, however, to in discussions of deposit insurance, I want to reinforce know whether the new arrangements will work as some of the paper's commnents about supervision and intended. Allan H. Meltzer is university professor of political economy at Camegie Mellon University and visiting scholar at the American Enterprise Institute. 56 Allan H. Meltzer One virtue of these plans is their recognition that domestic residents or foreigners concerned about the widespread use of deposit banking and orga- domestic conditions, such as inflation or political nized lending markets has social benefits. Small instability, shift into foreign assets. The larger is the depositors, however, have little incentive to monitor initial reduction in foreign exchange reserves, the their banks and lack the ability to do so. By writing greater a subsequent reduction is likely to be, unless rules that require banks to close before all capital has the authorities act promptly and effectively. Other been lost, market-based regulation tries to maintain factors contribute to the problem, including the fre- the benefits that come from wide use of banking ser- quency of past devaluations, the size of current bud- vices while avoiding some social costs. Chile has get or current account deficits, the populist rhetoric gone further by supplementing the rules that seek to of a newly elected government, or weakness in the close banks before they fail with rules that require financial system. future repayment of any taxpayer funds that may be The second problem requires a lender of last resort used to restore solvency. These arrangements are not even if deposits are backed by 100 percent reserves. perfect, however. Losses may accumulate too The reason is that, in current financial practice, there quickly or governments may engage in forbearance, is a large volume of overdrafts. Thus an institution's in effect waiving the rules. gross purchases or sales, hence its exposure, may be large relative to its cash or net worth. Since most Pitfalls of Narrow Banking transactions are not settled bilaterally but clear at the end of the day (or in the foreign exchange market, at Sjaastad proposes 100 percent reserves, or narrow the end of the second day), there is settlement risk. If banking, as one solution to these weaknesses. a large institution failed, settlements would be dis- Although he does not provide many details, narrow rupted and other institutions might default. banking proposals usually separate deposit-taking In principle the failure to pay at settlement is no from lending. All deposit-taking and payments are different from the failure that occurs when some- made by banks that hold only default-free assets body pays with a check drawn against insufficient (such as short-term treasury bills). Depositors pay a funds. The buyer has the asset purchased. The seller fee equal to the cost of providing payment services has a claim against that asset. If the seller defaults on net of the earnings on treasury bills. All lending is payments during the interval in which the buyer has done by capital market institutions that finance their not discharged the debt, other defaults may occur. activities by selling bonds and equities. Banks and This hypothetical and improbable sequence in the lenders may belong to the same group or holding case of a bad check is more likely to cause systemic company, but services are provided separately and failure in a system with many users of overdrafts that independently. clear only once a day. Even so, three problems can cause banking or This problem currently arises most acutely financial distress. First, 100 percent reserves elimi- because collateral requirements do not fully nate default risk when holders convert domestic remove default risk on foreign exchange transac- deposits into domestic currency, but under a fixed tions, and central banks have been reluctant to exchange rate they do not eliminate the risk of a run underwrite the risk by offering to serve as lenders on foreign exchange reserves if holders of domestic of last resort. By contrast, central banks have currency or assets shift to foreign currency or assets. reduced or eliminated systemic risk on many Second, a 100 percent reserve against deposits does domestic transactions. The Federal Reserve accepts not prevent gross claims from exceeding deposits or the risk on Fed Wire and holds a reserve against a reserves. A bank or its customers may sell one foreign default on the CHIPS network. The Bank of currency and buy another. The net deposit or reserve England requires a securities reserve as collateral position may be unaffected, but a default by one for users of the CHAPS network. Germany and party can cause default elsewhere-the so-called France are moving to central bank guarantees for Herstatt problem. Third, small countries may have to domestic wire transfers. Developing countries choose between insufficient diversification (if the have less technology but they also have fewer financial system specializes in domestic lending) and explicit guarantees. foreign exchange risk (if the financial system seeks to The third problem can arise if there is insufficient diversify its portfolio by investing abroad). diversification domestically. If a dominant domestic The first of the three problems is the easiest to dis- industry in a small country experiences a large nega- cuss, in part because it is familiar from the recent tive shock that causes it to default on loans, many Mexican crisis. Under a fixed exchange rate regime lenders can fail. This problem can be reduced by 57 Comment on "Deposit Insurance: Do We Really Need It?" allowing branches of foreign lenders to compete in the for financial stability. The past two decades have domestic market and by allowing domestic lenders to taught us that financial regulation cannot compen- hold foreign assets or open branches abroad. sate for the effects of large changes in real exchange rates or a cycle of inflation followed by disinflation Policy Options in a Market-Based System or deflation. A welfare maximizing policy would reduce to a min- Some Final Comments imum the risk inherent in market-based systems. Financial firms would be allowed to fail, but sys- We should seek to get rid of deposit insurance. Doing temic failures of the payments and asset transfer sys- so will require institutional adjustments that both get tems would be reduced by the lender of last resort rid of time inconsistency and recognize that small function. The purpose of policy would be to protect depositors cannot be expected to monitor banks. the system, not the financial institutions. In the United States reform is relatively easy to To perform this function without subsidizing risk, design. If restrictions were removed on the number the lender of last resort should announce a penalty or size of checks that can be written on money mar- rate system-the modem equivalent of Bagehot's ket funds that hold only U.S. Treasury bills, there (1873) proposal. The central bank announces the would be a payments system with 100 percent range of collateral against which it lends. Each class reserves and no separate deposit insurance. This of collateral would be accepted at a penalty rate, that system could compete with regulated banks or other is, a discount rate in excess of prevailing market payments systems, including conventional banks rates. Borrowing would be a right. Financial institu- that now rely on the Federal Deposit Insurance tions, whether banks or nonbank financial firms, Corporation Improvement Act (FDICIA), a weak could borrow as long as they could offer acceptable version of the Benston-Kaufman proposal. The pub- collateral. Since a penalty rate is charged, the central lic could choose the preferred system by placing bank would only discount if there was a market deposits in money market funds with 100 percent panic. reserves in treasury bills or banks under FDICIA This proposal seeks to use market pricing of risk without deposit insurance. With changes to to avoid subsidizing or penalizing risk taking. strengthen the rules against taxpayer bailouts, as Electronic transfers would be subject to the same Chile has done, and to permit payments in foreign requirements as any other transfer. The central bank deposits, as in Argentina, the system could serve as would use open market operations to reduce risk of a useful model for many countries. systemic failure in times of distress and would offer discount facilities, at a penalty rate applicable to that References security, to anyone discounting acceptable securities. Chile and New Zealand have supplemented mar- Bagehot, W. 1873. Lombard Street. London: Scribners, ket pricing of risk with rules requiring public disclo- Armstrong. Reprinted in 1962 by Homewood R.D. sure of prospective losses. Bank examiners' ratings Irwin. are published in the press. And in Chile the Benston, G.J. 1973. "Bank Examination." Bulletin of the Superintendent of Banks is prohibited from offering Institute of Finance (May). New York University, forbearance. Graduate School of Business Administration. To encourage market pricing of risk, countries Benston, G.J., and G.G. Kaufman. 1988. "Regulating Bank should allow financial firms to offer deposits in domes- Safety and Performance." In W.S. Haraf and R.M. tic and foreign currencies and permit foreign curren- Kushmeider, eds., Restructuring Banking and Financial cies tobe used in payment, as Argentina has done. The Services in America. Washington, D.C.: American pricing of foreign and domestic deposits would pro- Enterprise Institute. vide useful information about perceived risks. Friedman, M. 1959. A Programfor Monetary Stability. New No system of regulation, supervision, or market York: Fordham. pricing of risks can eliminate risk. Furthermore, Simons, H. 1948. Economic Policyfor a Free Society. Chicago: macroeconomic stability is a necessary condition University of Chicago Press. 58 Floor Discussion Question: We have heard today from U.S. professors Paul Horvitz: The issue is really whether there is about adjustments that should be made to deposit a problem in empowering the government to take insurance schemes but cannot be made because of over a bank that is solvent. In the United States the political pressures. The Argentine experience is inter- ground rules for action say that the insurer must act esting because the economic authorities have a great when a bank has a net worth below 2 percent of cap- deal of political and media influence, and they did ital. So, in effect, the government would be taking what they thought should be done. Are you giving over an institution that is solvent. up too easily in the United States? How does the sit- At first glance banks appear to have been taken uation differ? Our view is that depositors have to go over when capital fell to those levels. In fact, it does through a learning process, and if you continue to not happen often, because banks that are truly sol- bail them out they will continue to exert pressure for vent come up with additional capital. Thus the gov- additional bailouts. emment takeover does not take place, avoiding the problem of taking away private property in a way Allan Meltzer: I agree with the first point. that is unconstitutional or undesirable. Professors should not make political decisions about what should be done. They should say what they Allan Meltzer: The Japanese case is worth dis- think is best, possibly taking into account some polit- cussing. It was not simply a bad set of banking cir- ical realities in trying to make the proposal accept- cumstances, but a bad set of banking circumstances able, and leave politicians to decide what can and coupled with bad macroeconomic policy. That is, the cannot be done. policy in Japan was deflationary. And because such policies hit asset prices first (real estate prices, stock Paul Horvitz: The United States is now in a situa- market prices), their decline made the banking prob- tion where pressure to bail out depositors will prob- lem worse. So, there was no sensible correction for ably not be a problem. With deposit insurance, large the Japanese system until monetary and fiscal poli- banks are unlikely to fail or put depositors in much cies began to turn around. Now that they have, it is danger. There will be losses faced by other creditors possible to get an idea of what the magnitude of the of banks but the political pressure to bail them out problem is, because as long as the monetary policy will not be insurmountable. was deflationary real estate prices were expected to be lower tomorrow than they were today. As a result Question: Since Japan is experiencing banking no one wanted to buy out any real estate today. It problems similar to those in Latin America, I am really took an anticipation of an end to deflation of interested in the discussants' views on the function- asset and output prices to get this problem under ality of deposit insurance corporations-in particu- control. Once the Japanese authorities turned the lar, their ability to seize and dispose of technically macroeconomic problem around, things began to insolvent institutions as quickly as possible. And sec- look a little better in terms of finding a place to sta- ond, how can deposit insurance corporations be bilize the banking system with rising asset prices, equipped to deal with banking problems in addition which is now occurring. to simply liquidating insolvent banks and directly My second point concerns debt overhang, a prob- paying off depositors? lem that was inherited by the current regime. Chile's 59 Floor Discussion of "Deposit Insurance: Do We Really Need It?" approach seems as close to a good solution as one is ure to provide information. But the paper fails to likely to find. Namely, they saved the banks that could specify who the insured is. Surely, the insured is the be saved and wiped out all the stockholders. The gov- depositor and the responsibility for insuring that ernment put in capital, but owners of banks will have adequate information gets disclosed should rest with to pay for it out of future profits. That is what should the insurer. Thus depositors are penalized when be done in Japan. Close the banks that cannot be sal- managers fail to provide information. That may not vaged and salvage the ones that can be saved. be the kind of insurance anyone would want to have. Anyone familiar with insurance and moral hazard George Benston: The U.S. savings and loan insti- problems knows that one straightforward, simple tutions failed predominantly because they were approach to resolving moral hazard is to impose investing in long-term, fixed-rate mortgages using some deductible. Most agree that coverage of liability short-term money. Paul Volcher, then Federal Reserve should be capped at some high value. But what about chairman, allowed interest rates to go up sharply, making sure that every depositor loses something in which destroyed the industry. The savings and loan the event of a crisis? How much could we improve institutions should have been closed down after the pool of depositors if we forced them to have a $500 interests rates came down in 1982. But the Reagan loss every time the system fails? This should not be administration chose not to do so because that would particularly politically difficult to achieve. have jeopardized the tax cut the president wanted. Real estate did not go bad, at least not at first. Comment: It is clear that banking failures are Mortgages went bad. Then the industry was encour- related to systemic problems that are related to macro- aged to grow out of it problems and was allowed to economic problems. Thus there should be separate do so when required capital ratios were lowered. discussions of systemic problems and individual bank Perhaps surprisingly to some, most of the thrifts that problems. This is not done. In addition, many papers made bad loans were not insolvent. Mike Carhill and discuss failures in Latin America, but they rarely dis- I revalued all the savings and loans' financial state- cuss the success stories, such as Panama. ments to market values. We found that the savings Panama introduced financial reforms in 1970 and and loan institutions that made bad investments has had total success. The basis of the system is totally were predominantly solvent and better capitalized. different from everywhere else. First, there is the Most of the insolvent savings and loan institutions strongest commitment to stability in the form of U.S. just stayed put and went deeper into the hole because dollars. Second, many foreign banks operate in of their negative equity. Panama, although there are many local banks. There is The Bush administration addressed the problem no macroeconomic crisis, the interest rate is 3 percent and created the Resolution Trust Corporation (RTC), for both loans and deposits, the system is totally stable, which sold the real estate held by institutions that there is no regulatory process, there are no reserve were taken over by the PSLIC. The RTC correctly did requirements for local banks, and the government not try to hold the real estate until prices went up. does not bail out anybody. Local banks occasionally There were large losses, but they would have been fail. Yet the government does not bail them out. much larger if the RTC had not gotten rid of the pack- Panama's system cannot be improved. Every possible age as quickly as it did. Still, it is best to avoid such objective of macroeconomic policy has been achieved. bailouts in the first place because there are large In Mexico, on the other hand, the authorities losses when any government agency or, for that mat- protect local banks under the assumption that pro- ter, bank takes over real estate. Neither government tection is essential to their development. In addi- agents nor bankers are good at managing property. tion, there is no foreign competition. As a result local banks never became efficient because they did Comment: I would like to pick up on the argu- not have competition. ment favoring multiple liability, which strikes me as Every time a guarantee like insurance is given, it a cop-out. If the premise is that failed banks needed goes against the market behavior of depositors larger amounts of capital, then there is a strong argu- because the system works better when depositors ment for raising the capital adequacy ratio. But why must assume risks. In the absence of insurance, depos- does it have to be recovered from the shareholders of itors will look for the banks that give bankers the incen- a failed institution? tive to behave in a responsible manner. With insurance, A second point that is mentioned in Larry this market mechanism is lost. Thus I agree with the Sjaastad's paper is about a situation where a private proposal suggested-the best approach is to make all insurer limits the insured's liability in case of a fail- depositors lose something when a bank goes under. 60 Floor Discussion of "Deposit Insurance: Do We Really Need It?" Larry Sjaastad: Let me take up the issue of multi- at least two years so that debt holders cannot remove ple liability. Presumably, these shares would be pur- their funds before the bank is closed by the authori- chased voluntarily. There would be no coercion. As ties. The interest rate at which the debt trades in the long as they are purchased voluntarily, there is little market would provide the authorities with the mar- difference between buying shares and buying milk. ket's perception of the bank's risk. When the debt is If people want to buy shares of multiple liability, and refinanced, banks would have to go to the market- they do it voluntarily, there is nothing wrong with it. place; interest paid on that debt would be the same On the issue of capping liability, it is more of an as a variable insurance premium. Furthermore, unin- empirical issue than a logical one. The idea that sured, subordinated debt would not be costly to a everyone should lose something has a certain appeal. bank, except in that it would lose a deposit insurance But it seems to run counter to the whole reason for subsidy. Similar to any other corporation, the bank talking about deposit insurance in the first place, would have debt that is not government insured and because the whole point in having deposit insurance it could deduct interest on the debt for taxable is that we do not need to use it. We hope that it pre- income. Thus only banks that expect to take risks that vents runs. could impose costs on the deposit insurance fund If I expect to lose 10 percent of my deposits then I should oppose holding subordinated debt. will behave in much the same way as if I expect to lose 100 percent. There is not much of a difference. Question: An issue closely related to deposit Monitoring is expensive compared with the cost of insurance is state-owned banks that have, automati- pulling my deposit. So, if I think there is a chance of cally and at no cost to the bank, 100 percent insur- losing 10 percent or even 1 percent, I will be first in ance. After all, commercial banks have to compete line to withdraw my deposits-defeating the objec- with those banks. Are there any observations to be tive of preventing banking panics. As I said, it is an made on this issue, other than the obvious recom- empirical issue. We really do not know how people mendation to privatize the banks? would behave. In Argentina in early 1980 deposit coverage was 90 percent. It failed and was restored Allan Meltzer: Why should the state have a role? to 100 percent. In a well-developed economy the That is, what comparative advantage does the state threat of widely distributed losses could lead to the bring to the banking system? Does it performn some development of private insurance, or co-insurance. function that is not being performed? Is there an That is, the govenmment might insure 90 percent, and extemality available that it can somehow internalize? depositors could go to a private agency for the other When the state gets into the banking business, it 10 percent. creates risk. In every country where the state has played a role in banking it has made loans to favored Paul Horvitz: I think Sjaastad is exactly right on customers for political reasons at below-market the matter of imposing losses on everyone. If the goal interest rates. So, the answer is no. The state should is to eliminate instability in the system, that is exactly not play a role in banking. the wrong way to do it. With respect to private insurance, the problem is Comment: It has been mentioned that the lender that to make things work, insurers would need some of last resort is for liquidity and deposit insurance is control over the closing of the bank. Since they can- for solvency. While this is true in theory, the distinc- not get it, the whole project is unfeasible. tion is blurred. The lender of last resort facility some- Under the U.S. FDIC Improvement Act it may be times helps to improve solvency and deposit plausible for private insurers to enter the picture. Its insurance sometimes helps to improve liquidity. capital requirement provisions may be one way to It was well known in Argentina that if the Central move toward co-insurance. Bank had provided some liquidity to the system through reductions in reserve requirements and George Benston: There already is a way to insti- through real discounts, interest rates would have tute a private insurance scheme-subordinated been higher. As a result more banks would have debentures, debt that is explicitly not insured by gov- become more insolvent. In that case a lender of last ernment. Such debt would have only downside risk; resort facility helped solvency. Likewise, in the mid- unlike equity, there is no prospect for upside gain. dle of the crisis the government adopted deposit This debt should be part of the required capital of insurance, and that again helped keep liquidity in banks, as it would be available to absorb losses, as is banks as they were losing deposits. It was also equity. Subordinated debt should have a maturity of inmportant in improving solvency. 61 Market-Based Banling Regulation Peter Nicholl In the past ten years or so, many countries have Policy interventions can be justified only if they suffered recurrent and extremely costly banking are likely to improve the soundness and efficiency of crises. Latin America and the Caribbean were a system compared with a situation in which there is particularly hard hit, but no region has avoided no intrusion. One problem is that it is almost impos- them. These crises have led to an extensive reexami- sible to find a banking system where there is no inter- nation of the role of banking supervision. vention, so it becomes academic to describe how an New Zealand has gone through such a reexami- intervention-free banking system would look and nation. Unlike many countries, however, it has behave. opted for less regulation and more reliance on mar- The age-old problem of moral hazard that besets ket information and market discipline. By incorpo- the traditional approach to banking supervision rating monitoring in the marketplace by the means that the issue of whether more is better than supervisory authority and by making the responsi- less supervision is not clear-cut. A background bilities of bank directors and management clear, this paper prepared for this conference listed the approach has the potential to be a stronger method bailout costs for banking crises in Latin America of supervision. But how and why did New Zealand and the Caribbean. As that paper said, "the large reach those conclusions? economic and fiscal cost of resolving these crises is put into perspective by the dramatic difference that Should There Be Any Banking Supervision? could have been made had equivalent resources been available for investment in human capital When banking supervision in New Zealand was (health and education), infrastructure, and poverty reviewed in the early 1990s, the initial question was: alleviation." Is there a public policy justification for banking At the very least, there is enough evidence from supervision? Or, put another way, why isn't the core around the world in a variety of regimes to show legal code adequate to regulate banks? that supervision has not prevented bank failures Although it was not unanimous, the majority con- and, because of moral hazard, govemments in clusion of the review was that there was indeed a many countries have met some or all of the costs of public policy justification for treating banks differ- bank failures. The initial response of many govern- ently from other corporate bodies. That conclusion ments to banking crises is to impose more regula- was based on the central role banks play in the econ- tion, and more supervision. Along with this, of omy and the risk that a banking sector problem could course, goes greater moral hazard and greater become systemic. potential financial risk to the government. Banks allocate credit. They play a maturity trans- The Reserve Bank of New Zealand considered formation role. They provide liquidity. They are a big ways to reduce moral hazard and shift more respon- part of the payments system, and so on. But because sibility for the soundness of banks and for moni- banking is strategically important, that does not nec- toring the safety of deposits onto directors and essarily mean that public policy intervention to reg- management. The review concluded that there was a ulate banking is warranted. public policy role because of: Peter Nicholl is executive director for Australia, Cambodia, Kiribati, Republic of Korea, Marshall Islands, Federated States of Micronesia, Mongolia, New Zealand, Papua New Guinea, Solomon Islands, Vanuatu, and Western Samoa at the World Bank. 63 Market-Based Banking Regulation * An information problem. Depositors do not * The Post Bank, also state-owned for more than have unrestricted access to information about 100 years, was sold to an Australian bank. a bank. * The explicit government guarantee of Trustee * The payments system. The failure of a major Saving Bank deposits was removed and a bank may have a crippling effect on the pay- Scottish bank became a minority shareholder. ments system. * The Rural Bank was sold to a New Zealand cor- * The potential withdrawal of credit lines. A credit poration that has since sold it to the National contraction could spread the impact of a major Bank of New Zealand, a wholly owned sub- bank failure quickly and widely throughout the sidiary of Britain's Lloyd's Bank. economy. * The Housing Corporation has been scaled * The freezing of transactions balances. In New down and most of its mortgage portfolio sold Zealand more than 90 percent of the narrow to private banks. money supply is in bank deposits. Thus New Zealand has gone from one of the most to * The possibility of contagion. one of the least regulated financial systems in the developed world. Aims of Banking Supervision Bank Supervision Techniques Having decided that it was not getting out of bank- ing supervision altogether, the Reserve Bank had to New Zealand is using other bank supervision poli- decide on the objectives of its regulatory and super- cies as well. First, there is still a system of bank reg- visory regimes. These were determined to be pro- istration. There is no limit on the number of banks moting and maintaining the soundness and that can be registered, however, and competition is efficiency of the financial system (it does not seek to fostered through an open-door policy. But the door prevent bank failures or protect deposits per se) and, opens both ways-banks can go out as well as come in the event of a bank failure, minimizing damage to in. Some foreign banks that initially came into New the financial system. Zealand have found the market too small and the Although the aim is not to prevent bank failures, established banks too well entrenched. Some have it is hoped that the probability of bank failures will quietly closed down and left. The number of regis- be reduced. The aim is not to protect depositors tered banks peaked at twenty-two and through either, though it is hoped that the probability of mergers, closures, and one failure is now down to fif- depositors losing money is reduced (because there teen. Despite the open door, New Zealand's registra- would be fewer bank failures and depositors would tion policy endeavors to ensure that only institutions be better informed). of appropriate standing with the ability to carry on In addition to these two broad aims, the Reserve business in a prudent manner are registered banks. Bank review also sought to reduce the direct trans- So there is still an entry hurdle. mission mechanisms for systemic problems; use Second, the Reserve Bank of New Zealand still has market incentives wherever possible; maintain an statutory responsibility for maintaining the sound- open, competitive, and flexible banking system; and ness of the banking system. Under normal circum- keep compliance costs down and put all aspects of the stances the Reserve Bank monitors banks using regimne to a "positive net benefits" test (even if this can- public disclosure statements, whereas previously it not be wholly objective). received private prudential returns. The Reserve T'he review of banking supervision took place at Bank still conducts annual consultations with banks the same time as a wide-ranging public sector reform and coordinates with parent bank supervisory that focused on the appropriate role and objectives of authorities where appropriate. There are no (and the state and how these could best be achieved. At have never been) regular on-site inspections in New this point everything has been reviewed, and almost Zealand. everything has been reformed. In 1984 the New Third, minimum capital requirements based on Zealand goverrnment owned or explicitly guaranteed the Basle agreements have been maintained for more than half of the banking system. Since then vir- locally incorporated banks. Although the Reserve tually all state banks have been privatized. Bank believes that disclosure alone should provide * The Bank of New Zealand, under government sufficient incentives for banks to maintain the 8 ownership for 100 years and the country's percent capital-asset norm established in the Basle biggest commercial bank, was sold to an agreements, it enforces this rule because it believes Australian bank. that the capital requirement increases banks' 64 Peter Nicholl international credibility at little, if any, marginal In 1984-85 New Zealand got rid of most regula- cost. The average capital-asset ratio of banks in tions imposed in the name of monetary policy: New Zealand is now close to 11 percent. reserve requirements, interest rate controls, qualita- Fourth, a comprehensive review of the payments tive lending guidelines, and exchange controls. In system is under way aimed partly at reducing risk, addition, the exchange rate was floated. particularly by shifting to real-time settlement for large transactions. This change should greatly reduce Market-Based Elements the impact of a bank failure on the payments system, though it will not eliminate it entirely because some The two main strands to the market-based elements transactions will still be settled on a deferred basis. of New Zealand's supervisory regime are disclosure The central bank will most likely run the real-time and incentives. settlement system, though some commercial banks are disputing that this is a central bank role. Disclosure Fifth, the Reserve Bank will retain a limit on the amount a bank can lend to related parties-that is, The new regime requires all banks to issue public dis- any party controlling or exercising significant influ- closure statements quarterly. Banks are required to ence on a bank. This is because a related party might make these available on request, and to display a coerce a bank to make loans on noncommercial one- or two-page "Key Information Summary" in all terms. Under the new arrangements the limit will be branches. based on a bank's tier one capital rather than total Disclosure requirements are comprehensive and capital (as is now done). Tier one capital is the only include: capital capable of keeping a bank's doors open while * An income statement and balance sheet. absorbing losses. As a corollary, banks must make * Information on the composition of the board of public their exposure to related parties, and directors directors and any conflicts of interest that direc- must sign attestations that the exposure is not con- tors may have. trary to the interests of the bank. * Detailed information on asset quality and Sixth, the Reserve Bank retains extensive crisis provisioning. management powers, including the right to appoint * Information on exposures to individual coun- an investigator, give directives to a bank, and recom- terparties, measured in bands relative to the mend that it be placed under statutory management. bank's equity (that is, the number of exposures The Reserve Bank also has extensive powers for deal- between 10 and 20 percent of a bank's equity, ing with breaches of disclosure requirements and the number between 20 and 30 percent, and so conditions of registration. on). These concentration exposures will be The Reserve Bank sees the power to appoint an taken both at their peak and at the end of the investigator as a substitute for, and a preferred alter- quarter to prevent figures from being tem- native to, regular on-site inspection. It is certainly porarily lowered on the reporting date. cheaper. And it should be more effective. After all, * Information on exposures to related parties. almost half of the U.S. savings and loan institutions * The bank's risk management systems. that got into difficulties in the 1980s had received * Sectoral exposure information. clean bills of health at on-site inspections. * Detailed information on the bank's capital Some other banking regulations will be abolished, adequacy, including its off-balance sheet including limits on banks' exposures to individual exposures. counterparties and on open foreign exchange posi- * Information on market risk exposures, both at tions. The Reserve Bank considers the disclosure their peak and at the end of the quarter. regime to be incentive enough for banks to maintain * Details on whether a bank's obligations are prudent risk positions, making regulatory limits guaranteed and the nature of the guarantee. unnecessary. The Reserve Bank also withdrew its There are also to be disclosure requirements for guidelines on banks' internal controls. It was satis- market risk.1 Banks will have the option of calculating fied that the new disclosure framework and direc- interest rate risk using the Reserve Bank model (based tors' attestations removed the need for such on the Basle market risk model) or using their own guidelines. The bank was also concerned that guide- model, provided it produces a result that is at least as lines could become standard practice, introducing conservative as the Reserve Bank's. Market risk dis- inflexibility and allowing banks to avoid responsi- dosures are for the bank's whole book-that is, the bility for their own judgments. banking book and the trading book-in contrast to the 65 Market-Based Banking Regulation Basle approach, which confines market risk measures agement systems are adequate and are being properly to the trading book. Banks will not be required to hold applied. The Reserve Bank thinks that this requirement capital against market risk exposures. will sharpen the incentives for directors to ensure that Disclosure serves a number of purposes. It is a their bank has appropriate systems to identify, moni- way of reinforcing incentives for bank managers to tor, and manage its business risks. It also reinforces the adopt and maintain prudent risk positions. Some Reserve Bank's desire to ensure that responsibility for bank managers and directors were concerned about management of a bank rests with the directors and not depositors' reactions if they were to publish worri- the banking supervisor. some information. Their concern alone may be a strong enough incentive to avoid that situation. It Response to breaches of capital requirements also illustrates the moral hazard risk that occurs when only the bank and the banking supervisor have A more structured approach to a breach of the mini- access to worrisome information. mum capital ratio requirements also has been Disclosure also should encourage directors to take adopted. This new approach is likely to reduce the ultimate responsibility for the management of their scope for regulatory forbearance by the banking bank and focus attention on risk monitoring and supervisor and therefore to reduce the risks associ- management. Moreover, disclosure should provide ated with such forbearance. If a bank's tier one capi- depositors with better and more timely information tal falls below 4 percent of risk-weighted exposures with which to assess the prudential condition of their or its total capital falls below 8 percent, the bank is bank and to compare it with others. Disclosure will required to submit to the Reserve Bank a plan for lower crown risk by reducing private information in restoring its capital to the minimum required levels. the hands of banking supervisors and giving depos- The bank is also expected to disclose the plan in its itors and others greater responsibility for investment public disclosure statement at the first opportunity. decisions. The plan must include certain elements: Disclosure statements will be subject to external * No distributions can be made to shareholders audit twice a year. Recognizing that external audits until minimum capital requirements are met. can be costly to banks, the half-year audit will be a * The bank's exposure to a related party cannot limited review. increase from the level prevailing when the breach first occurred. Where a reduction in cap- Credit rating disclosure ital results in a bank being in breach of the limit on related party exposure, it must reduce such Banks with a credit rating applicable to long-term exposure to a level that complies with the limit. senior unsecured debt will be required to disclose the * If a bank's tier one capital falls below 3 percent rating prominently in disclosure statements. If a of risk-weighted exposures, gross credit expo- bank has no such rating, that too must be displayed sures must not be increased from the level pre- prominently. This requirement is expected to vailing when the breach first occurred. strengthen market discipline on banks and provide If necessary, the Reserve Bank will use provisions creditors with a simple means of comparing one in the Reserve Bank Act to enforce this policy by giv- bank with another. Initially, ratings were to be ing a direction to the bank. mandatory, but smaller banks argued that this would impose unnecessary costs on them. Conclusion Directors' responsibilities As the background paper for this conference said, "it is still very recent to draw definitive conclusions Bank directors or their agents must sign disclosure from New Zealand's experience with market-based statements as being true and fair. The consequences regulation." Disclosure provisions only took effect of producing a statement that is false or misleading on January 1, 1996. are serious and include fines and imprisonment. There have been mixed reactions to the new Moreover, if creditors lose money as a result of approach. For the most part consumer lobby groups, reliance on a false statement, directors face poten- academics, business commentators, and politicians tially unlimited personal liability. in New Zealand have been supportive. The Reserve In addition to signing the statements, directors must Bank spent much time explaining the new approach make certain attestations. These must state whether to these people, believing that it was important to the directors are satisfied that their bank's risk man- have widespread confidence in the new regime. 66 Peter Nicholl The reaction from New Zealand's bankers and market driven. The market elements are important, overseas supervisors has been more hesitant. but they are a complement to-not a substitute for- Although most banks support the general direction bank supervision. The prudential supervision ele- of reform, some are uncomfortable with the degree ments of the regime are: of transparency in the new disclosure regime. Some * Qualitative entry criteria of their comments and reactions imply that they * Minimum capital ratios think a regime based on market judgments will be * Limits on connected lending tougher than orre where they deal primarily with a * Monitoring by the Reserve Bank, based pri- supervisory authority, marily on publicly disclosed data Some doubts have been raised about the degree of * Scope for investigations responsibility put on directors to ensure that their * Crisis management. banks are sound. In January 1996 an Australian news- The regime aims to make bank directors clearly paper (The Australian) said, "it is hard for independent responsible for the prudential soundness of banks. directors to understand and monitor the risks" and Whereabankfallsbelowtheminimumcapitalrequire- "you'd wonder why anyone would want to be one [a ments, the response by the supervisory authority is bank director]." Both sentiments are probably accurate. predetermined, eliminating regulatory forbearance. But if banking risks are hard to understand and moni- The New Zealand approach clearly will reduce the tor, are bank supervisors in a better position to under- moral hazard problem. But will it improve the pru- stand and monitor than bank directors? Probably not. dential soundness of the banking system? The By assuming they are, supervisors increase moral haz- Reserve Bank believes that it will-and at least as well, ard and potential financial liability for the government. if not better than, conventional banking supervision. Moreover, placing greater responsibility on bank direc- Is the model exportable? Maybe. But at least the tors is consistent with developments elsewhere. For major elements of the model-public disclosure, example, directors of securities- issuinghouses are sub- director responsibility, predetermined responses to ject to similar disciplines under New Zealand's breaches-should be accommodated within other Securities Act. And the 1993 Companies Act strength- countries' bank supervision regimes. ens disciplines on company directors. New Zealand has been criticized for free-riding on References overseas supervisors. That criticism has some valid- ity. Just over 90 percent of the New Zealand banking Archer, David. 1992. "The Rationale for Prudential system is now foreign-owned (about 66 percent by Policy." Paper prepared for the Reserve Bank of New Australian banks). But New Zealand would almost Zealand's Prudential Policy Committee. certainly have proceeded with change even if the Brash, Don. 1992. "Banking Supervision: Defining the domestically owned proportion were greater. The Public Sector Role." Address to the Institute of Policy Reserve Bank believes that the new system will pro- Studies Seminar, August 25, Wellington, New mote soundness and efficiency in its banking system Zealand. while significantly reducing moral hazard. Ledingham, Peter. 1995. "The Review of Bank Supervision Another criticism is that most depositors will Arrangements in New Zealand: The Main Elements of either ignore or misunderstand disclosed informa- the Debate." Paper presented to the OECD Committee tion. The Reserve Bank does not expect most bank on Financial Markets, June 20-21, Paris. depositors to carefully study disclosure statements Mortlock, Geof. 1992a. "Banking Supervision: Monitoring or to base investment judgments on them. But there and Compliance Enforcement." Paper prepared for the will be people that do so (financial advisers, business Reserve Bank of New Zealand's Prudential Policy joumalists, and the like) who will, in tum, convey the Committee. information to depositors. - . 1992b. "Strategic Review of Banking Supervision: It is possible that these observers will misinterpret Alternative Approaches to Meeting Systemic information and start a panic. But that risk also existed Objectives." Paper prepared for the Reserve Bank of in the past. Market rumors cropped up. Banks and the New Zealand's Prudential Policy Committee. authorities had to decide whether to dispel the rumors Reserve Bank of New Zealand. 1995a "Disclosure and, if necessary, support the institution. With more Arrangements for Registered Banks: Reserve Bank's information in the marketplace, there should be less Conclusions." risk of unfounded rumors, not more. -. 1995b. "Review of Banking Supervision: Reserve The New Zealand approach sets minimum pru- Bank's Policy Conclusions." Reserve Bank ofNew Zealand dential standards. It is not completely hands off and Bulletin 58 (2). 67 Market-Based Banking Regulation White, Bruce. 1991. "Banking Supervision Policy Issues." Note Paper prepared for the Reserve Bank of New Zealand's Prudential Policy Committee. 1. It has been difficult to reach agreement with banks - . 1992. "The Basis for, and Objectives of, Banking on the valuation of some of these market risk items. The Supervision." Paper prepared for Reserve Bank of New disclosure regime commenced on January 1, 1996, but the Zealand's Prudential Policy Committee. market risk disclosure was postponed to April 1, 1996. 68 Comment Rolf J. Liuders T here are two kinds of banking crises. One is sion of one or two bank failures to the rest of the related to economic recession, such as that in system.' Chile in the early 1980s, which cost-effective During an extreme economic downturn business regulations probably cannot help avoid. The other cash flow becomes insufficient to service normal debt results from one big bank failure that extends to the levels and asset values fall drastically, often well below rest of the financial system, in which regulation can the debt they were guaranteeing. When that happens make-and has made-a major difference. Market- most financial institutions will face liquidity problems based regulations are likely to be cheaper for gov- and often become insolvent. This is what happened in emments because they tend to reduce moral Chile in the crisis of the early 1980s. It is probably also hazard and agency problems. Market-based sys- the main reason for the financial crises in Argentina tems such as that instituted in New Zealand share (1981), Colombia (1982), Uruguay (1982), and many common features with most prudential reg- Venezuela (1994; Gavin and Hausmann 1996).3 In ulatory systems. Likewise, Chilean prudential reg- other words the primary cause of recent deep financial ulations include many market-based elements. In crises in Latin America has been macroeconomic and practice both systems are strikingly similar, proba- can be avoided in the future only through much stiffer bly because they were designed by pragmatic offi- regulations. But it is often argued that tougher regula- cials who shared a common set of objectives and tions would generate an additional cost in lost eco- economic principles. nomic opportunities that would probably vastly exceed any benefits derived from occasionally less Banking and Financial Crises and the Role of severe financial crises (Gavin and Haussman 1996). Regulators Of course, weak financial institutions aggravate these crises and the cost to governments of bank fail- There is broad agreement that bank failures differ ures. For political reasons governments usually from the collapse of most other firms. In particular, intervene to bail out failed banks. This has prompted bank failures can affect competing banks and the most, if not all, governments to regulate financial payments system. Moreover, moral hazard (which institutions to try to minimize expected losses, both arises because banks mostly lend resources belong- for society and for government. Regulatory improve- ing to others) and information asymmetry (which ments are usually synonymous with more regula- exists because banks do not have all the information tion. As deregulation has become the fad, however, available to borrowers and depositors have little market-based regulation has been advanced as an information about the credit portfolios of banks) are alternative, and one country, New Zealand, has also more important in the financial sector. partly adopted such a system. For societies and governments these characteris- Prudential and market-based regulation have tics of banking have led to costly failures all over sinmilar aims. They reduce systemic risk, that is, that the world. Most countries also have experienced the failure of one institution extends to the rest, and serious financial crises, which tend to be related to limit the bailout costs for government. The market- severe macroeconomic crises and not to the exten- based approach, however, is more concerned with Rolf J. Luiders is professor of economics at the Pontifical Catholic University of Chile and director general of the International Center for Economic Growth. 69 Comment on "Market-Based Banking Regulation" efficiency, both in the effect of regulations on the precisely because deposit insurance exists. Such workings of the entire financial system and with schemes have nevertheless been effective in avoid- respect to the operational costs of regulations to ing the spread of isolated bank failures to the entire financial institutions. Thus any comparison of the financial system, although most specialists would two regulatory systems should be made in terms of argue that this has been achieved at relatively high all three objectives, bearing in mind that bank fail- administrative and resource allocation costs. ures related to deep macroeconomic cycles are probably unavoidable, except at sky-high costs. Market-based regulation Advantages and Disadvantages of the Two The high costs of prudential regulation have Regulatory Systems prompted the search for an alternative. Market- based regulations aim to achieve the objectives of financial regulation by eliminating subsidized Prudential regulation deposit insurance and forcing financial institutions to publicly disclose information that is normally Current regulatory systems are usually prudential, made available only to official regulators. In par- characterized by restrictions on entry, mandatory ticular, depositors as stakeholders would have to credit risk assessments, and minimum risk-based bear bank losses when they exceed net worth. The capital requirements. Regulations are enforced by an authorities would only enforce property rights- official agency and have significant administrative those of banks if clients do not repay loans and costs, both for financial institutions (estimated at those of depositors when a bank does not honor 6-14 percent of noninterest operational costs) and for deposits and has not provided accurate and timely the agency (Jordan 1993). Prudential regulations also information as required by law. have allocative costs, which are difficult to quantify In principle, market-based regulation reduces the but are significant (Jordan 1993, Nicholl 1996). operational costs of banks. It also eliminates the No reliable cost-benefit comparisons exist insurance or bank bailout cost to government, or between administrative and allocative costs and the both. Moreover, it does away with the resource allo- benefits from fewer bank failures derived from reg- cation costs of prudential regulation. Market-based ulation. It is likely that the well-known agency prob- regulation does not require a specialized official lem, in this case produced by a divergence between agency to monitor banks and enforce regulations the agency's purpose and the personal interests of since agents, if they feel so inclined, can take their regulators, often reduces the effectiveness of pru- case to the courts. That is, market-based regulations dential regulation. If so, the system might become are in principle less costly to banks, to governments, cost ineffective. and to society as a whole. In principle they eliminate Prudential regulations are, more often than not, the agency problem and reduce the moral hazard accompanied by government-financed deposit insur- problem because there is no subsidized deposit ance schemes. These have been costly for govern- insurance. But in principle does not mean in practice. ments, although they have effectively reduced If the aim is to achieve the same degree of protec- systemic risk. Moreover, if for macroeconomic reasons tion against bank runs as those offered by reasonably there are generalized bank failures, deposit insurance effective prudential regulations, the amount and has helped avoid a breakdown in the payments sys- quality of information to be provided to the public tem. As before, however, it is difficult to make a reli- will also require substantial extra work for banks. able cost-benefit analysis of such insurance schemes. This is in addition to special auditing efforts, which Theory says that deposit insurance schemes that are expensive and sometimes unreliable, and means are government financed or subsidized exacerbate the agency problem is not necessarily eliminated. If the moral hazard problem of the financial sector. so, it would not be surprising that governments, Banks that roll over bad loans can do so with faced with an unexpected and significant bank fail- impunity, financing them by increasing interest ure likely to spread to the rest of the system, might rates to depositors who, in turn, do not care about be forced to incur bailout costs not too different from the quality of the bank's loan portfolio because of those under prudential regulation. the insurance. This has usually generated political Generalized bank failures are almost certain to demands for more regulation to limit govemment happen in deep recessions, no matter which regula- losses. This implies that part-perhaps a signifi- tory system is in place. Under such circumstances it cant part-of current regulatory costs are incurred is unrealistic to expect a hands-off policy from 70 Rolf I. Luders government. By intervening, governments expect to ratio to operate abroad without having to go contribute to a speedier recovery. In such cases through the time-consuming authorization. bailout costs are unlikely to differ significantly In Chile, as in New Zealand, banks are free to between the two regulatory systems. decide their own loan portfolio (clients, sectors, and so on). In Chile, however, no bank can lend to Market-Based Regulation in New Zealand, any client, person, or firm more than the capital of Prudential Regulation in Chile the bank. This limitation also affects related parties, whose firms are considered for these purposes as In New Zealand it was decided early that the Reserve one person. These regulations are similar to those Bank would still play a role in supervising the finan- in New Zealand, which also restricts credit to cial system-in bank registration and monitoring, in related parties. A significant difference, however, is requiring minimum capital, and in limiting loans to that as a way of limiting market or financial risks, related parties. The main market-based elements banks in Chile must maintain relationships relate to comprehensive public disclosure require- between foreign currency deposits and loans and ments, to bank directors' responsibilities, and to between the term structure of deposits and loans. responses to breaches of capital requirements. New In New Zealand banks are free to choose the level Zealand does not have an explicit deposit insurance of risks they want to assume, but they have to dis- system. close them. In addition to the risk-adjusted mini- By contrast, Chile, heavily influenced by the mum capital requirement, Chilean banks have to effects of the 1982 recession and financial crisis, has set aside a proportion of their loan portfolio as a liq- developed a system of prudential regulation that uidity reserve, a regulation that does not exist in includes market-based elements. In Chile (as in New New Zealand. Zealand) the aim has been to use market incentives These additional regulatory requirements in Chile whenever possible; to maintain an open, competi- are in a sense compensated by a Central Bank guar- tive, and flexible financial system; and to make reg- antee on all sight deposits. This guarantee ensures ulation cost effective. the smooth functioning of the payments system even In Chile and New Zealand access to the banking when the system comes under deep stress. It is also industry requires official authorization. During the compensated by an explicit time deposit insurance of 1970s banking permits in Chile were liberalized and about $4,000, payable only to Chileans and only once the number of financial institutions increased a year for the entire system. The idea is that such rapidly. Since the early 1980s (under basically the insurance eliminates popular pressures to bail out same legislation) no new banks have been autho- banks once they fail. rized, and officials favor mergers of existing banks. Given that the proportion of time deposits belong- Under the pretext of making sure that only prudent ing to Chileans and firms that have more than $4,000 institutions have access to it, this entry restriction has deposited is high and that time deposits exceed sight been defended on several grounds, especially deposits, the potential risks incurred by depositors economies of scale and the need to allow banks to are still significant, although not as much as in New recover from past crises. As a result financial institu- Zealand. Thus regulations in both countries provide tions in Chile enjoy high rates of return. This protec- incentives to banks to be prudent. tion, which is not legally based, should be Public disclosure requirements in New Zealand are abandoned to generate more competition. more comprehensive than in Chile. Even so, among New Zealand and Chile more or less follow the other things, banks in Chile have to disclose publicly Basle capital accord. In New Zealand it is argued and usually monthly income and balance sheet infor- that banks would likely maintain at least as much mation, composition of the board of directors, and capital as the accord requires, but the minimum information about the quality of assets. Although not capital requirement is convenient for attracting a public disclosure, but perhaps more important, and foreign resources. In Chile it is argued that require- as a requirement of its pension system, in Chile inde- ments higher than those of the Basle accord, pendent firms classify bank deposits every month for although desirable for reducing bank failure risks risk. These ratings are reviewed by a risk classification and the moral hazard problem, might mean the conunission that is private but that includes represen- loss of intemational competitiveness. To induce tatives from the Central Bank and some supervisory higher net worth, however, the Chilean govem- agencies. Moreover, the Superintendencia de Bancos ment has submitted to Congress a measure to e Instituciones Financieras (SBIF) makes its loan port- allow banks with a 10 percent risk-adjusted capital folio classification public once a year. 71 Comment on "Market-Based Banking Regulation" In Chile (as in New Zealand) regulation makes common, and both countries approach regulation on commercial bank directors responsible for providing the basis of similar general economic principles. the correct information and, in general, for comply- ing with regulations. In fact, under a proposed legal Notes reform directors of financial institutions in Chile will be presumed by law to always be perfectly informed 1. The crisis of the early 1930s in the United States is of all aspects of the operations of these institutions perhaps the best example of a few bank failures eventually and will be totally responsible for all management spreading to the rest of the system and later generating a decisions. In both countries any false or misleading worldwide depression. information disclosed by a bank has serious conse- 2. The Mexican financial crisis that started in late 1994 quences for directors, including fines and imprison- probably also had an important macroeconomic component. ment. And if depositors lose money, directors may be personally liable for those losses. References In both countries, if banks do not fulfill minimum capital requirements because, for example, they have Gavin M., and R. Hausmann. 1996. "Las Dimensiones incurred significant losses, they will be forced to fol- Macroecon6micas de las Crisis Bancarias." In Politicas low one or more steps to restore solvency. In New de Desarrollo. Washington, D.C.: Inter-American Zealand these banks cannot distribute dividends, Development Bank. cannot increase exposure to related parties, and Jeftanovic, P. 1996. "Dinero y su Institucionalidad en Chile eventually might have to present a plan to increase en el Perlodo Anterior a la Fundaci6n del Banco Central net worth. In Chile owners of banks might put in new de Chile: 1861-1925." Part of a draft chapter of the forth- resources, negotiate the capitalization of deposits, or comning Monetary History of Chile, P. Jeftanovic and R. even get credit from the rest of the banking system. Liiders, Santiago, Chile. T hese are considered part of net worth (for up to two Jordan, J. L. 1993. "A Market-Based Approach to years) to provide incentives for the financial system Regulatory Reform." In Economic Commentary. to bail out its own member banks. If necessary, the Cleveland: Federal Reserve Bank of Cleveland. Reserve Bank of New Zealand has the power to give Liuders, R. 1986. "Auge y Desaparici6n de los Grandes directions to banks to restore solvency In Chile this Conglomerados Chilenos: 1975-1982." Santiago. power is wide and ranges from giving specific direc- Valdes, S. 1992. "Ajuste Estructural en el Mercado de tions to bank managers, taking over management, Capitales: La Evidencia de Chile." In El Modelo and even closing banks down. Econ6mico Chileno, D. Wisecarver, ed. Santiago: Although in New Zealand regulation is called Pontifical Catholic University of Chile and Centro market-based and in Chile is thought of as pruden- Intemacional para el Desarrollo Econ6mico (CINDE). tial, the differences between the two systems are not Wagner, G. 1992. "Competencia y Regulaci6n en el great. Both countries have decided to use a mixture Mercado Chileno." In El Modelo Econ6mico Chileno. D. of elements from both regulatory systems. Thus it is Wisecarver, ed., Santiago: Pontifical Catholic University perhaps not surprising that both regulatory systems of Chile and Centro Internacional para el Desarrollo are similar. The objectives of banking regulation are Econ6mico (CINDE). 72 Comment Jacques Trigo Loubiere S ome 95 percent of the assets in Bolivia's finan- This is not to say that the superintendency does not cial system are managed by thirteen national disseminate information. We publish weekly, monthly, banks; the rest is controlled by four foreign and annual banking information to develop a critical banks. Thus, unlike in New Zealand, foreign banks sense of analysis. But in many cases it is difficult to ver- play a minimal role. Moreover, ownership is concen- ify the quality, commitment, and reliability of infor- trated in a few powerful economic groups. mation. Although monitoring can help improve Concentrated ownership can distort market out- banking system stability, there are problems and defi- comes and makes appropriate supervision based on ciencies in accounting registries, induding double the market more difficult. counting, high-risk hidden operations through other Since 1995 Bolivia's Superintendency of Banks bank accounts, inappropriate provisions for unreal- and Financial Entities has focused on improving the ized income or impaired assets, and so on. capital base and technology of the banking sector. The superintendency has created a more favorable External Audits environment for banks and has opened up the sector to intemational investment. Citibank is negotiating As with other supervision models, the market-based to become partners with a local bank (BHN- approach depends on external audits, though to a Multibanco), and the Andean Foment Corporation greater degree. Bolivia requires that banks undergo and the International Finance Corporation are part- two supervisions and two opinions each year. These ners with another local bank, Industrial Bank. Two audits allow problems that weaken the credit stand- other local banks, Union Bank and BBA, have joined ing of banks to be detected by questioning the effec- capital with Chilean banks. These efforts are tiveness of the objectives and approaches they use. In expected to change the Bolivian banking sector from many countries, however, banks that relied on the a familiar to a corporate system. If this happens, mar- opinions of prestigious auditing firms still had to be ket supervision will become more feasible. liquidated. Another aspect that determines the contribution Spreading Information of extemal audits to banking supervision is the abil- ity that authorities have to establish goals for audit- Such a transition can create problems-particularly ing. Effective auditing depends on this authority and in the dissemination of information, which is essen- on the degree of responsibility held by and sanctions tial for market adjustment. Bolivia has no institutions on external auditing firms. Without these aspects the or public agencies to monitor the performance of the market-based model would have difficulties. banking sector. Moreover, there are no penalties for journalists who misrepresent the financial health of Responsibilities of Managers and Directors banks. Another problem is the political mishandling of information. At times congressmen have made Effective supervision requires that managers and their personal opinions about specific banks known directors be held accountable for their actions. for purely political purposes. Developing sanctions for managers and directors is Jacques Trigo Loubiere is superintendent of banks and financial entities at the Superintendency of Finance in Bolivia. 73 Comment on "Market-Based Banking Regulation" an essential element of the self-regulation that is Conclusion being implemented by the Superintendency of Banks and Financial Entities. Making this change will take New Zealand's banking supervision includes a time, however, because of the institutional culture in number of elements that are based on the market; a concentrated system that lacks adequate human information activities, directors' responsibilities, resources. Moreover, bankers must learn to respond and predetermined answers. Some of the norms of to professional principles and not to personal inter- the New Zealand supervision model are similar to ests or economically powerful groups. The superin- those in the Bolivian model. Supervision in Bolivia, tendency has helped control negligence by initiating however, assigns an important role to on-site criminal proceedings against directors and man- inspections, which do not happen in New Zealand. agers that were involved in fraud and malpractice. Inspections are essential in Bolivia because they At the same time the superintendency is removing its track accounting irregularities that would other- intemal and manual controls by leaving approval wise be difficult to identify. responsibilities and obligations to the directors of Another difference between the two systems is financial firms. that New Zealand's central bank has extensive inter- vention rights. Bolivia's Superintendency of Banks Predetermined Answers and Financial Entities intervenes only to liquidate. A market-based system assumes that the informa- Like New Zealand, Bolivia has established predeter- tion that has been disseminated has been properly mined answers to breaches of minimum prudential treated. But problems exist in many countries: procedures, essentially imposing capital sufficiency * Depositors are not educated enough to perform requirements. In addition, the superintendency is analysis that allows them to make decisions. lifting some of its internal and manual controls, leav- * Private economic and financial institutions ing approval responsibilities and obligations to the with sufficient capacity to evaluate the banking directors of financial entities. system do not exist. - The media do not have journalists and Crisis and Deposit Insurance reporters with the specialized skills or knowl- edge required to disseminate information One difference between New Zealand and Bolivia's about banks. approach is the broad influence that the New Zealand The role that directors and executives play in a Reserve Bank has in a crisis. Unlike the Superinten- market-based system assumes that the country has dency of Banks and Financial Entities, it can intervene an enforcement capacity such that any false declara- only in involuntary settlements of financial firms. tion from a director would result in fines and other It is more feasible to provide market-based penalties. But since judicial systems in most Latin supervision of banks when private deposit insur- American countries are weak, it is unlikely that such ance mechanisms exist and are another element on penalties would be enforced. Bolivia's supervision which depositors can base their analysis of banks. model is moving toward self-regulation, giving By contrast, state insurance for bank deposits greater responsibilities to the directors and managers irnpedes the market approach. It is important, how- of financial firms. This is a gradual process, however, ever, to remember that state-owned insurance can to which the financial system and the depositing help reestablish depositors' confidence after a public must adjust. This model is easy to apply to sta- financial crisis and help avoid massive deposit ble and consolidated financial systems but presents withdrawals that would put the stability of the greater challenges where systems are stil making financial system at risk, not to mention a country's adjustments and where stability is weak and vulner- entire economy. able to internal and external factors. 74 Comment Paul L. Bydalek In January 1996 New Zealand implemented a How does New Zealand's system compare with new system of market-based regulation for its Brazil's? And, more specifically, how might it have banking system. Under this system regulation been applied in the cases of the three problem banks? essentially has been privatized. Each bank is required to make quarterly disclosures to the mar- Objectives and Similarities ket of the information it deems sufficient for sophisticated clients-including the central bank- Macroeconomic problems in the late 1980s and to make sound judgments about it. This remark- early 1990s forced a rethinking of the role of the able, free market approach to regulation is public sector in New Zealand. Even there, the core provocative, particularly when juxtaposed with legal code was judged inadequate for regulating the Latin American tendency to overregulate mar- banking. kets. It is easy to dismiss as irrelevant what hap- Similarly, Brazil is rethinking the government's pens in New Zealand, a country of 3.5 million role in the economy. Since the attainment of eco- people and GNP of $56 billion, where foreign- nomic stability in July 1994, four of Brazil's ten owned institutions hold 90 percent of banking largest commercial banks have undergone Central assets. Brazil, with its 155 million inhabitants, has Bank intervention, restructuring, or closure. As in a GNP of $677 billion, and more than 90 percent of New Zealand, the authorities have worked to pro- banking assets are owned by local investors. mote soundness and efficiency in the banking sys- Obviously, vast cultural differences exist as well. tem, reducing damage to it and to the economy. The Nonetheless, there are similarities and lessons to be authorities have admitted that past regulation was learned. Good business practices transcend size deficient. At the same time Congress is approving a and culture. revamp of the public sector, encouraging privatiza- The changes in New Zealand's banking system tion and lowering entry barriers to private investors. provoke thoughts about three major crises in In addition, the duties of external auditors recently Brazilian banking: the Central Bank intervention at were redefined and tightened. the end of 1994 in Brazil's second-largest commercial New Zealand focused on reducing moral hazard bank, the state-controlled Banco do Estado de Sao by downplaying overt control by authorities and Paulo (or Banespa), and failures in the second half of placing responsibility for proper conduct on bank 1995 of two of the largest private banks, Banco directors, managers, and depositors. No new direct Economico and Banco Nacional. These three crises responsibility was given to shareholders, perhaps in are each unique and should be the test for any an effort to acknowledge that shareholders should changes contemplated in the Brazilian regulatory suffer the automatic consequences of poor steward- system. To put these problems into perspective, the ship by the executives they designate. last audited, consolidated total assets for each bank In Brazil voting control of most banks is held by a were Banespa, $28.4 billion (September 1994), Banco few individuals or families that are active in manage- Economico, $8.7 billion (June 1995), and Banco ment. Ownership and management are frequently Nacional, $13.1 billion (September 1995). indistinguishable, even for banks listed on stock Paul L. Bydalek is president of Atlantic Rating, located in Brazil. 75 Comment on "Market-Based Banking Regulation" exchanges and having substantial minority participa- publicly owned and listed banks (and companies) tion. In Brazil responsibility for distress should be publish quarterly financial statements and provide extended to controlling shareholders, who should bear information on asset quality and provisioning. Yet the brunt of sanctions for improprieties. For publicly most banks in Brazil are privately owned and so are owned and listed banks, noncontrolling shareholders exempt. The main regulatory disclosures are half-year (who are usually a passive minority) should be exempt audited statements. from sanctions. As with any other investment, how- Twice each year all banks undergo full audits- bal- ever, they would still suffer from any erosion in share ance sheet, income statement, reconciliation of net values caused by bad management. worth, and a source and application of funds state- ment, with notes and explanations and the auditor's Regulation Techniques opinion. These audits are published in local newspa- pers, usually within two or three months. Unlike in Some of New Zealand's regulatory techniques seem New Zealand, there is no requirement to report con- curious. There are no on-site investigations; only fficts of interest by board members, asset concentra- information disclosed quarterly to the general pub- tions, sectoral exposures, or peak market risk lic is used by regulators. Limitations on loans to exposures. Auditors and banks meet the requirements related parties and individual counterparties and on set by regulators; few offer data beyond that. foreign exchange positions have been abandoned. To In New Zealand quarterly disclosures bear attes- understand a bank's operations, discussions with tations by bank directors to the accuracy of disclo- management are essential. sure and to the sufficiency of risk management There may be no need for regulators to visit banks. systems. False or misleading data can mean fines and If regulators are as agile as bankers in understanding imprisonment for directors and personal liability for new products and procedures and are as technologi- losses by creditors. In Brazil published statements cally proficient, electronic and statistical approaches bear the names of the auditing firm and the bank can probably be devised to identify issues and to check accountant responsible for the figures. Directors do problems. Government employees everywhere, how- not attest to the accuracy of disclosure or adequacy ever, are less nimble than their private sector counter- of risk management systems. Presumably all direc- parts and almost inevitably are less well equipped. tors are equally responsible. Yet when two large pri- This regulatory approach would be inconceiv- vate banks closed in 1995, only the signatories of able in Brazil. The complexities of Brazilian published statements were held liable for faulty finance, the multitude of regulators overseeing a reporting. The exact liability assumed by these sig- bank's activities, and the almost daily changes in natories is arguable and will be tested in legal pro- regulations mean that regulatory reporting is volu- ceedings. Disclosure in Brazil is broadening, mostly minous and cumbersome. An outsider has little because of regulation and market pressures. At times chance of obtaining banks' data, much less under- disclosure is greater than in New Zealand. At the end standing it unless specially trained. Removing lim- of 1995, for instance, Brazil's second largest bank, itations on loans to related parties and Itau, and its third largest, Bamerindus, provided data counterparties and on foreign exchange positions on restructured loans, paving the way for better and is also alien to Brazil, where regulation is tight and voluntary disclosure. In related developments over detailed. This approach may date from times when the past two years commercial banks have been poor communications and geographic distance obliged to publish more meaningful income state- forced the government to issue detailed regulations ments to consolidate their overseas branches on a in order to develop a homogeneous banking sys- line-by-line basis instead of reporting only net equity, tem. Such excessive regulation will change only to disclose the ratio of capital to risk-adjusted assets reluctantly and slowly. using standards more rigorous than those of Basle, and to reveal swap and derivative exposures, includ- Disclosure ing the net positions in risk-adjusted asset calcula- tions. Off-balance sheet items are included in these In Brazil banks are regulated by several government calculations. agencies, as well as by the Central Bank. These agen- cies include the Securities Commission, the Insurance Credit Ratings Commission, the Pension Fund Comniission, the tax authorities, and others. Conflicts exist among these Unlike New Zealand, there are no long-term senior regulators. The Securities Commnission requires that all unsecured debt ratings for any Brazilian company or 76 Paul L. Bydalek bank. Since Brazil is a notch or two below investrnent impossible for executives to be aware of all the risks grade, no domestic company that does not have for- assumed throughout the bank. eign backing can secure an intemational debt rating. In a family-owned and -managed bank of reason- There are local currency ratings, however. able size, professional directors and managers will Ratings were virtually unknown until Bradesco have scant knowledge about dealings elsewhere in the and Itau published their short-term local currency organization. To hold all directors and executive man- ratings in 1993. Now about 20 of Brazil's 265 banks agers responsible for corporate actions, as New hold short-term, intracountry ratings issued by Zealand apparently does, defies imagination and Atlantic Rating and BankWatch. In 1995 Moody's seems impractical. Tremendous transformations issued its first financial strength indicators for would have to occur within Brazilian organizations for Brazilian banks with outstanding euronotes. Their directors to feel comfortable with such responsibilities. ratings on three local banks in difficulty received More focused responsibilities and consequences seem much publicity, causing all of Brazil to become aware preferable. Holding the marketing director responsible of the idea, value, and responsibility of rating. for the actions of the foreign exchange trader or the Local capital markets, however, are still uncertain derivatives desk seems impractical. about ratings. One function of rating is to uncover To work, laws holding directors and managers and promote unrecognized quality among less visi- accountable must function smoothly. Despite some ble banks. Thus high-quality banks wonder if disclo- twenty-five bank failures in Brazil since mid-1994, sure and the effort to obtain a rating are worthwhile. only two bank directors have been held accountable Quality debt issuers and mid-size banks are finding for their actions and arrested. In most cases this kind that a rating helps access new sources of funding, at of white-collar crime goes unchecked for years. better rates and at a time of capital market uncer- Punishment usually comes in form of nominal fines tainty caused by the spillover effects from recent and a ban on assuming responsibilities as a director bank failures. elsewhere. Breaches of Capital Adequacy Benefits to New Zealand New Zealand has clear rules on the consequences of Disclosure in New Zealand should improve the qual- exceeding capital adequacy ratios. Brazil, does not, ity of the financial system. Banks will disclose abun- and problems are dealt with quietly, on a case-by- dant information so that directors and managers can case basis. Until risk-based capital-asset ratios were claim later, in case of problems, that they are innocent introduced in 1995, the basic requirement was for of mnisrepresentations. Professionals with intimate third party funding to be up to fifteen times adjusted knowledge of local capital markets and bank reporting net worth. In 1993 and 1994 this ratio was chronically willbegin to interpret data and to render opinions. One overshot by a few banks. To suspend dividends until clear advantage of the New Zealand system is the capital ratios strengthened, as in New Zealand, stream of quarterly data and performance interpreta- would have pit Brazil's Central Bank regulators tion. Assuming that problems are readily disclosed by against Securities Commission regulators. Capital local accounting practices and that bank management requirements for conglomerate banks' operating does not conceal difficulties, observers gradually will insurance companies or pension funds can vary, become aware of deviations from averages. Festering depending on the regulator. problems will become more visible and surprises less The equity of Brazilian banks is now being weak- likely as quarterly disclosures proceed. ened by a dispute between bankers and the tax Full disclosure by Brazilian banks as called for in authorities. Bankers want to provision for identified New Zealand would have reduced the size of the loan losses, but the authorities refuse to recognize the problems at Banespa, Economico, and Nacional, new provisioning as tax deductible. which were ignored by the market. Any early warn- ing system would have helped. For years Banespa Market Reaction to New Zealand's Change operated on negative cash flow, lending prinarily to the state of Sao Paulo and its related companies. New Zealand's shift of responsibility for compli- Interest and principal were rarely paid; instead they ance and disclosure to bank directors and manage- were rolled over into new loans at better rates. ment, putting their personal reputations and wealth Growing funding expenses, salaries, bonuses, divi- in jeopardy, is understandably disturbing. In a dends, and taxes were paid from new funding. large, segmented institution it is difficult or even Problem assets at Banespa are estimated at more than 77 Comment on "Market-Based Banking Regulation" $15 billion. The Central Bank's intervention and the reaction of the marketplace will be the litmus test. assumption of day-to-day management at the end of Several quarters of reporting will be needed to prove 1994 surprised the market. The size and complexity of the validity of the approach, as will a banking crisis the problems that surfaced were an even bigger sur- (such as a major and unexpected asset quality prob- prise. The closing of Banco Economico in August 1995 lem) or a steep recession. With 90 percent of banking was foreshadowed by years of dubious financial state- assets held by international banks, local problems ments. Much of the current apprehension about may easily be absorbed. Brazilian banks stems from the unexpected failure of Some broad recommendations can be made for the country's fourth-largest private bank, Banco government-controlled and private banks in New Nacional, in November 1995. Nothing had suggested Zealand, Brazil, and elsewhere. All should supply that a large part of its loan portfolio was fictitious-yet the marketplace with quarterly information on the massive fraud had persisted for years. quality of assets, funding, and profits. The degree of The New Zealand system requires management to disclosure should vary with local necessities. confess problems, with stringent legal penalties for Stockholders, directors, and managers should noncompliance if the marketplace is deceived. In this assume full responsibility for the accuracy of disclo- way the system protects against fraud. By revealing a sure, suffering severe and immediate penalties for gradually deteriorating situation to the marketplace, noncompliance. External auditors should validate the system puts pressure on banks to cure their prob- the data disclosed and other professionals should lems. Presuming no catastrophic losses, problems issue reports on quality and risk. Rating agencies should appear slowly, permitting the public to shift should analyze data and issue reports of credit risk business elsewhere without panic. A major industrial changes and ratings for each institution. bankruptcy suddenly pressuring the banking system Government regulators worldwide cannot keep would, however, still require the central bank to pro- pace with changes in capital markets and in technol- vide liquidity to the system. ogy. They must rely increasingly on self-discipline by banks. There must be stiff penalties for noncompliance. Lessons Authorities should privatize as much as possible mon- itoring of capital markets, placing responsibility on So what conclusions can be drawn? Time will tell in trained professionals. Regulators from various agen- New Zealand. The first quarterly disclosure reports cies that are each responsible for part of a bank's oper- will be available soon. The quality of disclosure and ation should be combined into one agency. 78 Floor Discussion Question: Most New Zealand banks are branches of envision the exchange of information being incorpo- foreign banks. Do you think depositors view this as rated into the supervision and regulation of banks in implicit insurance, and so do not have an interest in New Zealand? monitoring banks? After all, reputation is an asset, and if a foreign bank fails it will probably be bailed Question: How easy is it to check the authenticity out by its main house abroad. Do you agree? Has the of the publicly disclosed information and how deep new system improved depositors' ability to monitor can market participants drill to verify the informa- banks? In particular, do differences in the interest tion? In other words, does a participant in the mar- rates paid by banks reflect the risk position of banks ket have a right to demand more information to better now than before? confirm what has been disclosed? Question: Also on the subject of supervision and Peter Nicholl: Most New Zealand institutions are monitoring, I think that it is important to realize that foreign-owned and most are set up as subsidiaries. the comments that have been made about disclosure Only one operates as a branch. Thus most depositors requirements, directors' responsibilities, and prede- probably think that foreign-owned banks fall within termined actions on the part of supervisors are fun- the too big to fail category because the New Zealand damental points that all supervisors would embrace. part of their operation, for the biggest of them, rep- But Mr. Nicholl implied that supervision may actu- resents 10-20 percent of the entire bank. In terms of ally create a moral hazard. In my view it is the interest rates, it is too early to say how depositors absence of an effective bank supervisory function behave in monitoring or discriminating on that basis. that poses a threat to the stability of a banking sys- I do believe that supervision creates part of the tem. That has certainly been the case in many Latin moral hazard-that, as soon as a government sets up American countries. One element of the U.S. sys- a system of supervision and appoints a supervisor, at tem-at least at the federal level-is for well-trained, the very least there is a moral hazard for the govern- specialized bank examiners to actually evaluate, on ment. So, if the bank gets into difficulties, the depos- site, the condition of banks. Examiners are also itors will say: what were you doing and why didn't expected to use moral suasion to influence changes you warn us? To ensure that information is reliable, in an institution so that corrective action can be taken we have tried to set up a structure that gives strong at an early stage. incentives for people other than supervisors to do How does the central bank in New Zealand quality assessment. The main ones are the managers ensure that there is compliance with disclosure and directors of the bank. Because they are held per- requirements, that there is an adequate evaluation of sonally liable for false or wrong information, they the balance sheet and, in particular, of assets? That is, have much stronger incentives to do that job well how does it evaluate the accuracy of the information than supervisors. that is being disclosed? And what relationships are There is a worrisome aspect that we are still trying being established with foreign bank regulators, par- to sort out with the Australian authorities. Their leg- ticularly the Australian authorities, and how do you islation says that if an Australian bank fails, assets 79 Floor Discussion of "Market-Based Banking Regulation" will be used first to pay off depositors in Australia. It ity. In addition, on-site inspection is practiced. This is unclear what that stipulation means for depositors allows the supervisory authority to strictly control in New Zealand because it has not been tested. the risks assumed by the banking system. Chile's Depositors do not seem too concerned. It is ironic banking system has only recently been opened up to that there is more cooperation and consultation with other financial business, so it is difficult to gauge the the Australian authorities today than there was benefits of consolidated supervision. before we adopted this regime. That is partly because the Australians are paying more attention to the New Question: What exactly does a market-based sys- Zealand part of their banks. We never used to have tem add? Auditors are still required. Filings of regular meetings. We would get in touch only if a audited accounts with the supervisors are still problem arose. Now our supervisors meet every six required. Banks do not file a balance sheet that says months. they are bankrupt, if they are. Would they convey information to us if they So the banks are basically giving the public the believed an Australian bank was getting into diffi- same information they were giving to the supervi- culties? We do not know because it has not been sors. The problem develops when people want to tested. But we believe they would. However, at one challenge the information that has been provided. of the world banking supervisor conferences people There are three areas where the problem is localized: were talking about such cooperation, and most said portfolio classification, portfolio provisioning, and that if a U.S. bank operating in a foreign country were accrued interest. How does a system that focuses on getting into difficulty, they were not certain that the public disclosure improve the ability of auditors or U.S. authorities would make that information known anybody else to assess these things? to overseas supervisors. How far can market participants go beyond what Question: In Jamaica we are developing a deposit is in the public disclosure document? It will become insurance scheme that will be put in place sometime a question of how much pressure depositors, or more next year. Could Mr. Nicholl briefly sketch the type likely their agents, can put on a bank they have ques- of scheme he had in mind? tions about. As one example, we have asked banks to disclose the concentration ratios of the borrowers Peter Nicholl: We never came up with a final they have in various categories. They do not have to design for a deposit insurance scheme in New name them, they just have to list them. So, say they Zealand. But the only justification for deposit insur- have four borrowers that are exposed to more than ance in an economy like New Zealand's is the so- 25 percent of their capital. I suspect that some people called small savers justification. Even with all the in the marketplace will want to know who those four information in the marketplace, small savers cannot are, creating market pressure for the names to be dis- (or will not) make judgments, and the impact of the closed. But it will be something for the market to sort crisis on them is more severe than it is on large depos- out. They have no right to know that information. itors. So we would have set an upper limit. It would not have been across-the-board deposit insurance. Question: Rolf, do you believe that supervision We were also attracted to the British idea-co- creates moral hazard? Also, do you feel that the insurance, where you only insure 75-80 percent of absence of consolidated supervision has been a prob- deposits. Even those depositors have an incentive to lem in Chile? keep an eye on their bank and not to move deposits to the riskiest bank because they could lose 20 per- Rolf Luders: Supervision is, as a rule, a reaction to cent of their deposits. Although we liked the idea of the increase in moral hazard generated by deposit co-insurance, it would have been funded by the insurance. But supervision in itself probably checks banks who, in turn, would have passed the fees onto moral hazard. The final outcome is unknown and depositors. probably varies from case to case depending on the effectiveness of supervision. In Chile after the early Question: Market-based regulations assume that 1980s supervision is likely to have reduced moral markets work. That means that market-based regu- hazard, at a cost of course. lation will be no better than the market. Here there One element that distinguishes Chile's supervi- are two important elements. One is regulation itself. sory system from those of most countries, especially Laws make markets work but people must also New Zealand's, is that the information system of the accept the rules of the game. This should be a warn- banks is shared online with the supervisory author- ing against installing market-based regulation in 80 Floor Discussion of "Market-Based Banking Regulation" countries where the market does not work. Also, tors were Japanese, not small New Zealand depos- governments should be aware that once they have itors. Even so, I am confident that the government market-based regulation, they cannot make any will not step in to bail out depositors. false steps. Any false step is a false signal and peo- ple will simply think that they do not have market- Question: Mr. Nicholl said something to the effect based regulations and that the government will that inspectors had been replaced by investigators. save them. Do the people in New Zealand really What is the role of these investigators? What is their think that the government is not going to come and scope and to what extent are they replacing inspec- save them next time there is a problem in the bank- tors on a selective basis? With respect to external ing system? auditors, what is their liability when they fail? Peter Nicholl: We do not know whether people Question: Mr. Nicholl, as a central banker in New accept that they must look out for themselves, but the Zealand you have privileged information about policy that has been put in place is not independent what is going on in commercial banks through mon- of other things New Zealand's government has done. etary policies or through bank accounts that you The government has corporatized most state-owned hold for commercial banks. If you had an inkling that trading enterprises. They have been set up with their something dangerous were happening within a com- own balance sheets, capital, and boards of directors, mercial bank, would you step in and ring the alarm, and people that make loans to them have been told or would you leave it to the market to find out about that they are not lending to the government of New the problems? Zealand. If those institutions collapse, even though they are owned by the government, people lose their Peter Nicholl: Investigators are not replacing inspec- money. tors. We have never had inspectors, but we have The one institution that did collapse, DFC, had always had that power if we had doubts about infor- been privatized, although the major shareholder mation. We would probably use an external auditor was the govemment Supernation Fund. The audi- as our agent rather than a member of our own staff. tor was the govemment auditor and it was super- Auditors do not always know what is going on, and vised by the central bank. So the government links neither do supervisors. But in most cases they know were fairly close. Of all the creditors, 80-90 percent the direction of things. were Japanese, and there was a lot of pressure from It is true, we do get privileged information simply Japanese institutions that basically expected the by being the banker to the banks. If we had cause for government to cover all their losses. In the end the concern, we would require that it be disclosed in the government contributed to the package that was next quarterly statement. If capital had gone below put together, but ordinary depositors still lost the requirement, there would have to be a plan to about 20 cents on the dollar and the subordinated restore it. If capital was close to evaporating, then we debtors, who were all Japanese, only got about 25 would consider using our statutory power to either cents on the dollar. Still, they accepted that pack- give the bank a directive order or put it under statu- age. The government contributed to the solution tory management. If it were a minor breach we but did not provide a total rescue. However, the would require them to disclose it and correct it. We politics were different because most of the deposi- certainly would not sit on the information. 81 Roundtable Discussion: Systemic Banking Crises T he panelists for this roundtable discussion were tional reserve banking system. These failures usu- Robert Eisenbeis, senior vice president and ally involve one or more of three activities-pursuit director of research at the Federal Reserve Bank of inflationary policies designed to monetize gov- of Atlanta; Valeriano E Garcia, principal economist in ermnent debt, governmental forbearance when the Technical Department of the Latin America and institutions have financial difficulties, or use of Caribbean Regional office at the World Bank; Sergio banking system resources to cross-subsidize fiscal Ghigliazza, director of the Center for Latin American policies to the point of systemic collapse (either Monetary Studies; Allan H. Meltzer, professor of polit- because of withdrawal of foreign-owned funds, cre- ical economy and public policy at Carnegie Mellon ating an exchange rate problem, or because of flight University and visiting scholar at the American of domestic capital to sounder currencies and real Enterprise Institute; Jose Evenor Taboada, president of assets). the Central Bank of Nicaragua; Roberto Zahler, presi- Crises are most likely to occur when markets are dent of the Central Bank of Chile; and Rodrigo Bolafios not convinced that credible macroeconomic mone- Zamora, president of the Central Bank of Costa Rica. tary and fiscal policies are being (or will be) followed. That is, excessive inflation or fiscal policies that cre- Robert Eisenbeis ate too great a role for government in the manage- ment and allocation of resources are not credible Financial crises have their roots in monetary and fis- policies and will lead to a financial crises. In the cal policies that are not viewed as credibly contribut- United States at least, a decline in real economic ing to economic stability and low inflation. Credible activity triggers a financial crisis rather than the monetary and fiscal policies are hard to sustain, and other way around. the potential for crises increases when government attempts through direct ownership to operate financial Government ownership intermediaries. Moreover, governmental forbearance and a lack of credible policies on the closing of troubled Temptations are great and incentives are huge to institutions increase the possibility of financial crises. engage in cross-subsidization of activities when gov- There is also a link between supervision and interven- enmuents own and operate financial institutions. tions in individual institutions and the likelihood of Profit maxinmization incentives are dominated by financial crises. The goal of policies should not be to political and other concerns, and the potential for prevent failure or limit the costs to liability holders but mischief and abuse is great. to make failures independent events. Finally, a lack of transparency in the accounting of market values of on- Credible policies and off-balance sheet activities can increase the prob- ability of financial crises in both private and public sec- Credibility relates to concerns that governments tor institutions. will not follow the rules of the game, whatever they may be. For example, it is useless to design a Origins offinancial crises deposit insurance system, however elegant, if eco- nomically insolvent institutions are not closed or Most crises stem from govemment policy failures costs are not imposed on those who supposedly rather than from an inherent instability in a frac- had their funds at risk. Not following the rules is 83 Roundtable Discussion: Systemic Banking Crises forbearance. Experience has proved that forbear- sponsored enterprises with implicit guarantees and ance is a bad bet. It increases taxpayer losses and no monitoring of risks to protect and limit taxpayer creates incentives for moral hazard behavior. exposure to risks. Contrary to some views, a properly designed and So what conclusions can be drawn from all of this? instituted deposit insurance system does not carry Financial crises have their roots in misguided gov- implicit moral hazard. Rather, moral hazard results emment policies, and the situation can worsen from forbearance, from implicit guarantees, and where financial systems involve government-owned from mispricing of explicit or implicit guarantees. entities as active participants. Lack of credible poli- This is another type of policy failure. It is also why cies and lack of will to follow the rules lead to for- academics like Professor Edward Kane and others bearance and increase moral hazard and taxpayer have argued strongly for control of government risks. Finally, focusing on creating a riskless asset for incentives as part of the broader package of finan- payments purposes misses the critical point that gov- cial reform. ernments will still be likely to intervene-at taxpayer expense-when financial institutions experience a Transparency crisis. The potential for crises is greater when markets lack Valeriano F Garcia information on the quality of assets and the true value of an institution's net worth. When the owner Macroeconomic weaknesses are the main cause of of financial institutions (whether the govemment or banking distress and banking crises. Some of these a private entity) can hide from depositors and weaknesses develop from poor policy mixes-such investors the true value of the institutions' net worth, as lax fiscal policy coupled with tight monetary poli- then the owner can engage in subsidized risk taking. cies-that result in sky-high interest rates, increased Many reforms now recognize the role that disclosure quasi-fiscal deficits, and, of course, a day of reckon- can play in enhancing market discipline. ing. Latin America provides many examples of these time-inconsistent policies. Think of Argentina's Narrow banking experience in the late 1970s and early 1980s, when fis- cal policy veered to the left and monetary policy to The fascination with narrow banking and its focus on the right. Interest rates shot through the ceiling while creating a riskless transactions deposit is the mental bank portfolios deteriorated and exchange rate cred- equivalent of twiddling one's thumbs. There are ibility was called into question. The result was a clear externalities to a well-functioning financial sys- banking crisis. tem. It enables the separation of real savings deci- Even so, banking distress is not due exclusively to sions from investment. It facilitates the intertemporal inconsistent monetary and fiscal policy. Some of the transfer of consumption and investment. It enables blame lies in adjustment policies designed to treat the risk shifting that characterizes modem financial inflation and macroeconomic instability. The bank- institutions and markets. Promotion of economic ing system, accustomed to enjoying a share of the efficiency and specialization of labor are what dis- inflation tax, is shaken when a stabilization plan sud- tinguish modem economies from barter economies, denly removes this source of profit. and achieving these benefits depends on a well-func- So, in Latin America at least, the initial source of tioning financial system. Were financial institutions the problem has been macroeconomic-either bad not involved in payments (if, for example, the gov- macroeconomic policies or, in the transition from bad ernment permitted the exchange of infinitely divisi- to good policy, sharp changes in relative prices and a ble treasury obligations), these other extemalities reduction in the inflation tax shared by the banking would exist. Threats to these positive extemalities system. These macroeconomic problems then inter- would trigger rationales for governmental interven- act with the banking sector, deepening the crises. The tion in times of crisis, even without direct involve- failure of some banks, coupled with deposit insur- ment in the payments system. ance, implies a feedback from banks to the macro- To argue for narrow banking misses the point economy, aggravating the fiscal dilemma. It is at this entirely. Narrow banking solves no problems in point that govermments usually intervene to indem- today's financial system, nor does it remove the pos- nify depositors, increasing government expendi- sibility that taxpayers will not be put on the spot tures and money supply. when there is a crisis. In fact, a move to narrow bank- In Latin America and the Caribbean the provision ing may create an entire system of government- of deposit insurance is cyclical. As we all know, 84 Roundtable Discussion: Systemic Banking Crises deposit insurance creates the potential for moral system will require more than "better" laws: some- hazard behavior, which is inconsistent with a healthy thing has to be done about enforcement and, as his- banking sector. In the past when banking distress tory tells us, the best enforcement agency is the emerged, deposit insurance was quickly blamed, market. and was either eliminated or reduced in scope. But It is vital to distinguish between rules and discre- when banking crises resurfaced, deposit insurance tion. Automatic systems and rules must be imposed was reinstated. There seems to be a love-hate rela- and discretionary power removed from regulators. tionship with it. When there is deposit insurance Regulators often try to solve problems by using moral hazard can wreak havoc, and there is a ten- moral suasion or by trying to isolate and hide things dency to eradicate it from the banking system. from the market. But in Latin America this behavior Conversely, when there is a threat of a domino effect, leads to worse outcomes than informing the market deposit insurance is seen as the answer. in advance. The basic dilemma for authorities is related to Two things usually happen in the region's bank- fractional reserve requirements, and the best solu- ing crises: a liquidity crisis and a reduction in the tion is to eliminate fractional reserve requirements real amount of credit. Argentina, Brazil, Mexico, and and adopt a narrow banking system. A narrow sys- Venezuela all illustrate the liquidity effect and the tem does away with deposit insurance and, at the effect of crisis on real money and credit. In Argentina same time, eliminates the threat of a domino effect. in the first quarter of 1995, real (and nominal) money Although macroeconomic causes of banking declined by about 20 percent (figure 1). Suffice to say, crises are the most important, a sound microeco- this was a grim picture. In the United States during nomic environment can help mitigate the crises. A the 1930s the monetary base went down by 33 per- three-legged structure is needed: a sound regulatory cent-but that was over three years. Argentina's framework, efficient supervision, and early enforce- banking sector also experienced a liquidity problem: ment and resolution. the ratio of MI to M3 increased, although this Most Latin American and Caribbean countries growth was not as significant as the decline in total have an adequate regulatory framework. Codes of money stock. The same traits are evident in other law are satisfactory. Supervision in many countries is countries. not bad. What is lacking is tough enforcement, up to Brazil's experience is also interesting. In the past and including bank closings. Supervisors are usually decade it has undergone many changes in economic aware of problems but distrust the market. They are policies and seen myriad failed stabilization plans. In afraid of immediate enforcement because they think response to its economic crisis, it developed the Real that if the market perceives what is happening, there Plan. This plan drastically reduced inflation, but it will be a domino effect. So, they fail to enforce the also drastically reduced an important source of bank- rules. As a result the problem escalates to the point that it can no longer be concealed from the market- Figure 1 Real money and credit in Argentina, a much worse situation than if the market had been 1991Q1-1995Q4 informed at the outset. This tendency is not confined to the banking sys- Real money tem. For example, the Mexican crisis of late 1994 and (billions of U.S. dollars) Liquidity ratio early 1995 would have been less severe if markets 6 .40 had been privy to all data on a real time basis. In December 1994 data were roughly four or five 5 money\ months out of date. It is imperative that the market .35 be kept abreast of all pertinent (even negative) infor- 4 / mation. In the long run this approach helps avoid I graver repercussions. 3 Lack of enforcement is endemic in Latin America. / V Liquidity .30 ratio Take, for instance, the 1853 Argentine constitution, 2 (Ml/M3) which clearly was "imported" from the United States. Its general democratic principles were good 1 l .25 but they were never enforced with strength. 1991 1992 1993 1994 1995 Throughout the region's history, despite the presence Q1 Q1 Q1 Q1 Q1 of institutions, laws, and regulations, compliance and enforcement are absent. Changing the banking Source: Central Bank of Argentina. 85 Roundtable Discussion: Systemic Banking Crises ing profits: the inflation tax. As a result Brazil's cur- Sergio Ghigliazza rent banking problems are related to the transition to low inflation from an environment that provided To impute banking crises to macroeconomic imbal- banks with huge inflation-related profits. ances only means that, by adjusting these imbal- Mexico, on the other hand, did not suffer from a ances, the banking system could operate efficiently liquidity crisis. In January and February 1995 Mexico and that it would be free from bank crises. was able to avoid a run on its banks: there was no liq- But market failures also play an important role in uidity problem but, as in Argentina, the real quantity bank failures, particularly when they allow for the of money dropped severely (figure 2). This decline possibility of moral hazard behavior. Official inter- reduced real credit and intermediation and, together ventions, such as deposit insurance, that are with the sharp increase in interest rates and changes intended to stabilize the financial system often in relative prices (exchange rate devaluation), aggra- become obstacles to efficient market operations. vated the banks' problems. Bank supervisors are responsible for ensuring the The so-called tequila effect experienced by Mexico smooth functioning of the financial system. But they also touched Argentina, although Argentina has are not the only ones. The central bank and other gov- recuperated from the blow mainly because it emient agencies also must play a role in making the avoided the sterilization of capital outflows, follow- market work. In many countries where the financial ing rules rather than discretion. Mexico faithfully sector was liberalized, such as Mexico, banks were adhered to discretionary policies, and its sterilization left without reserve requirements, interest rate con- of capital outflows finally contributed greatly to the trols, and required asset allocation channels. The debacle. authorities believed that, once liberalization was In sum, macroeconomic shocks, the main culprits introduced, the market would respond wisely allo- of banking distress and crises, can originate from bad cate resources efficiently. policies or from good policies that change relative But liberalization is just one part of the task. In a prices. The problem is that a bank's portfolio that is country like Mexico, where at the moment of liberal- good with one set of relative prices can be bad with ization the financial sector was closed to foreign com- another set. Thus narrow banking and market-based petition, there was implicit unlimited deposit regulations should be seriously considered as good insurance, and two banks controlled 60-70 percent of alternatives. Given that the banking system cannot the financial resources, the market could not auto- be shielded from macroeconomic shocks, a more matically be expected to operate efficiently. resilient banking system (narrow banking) and bet- Intelligibility is a prerequisite. It is unreasonable to ter regulation, supervision, and enforcement (based expect market results if market conditions are not on market forces) must be applied. fully in place. Long-term stability-in growth, prices, Figure 2 Real money and credit in Mexico, exchange rates-is essential to the development of 1991Q1-1995Q4 markets. Although some countries have been able to develop financial systems despite market imper- Real money fections and imbalances, sustainability and stabil- (millions of new pesos) Liquidity ratio ity go hand in hand, because stability can overcome 3.5 Real .40 or lessen the inefficiencies that derive from market moneyA imperfections. .35 3 .30 Allan H. Meltzer .25 Capitalism without failures is like religion without 2.5/Liquidity sin. It does not work. Preventing banking crises is ratio .20 easier said than done. There are three aspects of the (MI/M3) problem. The first two are macroeconomic and 2.0 .15 microeconoic (moral hazard, risk, incentives, and 1991 1992 1993 1994 1995 how to deal with them). Q1 Qi Q1 Q1 Q1 The third one, which requires much more thought, is political economy. What is feasible in practice? Is it Source: Central Bank of Mexico. possible to design a system that minimizes social 86 Roundtable Discussion: Systemic Banking Crises losses? It is important to distinguish between private Banks and other financial institutions exist almost and social costs. The aim is not to eliminate all future entirely because they acquire assets that trade infre- banking failures. It is to avoid subsidizing risk while quently or not at all. Bankers make judgments about not penalizing it. whether and when prices or interest rates will rise or Some suggestions, on limited liability for instance, fall, whether firms will fail or delay repayment, and would clearly penalize risk. There is a real danger of so on. Regulators cannot make these judgments making banks so risk averse that their functions will unless they become bankers. There will always be end up being performed somewhere else and shifted differences of opinion between the two because reg- to another part of the system. It is a question of strik- ulators are usually more risk averse than bankers. ing the balance between too much and too little risk. Those judgments usually have to remain with the Finding that optimal balance for each country is not banks. Problems, arise, however, in markets where easy. For example, it would not be in the interest of many evaluations are imprecise. This weakness most countries to return to a lot of private lending alone creates an argument for political interference and borrowing outside the banking system. These and forbearance when there are problems. activities should be kept within an active and devel- Modem portfolio theory stresses that risks arise oping financial market. That will not happen if the from the structure of portfolios. Many people at this effort to prevent bank failures eliminates all or too conference have talked about the use of capital in the much risk in the banking system. Basle agreements. Nevertheless, the Basle agree- ments are seriously flawed because they assign risk Some issues to individual assets and not to portfolios. People can change the risk position of their portfolios without Some broad generalizations and conclusions can be making much structural change in their assets. They drawn from the presentations and discussion at this can make the change by selling short or selling for- conference. First, there are social benefits from the ward, by trading in derivatives, and so on. A regula- widespread use of payments and financial systems tor could examine portfolios one minute and apply and the development of markets for lending and bor- the Basle criteria, and twenty minutes later the whole rowing. Deposit insurance, among other things, pro- exercise would be irrelevant. The Basle agreements motes these social benefits. Although much has been miss what seems to be the most important implica- said about the problems of deposit insurance and tion of modem financial theory-that the risk is in moral hazard, there are benefits. If it is credible, portfolios, not in individual assets. deposit insurance makes people want to use the There is evidence that auditors are slow to detect financial system more rather than less. That has some fraud and failure and have difficulty appraising risk. social benefits. So, how do we get those benefits with- That does not mean that audits are useless, however. out incurring the costs of the moral hazard? The inability of bankers to predict an auditor's Preventing systemic risk requires a lender of last arrival acts as a deterrent to wrongdoing and exces- resort, usually the central bank. In a currency board sive risk taking. Audits cannot solve all problems but arrangement there have to be lines of foreign credit they should not be eliminated. or, possibly, a fiscal authority that can sell treasury bills to assist solvent but illiquid financial institu- Some answers tions. Under colonial currency boards-a system highly developed in the old British empire-most What can be done to reduce risk in financial markets? banks in a country were branches of banks abroad. First, there must be a lender of last resort. Second, The home offices supplied liquid assets or foreign there must be protection for households and busi- exchange, and the possibility of a run on the banks nesses, while avoiding deposit insurance and moral was much less. Runs did not occur often, if at all. hazard. The goal of policy should be to reduce to a mini- Lender of last resort is a public good that the cen- mum risks that are inherent in nature, trade, and tral bank can supply at low cost, provided that the other parts of the institutional structure. There lender operates only when there really is a systemic would still be financial failures. The aim is to avoid problem and does not become the lender of first systemic runs that disrupt payment systems and resort. This is a great risk in all countries and is why financial flows. Problems usually arise in markets Bagehot developed a wonderful set of rules for Great when information is not readily available. These Britain. Later, the lender of last resort in Britain, who markets are not traditional, efficient ones where all only operated at a penalty rate against particular risks are priced correctly. classes of assets, became the lender of first resort who 87 Roundtable Discussion: Systemic Banking Crises operated every day. Hence the banks never had to States the stock market can fall 30-40 percent hold any reserves because they knew they could overnight, as it did in 1987. If that can happen in the always go to the lender of first resort. It is important United States,with its well-developed financial sys- to create the public good but also to restrict its use. tem, what is the variance of asset prices in other To provide protection for households and busi- countries, and how many problems will that create? nesses while avoiding the problems of deposit insur- Structured and early intervention must be tied to ance and moral hazard, the central bank lends at a capital requirements, which in tum must be tied to penalty rate on a variety of marketable assets-or eli- variability. gible securities. The penalty rate is set by letting the Finally, does the world economy need an intema- discount rate for each security float with the rate on tional lender of last resort or a bankruptcy agency? that security. It is always above the last quoted price The answer is a resounding no, and for two reasons. on the security. The discount window should be First, because a lender of last resort has to create base opened to anyone holding eligible paper. This money, and that money has to come from a central includes, specifically, nonbank financial firms that bank. Second, it would create all sorts of new moral trade securities or use wire transfers. None will come hazard problems. If an individual or group can to the window if the market is functioning correctly decide that a state is bankrupt, lenders face a new because business can be done at a better price in the risk. If that happens risk is created, not removed. private market. They will come only when there is a Who is going to lend? And how much will they lend problem in the market, and that is exactly when the if they think the International Monetary Fund can lender of last resort is needed. That sets a policing come along and disoblige countries from servicing role for the penalty rate. debt? Such an arrangement would adversely affect Monetary policy is usually conducted through lending, borrowing, and financial flows to develop- open market operations, not the discount window. ing countries. Rates in the market are set according to money growth or other criteria, but for an individual bank Jose Evenor Taboada there is always an opportunity to come and get more. To prevent discounts from affecting total reserves or In the past fifteen years at least two-thirds of the money, the central bank can sell in the open market. World Bank's member countries have experienced It can drain the market but provide reserves to the banking sector problems. These developments have bank that requires accommodation at the penalty led the Bank and the Inter-American Development rate. As Bagehot pointed out, these policies should be Bank to finance structural reforms in banking sys- stated in advance and followed in a crisis whenever tems and in supervision and to provide relevant there is a premium on base money. This will not hap- technical assistance. The International Monetary pen without rules, and there must be penalties on Fund (IMF) recently has been discussing, at great bankers (and on regulators) who do not follow the length, the implications of financial sector crises and rules. the need for greater cooperation and coordination For many developing countries, a few other issues between multilateral institutions to deal with such need to be addressed. One is insufficient diversifica- problems. The IMF also hopes to increase its techni- tion within an economy or financial market because cal support in this area. of a country's size and structure of production. That Most analysts now recognize the close relation- could be Chile and its reliance on copper in the past ship between banking system soundness and macro- or Mexico and oil. Where there is one dominant asset economic policy. Policymakers must take the and there is a change in the relative price of that asset, condition of the banking system into account when the country simply cannot avoid a financial problem. formulating economic policies, both as a key objec- Hence the question is how that problem should be tive for and as a constraint on policies. A stabilization dealt with. Should foreign banks be allowed into the program targeting a specific instrument may need to country, or should domestic banks hold foreign be adjusted if it faces an unsound banking system. assets and lend outside the country? Which is the Nicaragua is a prime example of banking reforms right approach? in a transition economy. Until 1991 the banking sys- There is also the issue of the size of capital in bank- tem was comprised entirely of state-owned banks ing. Many countries are now thinking about capital and, under Nicaragua's old constitution, insurance structure. To prevent losses, capital must be tied to also was owned by the state. Since then a stabilization the variability of prices. More research is needed on and structural adjustment program has been under how much capital is required. After all, in the United way, including the liberalization of the financial sys- 88 Roundtable Discussion: Systemic Banking Crises tem. Interest rates were liberalized and became posi- market discipline, direct mechanisms of intervention tive in real termns, banks' legal reserve requirements (such as reserve requirements), contingent and trans- have been reduced substantially, and directed credit parent government liability in some cases, and per- policies have been eliminated. In addition, an sonal accountability for directors and management. autonomous Superintendency of Banks was created and private banks were authorized. Banking regula- Roberto Zahler tion and supervision were introduced with the objec- tive of keeping a sound system compatible with most All banking systems maintain a mismatch between modem market-oriented systems. the value of assets and the value of liabilities. In other How did the financial system respond to these words the economic value of assets is variable and new policies? Unlike private banks, which have market-determined, while the value of liabilities become stronger through higher participation in the tends to be equal to rigid and fixed accounting val- deposit and loan markets, state-owned banks have ues. In financial systems where the difference been facing myriad shortcomings and limitations between the value of assets and liabilities is too big, stemming from their inability to overcome the prob- liquidity problems may turn into solvency problems, lems of the old operating framework. That is, heavy causing failure. state subsidies and intervention, open access to pas- sive central bank credit, and lack of competition. Basic considerations These problems have been exacerbated by the public banks' decreasing deposit and loan market partici- To minimize the mismatch, two basic considerations pation and by a high level of arrears from excessive should be taken into account. First, it is crucial to concentration of lending to the agricultural sector, reduce the intensity and the extent of economic itself plagued by difficulties. cycles, especially when those cycles are not antici- Adjustment and stabilization policies in pated. Macroeconomic instability makes bank Nicaragua have had an important effect on the debtors more vulnerable and leads to inappropriate growth of financial savings, as well as on other sec- or biased credit risk evaluation by the financial sec- tors. GDP grew 3.3 percent in 1994 and 4.2 percent in tor. In most countries monetary policy tries to have 1995, and the fiscal balance in the current account of some early warning systems that control for this pos- the nonfinancial public sector fell from a deficit of 18 sibility, cooling down the economy in order to reduce percent 1990 to a surplus of 6 percent in 1995. In addi- the amplitude of cycles. The magnitude and intensity tion, inflation plunged from 11,000 percent in 1990 to of these cycles hits hard on bank loans in an asym- 11 percent in 1995. metric way. When cycles are contractionary, bank State-owned banks' repayments to the Central debtors are hard pressed to repay; this clearly is not Bank, which have run into arrears, recently intro- symmetric with the expansive phases of the cycle. duced the possibility of another crisis in the financial Second, macroeconomic policy shapes the behav- system. However, preventive measures have been ior of key prices in the economy-interest rates, taken in the form of downsizing the state-owned exchange rates, wages, and asset prices. If these banks and in transferring nonperforming loans to a prices are outliers or divorced from their fundamen- collection agency. In the future both state and private tals, they may generate huge adverse impacts on banks will depend exclusively on their efficiency and debtors. Policy flexibility is crucial because it allows effectiveness in order to continue operating in a sys- shocks to be distributed among different markets. tem where the only source of funds will be from Chile has clearly learned this lesson. When foreign or deposits. Only time will tell if Nicaragua's state- domestic shocks occur, they are distributed among owned banks will survive or if their existence is the credit and financial market, the foreign exchange incompatible with market-oriented system. market, and the labor market. There are several ways to prevent banking crises, Asset prices tend to be a residual or equilibrium one of them being effective and efficient supervision. mechanism whenever too much weight is placed on This entails an independent supervisory authority, certain variables. Moreover, bubbles in asset prices on-site supervision, higher capital ratios create problems not only because of their wealth (Nicaraguan banks' capital adequacy ratio was effect on domestic spending but also because of the raised to 8 percent in June 1996 and will be further collateral value of bank lending-which may turn raised to 10 percent by 1999), stringent rules on asset out to be totally disproportionate to its real value evaluations, early exit and intervention strategies for once the bubbles burst. So, in addition to flexibility it banks in deep distress, more transparency to foster is important to understand the way asset prices 89 Roundtable Discussion: Systemic Banking Crises (especially prices of stocks and nontradable assets) Something similar happened in Mexico in 1994, tend to move. The possibility of movements that are when an excessive current account deficit was comnpletely out of proportion to reasonable funda- financed by short-term capital inflows and, finally, mentals should be taken into account in the design the exchange rate system broke. The financial system and implementation of policies and regulations for also faced serious problems, and the economy tum- the financial sector. bled into severe recession. The timing, speed, and sequencing of reforms is In the past five years Chile has maintained a another important issue. In the mid-1970s a mission macroeconomic situation where there has been a from an international financial organization was sent fiscal surplus of around 2 percent of GDP, mone- to Chile to work on financial liberalization. Real inter- tary policy has been quite restrictive, and the est rates on deposits were, for a second consecutive exchange rate has been used to help contain a cur- year, 45-50 percent. One member of that mission said rent account deficit of about 3 percent of GDP, a rea- that if interest rates were market- determined-he sonable and sustainable figure. There also has been never explained what market-determined meant- a huge effort of sterilized intervention in the for- then they would find their own best level. That eign exchange market. In fact, by the end of 1995 turned out not to be the case-in fact, they stayed at international reserves were around 25 percent of around 40 percent for four or five years, contributing GDP. Given this high stock of international to the bankruptcy of the financial system. Those rates, reserves, Chile was able to prepay external debt, of course, were never related to the price and mar- reducing the quasi-fiscal deficit. In addition, the ginal productivity of capital. They were clearly out- currency has appreciated by an average of 4.5 per- liers. It is a mistake to think that because the prices are cent a year over the past five years. As a result the "market-determined," everything will be fine. When economy is growing by 6-7 percent a year, the cur- outliers persist, there is likely a problem with regula- rent account deficit is less than 2 percent of GDP, tion, bank owners or directors, moral hazard, and so inflation is down to less than 8 percent, and there on. No one should maintain confidence in appropri- is huge foreign direct investment, good export per- ate market behavior in such a situation. formance, and strong long-term and medium-term Another example. In the late 1970s Chile's cur- capital inflows. rency was pegged to the U.S. dollar, at a time when Given the circumstances, Chile had three options. wages were fully backward indexed to past inflation, It could do nothing and simply allow the market to domestic inflation was 35-40 percent a year, and work, thus favoring de facto an increasing apprecia- international inflation was 4-5 percent. There was a tion of domestic currency, a higher current account clear incompatibility between the exchange rate pol- deficit, and a greater vulnerability of the economy to icy and wage policy, but there was a pretense that foreign shocks. It could gradually liberalize capital domestic inflation could rapidly and smoothly con- inflows. Or it could close the wedge between what verge to the level of world inflation. A disequilibrium was considered to be an equilibrium in the domestic in the exchange rate contributed to a huge current interest rate and the interest rate Chile faced in inter- account deficit. Moreover, the exchange rate peg national markets. stimulated external inflows, which were mostly Because the first option (aligning the domestic intermediated by banks domestically as dollar- interest rate with the foreign interest rate) threatened denominated loans to both tradable and nontradable stabilization and current account objectives, Chile sectors, thereby increasing the foreign exchange risk chose a combination of the second and third options, of debtors directly and that of banks indirectly. What the third by establishing a special reserve require- caused the appreciation of the domestic currency ment on short-term capital inflows. and the current account deficit? There was no fiscal Countries that appear to be more liberal than deficit and monetary policy was apparently nonex- Chile in financial integration with the world econ- pansionary, but there was high domestic spending. omy also tend to have higher domestic real interest The current account of the balance of payments rates. Why? Basically, those countries tend to have reached 14 percent of GDP in 1981, a figure that higher expectations of devaluation than does Chile understated the weakness of Chile's external or higher country risk premiums in international accounts because of the impact of an excessively markets-or both. Neither of these elements, which appreciated peso on GDP. But because there was no mainly reflect the international financial commu- fiscal deficit, many highly regarded economists nity's lack of confidence in a country's macroeco- believed there was no need to worry, that there was nomic policies, is the proper way to integrate with no problem. Reality proved them wrong. international financial markets. 90 Roundtable Discussion: Systemic Banking Crises Chile has tried to pursue policies aimed at a seri- banks (such as workers and housing banks) created ous, responsible, and sustainable strategy of finan- by special laws. All banks are regulated by the cial openness-witness the prepayment of foreign Superintendency of Banks, which is part of the debt, the accumulation of international reserves, the Central Bank. Distortions in the system over the past healthy composition of the capital account, the huge ten to fifteen years led to the development of a par- amount of foreign direct investment, and an increas- allel (offshore) banking system. Most domestic pri- ingly internationally integrated financial system. vate banks have a parallel operation in the Caribbean Moreover, there has been significant Chilean invest- or in Panama, mainly processing transactions in U.S. ment abroad. In that respect Chile has been cautious dollars with Costa Ricans. There are other interme- with the foreign lending and investment of banks diaries, including small nonbank financial compa- and pension funds because of prudential policies nies and the so-called securities market, which trades related to state guarantee schemes and systemic risk in money market instruments. In addition, an incip- considerations. ient private pension fund has developed as a com- plement to the mandatory state-managed pension Future challenges scheme and state monopoly in the insurance market. Costa Rica's financial system suffers from ineffi- The Chilean financial system will face two basic chal- cient state banks (with spreads of 14-16 percent lenges in the next few years. One has to do with the between average loan and deposit rates), bad loans need for consolidated supervision of financial and portfolios, and insufficient capital. State banks groups. Supervision is lacking, but this has not been hold a large but declining share of total bank assets a major problem because of the nature and scope of and liabilities and have been highly protected. Some banking. But with a law in Parliament that would of these distortions are being ameliorated by a state allow banks to do more domestically, such as factor- warranty on all sight and time deposits, without any ing and securitization, in addition to cross-border limits. Until recently state banks enjoyed a monop- lending and the intemational establishment of oly on sight deposits and had exclusive access to the branches, subsidiaries, and joint ventures, consoli- discount window of the Central Bank. These privi- dated supervision is now a must. leges have been taken away. The most important challenge in maintaining the Over the past eight to ten years Costa Rica has health of the financial system has to do with intema- been moving gradually to prudential supervision. tionalization. Once the new banking law is passed, The offshore banks remain a big problem, however. Chilean banks will be able to invest abroad, subject Monetary and exchange rate policy focus on reduc- to certain criteria, posing major challenges for super- ing the dependency on reserve requirements, which visory institutions. Chile must seek to limit country are being lowered from about 40 percent to 15 per- risk in the emerging markets where banks will make cent for sight deposits. Costa Rica is moving to an loans and buy shares from other banks. Contagion auction system in open market operations and for risks also must be considered. The authorities will ten years has followed a crawling peg, with daily have to be aware of how supervision is conducted in mini-devaluations based on the real exchange rate those banks, coordinate that supervision with its and the level of intemational reserves. Fiscal policy own, and, most important, seek consolidated super- and monetary policy interact in a special way: vision of the banking system. Chile must ensure that Central Bank losses are equivalent to 2 percent of regulatory arbitrage risks are controlled, especially GDP, mainly because in the past the bank was used in terms of possible margin evasions for large expo- to secure extemal debt and to perform several other sures and related credit. We do not yet have the quasi-fiscal operations. These losses have caused expertise to ensure that banks will not be exposed to problems with monetary control. Inflation has fluc- excessive risk if they invest outside Chile. But with tuated from 10-30 percent over the past twenty the irreversible trend toward intemationalization, years and interest rates have been positive in real we can try to ensure that these activities proceed in a terms but volatile. gradual and prudent way. Trade and capital accounts are open, allowing cap- ital to flow freely in and out of the country. Although Rodrigo Bolanios Zamora we would like to control some short-term flows from the offshore banking system, we have neither the Costa Rica's banking system comprises three large legal right nor the economic resources to do so. state commercial banks, twenty-two small, domesti- However, regulations for the offshore banking are, cally based private banks, and a few other public being considered. 91 Roundtable Discussion: Systemic Banking Crises As to narrow banking, Costa Rica is, in fact, mov- depositors to shift toward domestic banks and away ing in the opposite direction. Reserve requirements from offshore banks. are being, lowered in an effort to lower the cost In terms of supervision, the government closed a imposed on intermediation margins by the spreads poorly managed state-owned bank in 1994. It was the in the banking system due to unremunerated reserve third largest bank in Costa Rica and the trial of the requirements. We are also preoccupied with the managers is still pending. Discussion has begun on problems created by new financial instruments that how to improve management in the three remaining substitute for sight deposits and with the problems commercial state-owned banks. Market-based created by fiscal and electoral cycles that tend to supervision would allow for better information, increase the fiscal deficit near election time. especially from offshore banks. The new law gov- A change in the constitution is planned to restrict erning the central bank gives it some authority to the nonfinancial public sector deficit plus the central require more information from offshore banks that, bank deficit to not more than 1 percent of GDP. That in any case, it already requires from domestic banks. change has already been approved once by Thus Costa Rica is moving to raise bank supervi- Congress; to become effective it must be approved sion to intemational standards within the next twice. The possibility of introducing deposit insur- decade. Other Central American countries are mak- ance is also being examined. In the past year several ing similar efforts to reduce distortions, improve consultants in Costa Rica have recommended supervision, and address the problems of state deposit insurance, which might be a way of inducing banks. 92 Floor Discussion Question: How would you have dealt with the The people holding those securities were receiv- Mexican crisis? ing 20 percent or more in yields. Thus they already had received a substantial risk payment for taking Allan Meltzer: Mexico's problems began after those risks. They knew the risks. Sometimes risk the election. There was an outflow of funds in hurts-and sometimes people have to learn that March 1994 after the assassination of presidential you do not get a 20 percent return buying safe candidate Luis Tonaldo Colosio. Then things stabi- assets. lized-the exchange rate appreciated a bit, and interest rates fell. In October there were still 100 per- Question: Allan Meltzer, you said that it was pos- cent reserves in foreign currencies, so Mexico was sible to penalize bankers too much with, say, unlim- able to stabilize. ited liability. But do you want to just keep trying to Then two big events happened at about the same raise capital? And how high do you go? You talked time. One was further assassinations, which fright- about looking at asset prices like stock market prices, ened people. The other was an indication that the but there are many different variables in the regula- incoming government intended to do nothing about tory framework that influence how safe a bank the fiscal deficit. It also had no plans to cut the rate of should be made. Could you or Roberto Zahler elab- money growth. By December 1994, for example, pro- orate on this issue of whether you want to do it by ducer prices were rising by 9.3 percent, about twice increasing liability limits, or do you want to just arbi- the annualized rate of the previous December. So trarily raise capital? Then, most important, how do there was not only inflation but also evidence that it you fold this in with the lack of diversification if was beginning to take hold. some countries, like Chile, want to go slow in allow- What should the Mexican government have ing banks to invest overseas? Does that imply that done? It should have stabilized in October by you have to be more vigorous with capital or liabil- announcing a fiscal tightening. But failing that, what ity or something else to make those banks safe? were its choices? Either to impose a large debt bur- den on the Mexican economy by paying off foreign- Roberto Zahler: One argument about overseas ers or to spread the loss and default on the debt. That investment is that Chile's banking system is some- is, to say to the debtholders, "you thought you were what risky because of the size of the market and con- here for three months but you are going to be here for centration, and so Chilean banks should increase three or five years. So, let's discuss the terms of the their scope of activities within and outside Chile. But workout." to do that without jeopardizing the financial system, Of those two choices, the second seems better. we need reasonable assurance that supervisory insti- That is, given that macroeconomic policies were tutions can provide minimum standards to maintain destabilizing in an election year, the choice was the health of the system. between borrowing abroad to pay off mainly foreign Two issues are involved: consolidated balance and domestic lenders who had gotten into dollars, sheet supervision and better knowledge of what it dollarized securities, or other securities early or means to go outside the country-not only investing telling them that they would have to wait three or outside, but lending abroad. These are major chal- five years before they were paid back. I would have lenges and Chile is headed in that direction. Clearly, chosen the second option. the big question is how soon we will get there. And 93 Floor Discussion of "Systemic Banking Crises" that depends on how quickly are we able to increase Robert Eisenbeis: Banks should not only move supervisory capacity. toward real-time gross settlement but toward real- We are comfortable with capital bulk, capital time gross settlement only against good funds. requirements, and things of that sort, and we feel Efficiency comes from participants being able to comfortable with the criteria we are using now. For decide how to manage and monitor risk exposure. I example, banks wanting to go outside Chile must cannot rationalize, particularly given the current have at least a 10 percent capital-asset ratio (instead rates that are being charged for interday money, of 8 percent, which is required to increase the scope that there is anything but free capital being pro- of activities in the country). Ours is a pragmatic vided from the Federal Reserve or the Fedwire. And approach to what must be a gradual process. with the evolution of markets to a system that is moving increasingly toward twenty-four hour trad- Allan Meltzer: I will repeat something that has ing, the idea of daylight overdrafts effectively dis- been said here many times: if we continue with the appears. macroeconomic policies of the past, there will be no I am also not persuaded about the gridlock argu- solution to these problems. We must balance too lit- ment against real-time gross investment. Gridlock tle risk-bearing by bankers with too much risk-bear- only comes when institutions do not manage their ing by bankers. The goal is not to drive all the risky own liquidity. An institution has to make the neces- activity out of the banking system; the goal is to pool sary investment to be in the payments business. it. Striking the optimal balance is difficult, but under Banks must provide sufficient resources to that activ- better macroeconomic conditions the risks will not ity so that it becomes economically viable. One can be as large as they were before. always borrow from the lender of last resort if there is a liquidity problem. But what is worrying is that Question: I would like to hear more about the for systems with other than real-time gross settle- links between liquidity, reserve requirements, and ment with clearing only against good funds, the the lender of last resort function. In New Zealand main risks often come from initiators who are not there is an initiative to shift from deferred settlement under the supervisory responsibilities of those bear- to a real-time gross settlement that is processed, set- ing the risks. Namely, in the case of the Federal tled, and made final, irrevocable, and unconditional. Reserve, the issues often come from outside parties From the risk reduction viewpoint, real-time gross who are initiating payments and become the source settlement is obviously superior to deferred net set- of risks over which the Fed has no control. Under tlement. those circumstances I would rather see many of these Peter Nicholl said that New Zealand has abol- payments markets go to operating like a futures mar- ished reserve requirements. But careful examination ket, which is essentially what real-time gross settle- of these arrangements is needed to understand who ment clearing only against good funds is about. is providing liquidity and who is assuming risks. Real-time gross settlement can be designed in several Allan Meltzer: I agree. Many countries have ways. One is with central bank intraday credit that is tried to pass off risks to taxpayers only to find that daylight overdrafted, like that provided by the U.S. taxpayers are unwilling to accept them. Japan is a Fedwire, or without central bank credit, as with the good example, where the legislature is unable to Swiss Interbank clearing system. The second way is pass any legislation settling who is going to pay for to provide central bank credit, which was free of bank failures. But no one in Japan was protesting charge until last yearjust to keep the system running. when the legislature was paying off the banks' A third way is to provide a kind of grease to the sys- depositors. The experience has been similar in Latin tem with daylight overdrafts. In the Swiss system the America, North America, Europe, and elsewhere central bank does not provide any credit and pay- when governments bail out banks. And it is not easy ment is held on queue when funds are insufficient to for people to return to a system if they have been cover payment. This approach tends to create grid- forced to accept losses. lock. It must be stressed that there is a tradeoff between efficiency and risk. Question: Suppose that American taxpayers, at In Argentina high reserve requirements are the end of the 1970s, had been asked whether they required and the Central Bank's role as lender of last were willing to accept a mortgage system for savings resort is limited. Presumably, Argentina does not and loans that implied a risk that eventually could have any other realistic option under those circum- require them to pay $3,000 each. How would tax- stances. payers have answered? 94 Floor Discussion of "Systemic Banking Crises" Allan Meltzer. Clearly, they would have said no. Policy flexibility is another objective, allowing the They would have said that they want the system but cost of economic shocks to be distributed across differ- not the risk . That is why we need to find an incen- ent markets rather than concentrated in one segment. tive system that solves the problem in advance. It Deposit insurance and the threat of moral hazard is will not be perfect and there will be problems, but another important issue-how, without subsidizing they will be less severe than the problems now. No risk, to optimize the provision of deposit insurance? solution will work unless macroeconomic problems What is the best form of deposit insurance? How can are taken care of. different actors be made to bear responsibility for their actions? The answer to those questions are country- Question: Banks have lirnited capital. If there is a specific and depend on how well markets function, big change in the domestic discount rate, most of the extent of information asymmetries, and so on. banks' capital can be wiped out, and the usual incen- Market-based regulation is a good idea, because tives for shareholders to behave prudently are also in the information age transparency and disclosure wiped out. What kind of corporate finance mecha- are the only way to go. People bet against anything nism would allow for the replacement of sharehold- even slightly obscure, fearing the worst. That is why ers by others who would otherwise not interfere but transparency is essential. who could take over? Then there is the importance of accounting princi- ples, in particular those that allow for the valuation George Benston: That is the rationale behind of assets even in situations where functioning mar- structured early intervention and resolution. As a kets do not exist. What are the proxies for those val- bank's capital is depleted or as it grows beyond its ues, and how are they developed? In the absence of capital, its capital-asset ratio goes down. When the transparent accounting principles the 8 percent Basle ratio declines below prespecified "trip wires" the capital adequacy standard and all other guidelines authorities first may and then must take action to are useless, because the numbers mean nothing. prevent the bank from expanding, paying divi- Bankers must embrace the fundamentals, and that is dends,and so forth until the bank brings its capital- an area where Latin America and the Caribbean must asset ratio back up. Thus the bank rarely would have continue to make efforts. Adequate prudential regu- to be taken over by the authorities. Rapid depletion lation and effective supervision are also essential. In of capital could occur because of massive fraud or addition, given increasing cross-border operations of severely negative macroeconomic conditions. The banks, there is a need for regular consultation among supervisors should make fraud unlikely and the cen- regulators-not least for consistency. tral bank should control the macroeconomy. Finally, some countries in Latin America are developing policies on closing insolvent institu- Sri-Ram Aiyer: Three or four points from this ses- tions-but more and tougher policies are needed. sion are worth noting. A sound macroeconomic Inadequate enforcement does not provide the right environment is paramount. Economic stability and incentives. There are continued attempts to protect sustained economic growth should obviously be the institutions rather than the system. Every institution first and primary goal of public policy. Without is seen as a potential risk, as a potential threat, to long-term stability, any banking system could melt some kind of systemic problem. As long as that atti- down. tude persists, so too will banking problems. 95 Postscript Valeriano F. Garcia As noted in the introduction, this conference was Origins of Financial Crises organized to gain a better understanding of banking distress and crises, both common and costly occur- Unlike crises in individual financial institutions, gen- rences in Latin America. The main topics of the con- eralized financial crises typically stem from macro- ference were narrow banking, deposit insurance, economic disequilibria. But even though such and market-based regulation. Many attendees were disequilibria may trigger systemic crises, the inci- surprised by the discussion of narrow banking, hav- dence and depth of crises is strongly influenced by a ing considered it an issue of merely theoretical and range of institutional and regulatory elements, historical interest. When seen afresh, however-in including the asymmetries, rigidities, and extemali- the context of protecting the payments mechanism ties associated with fractional reserve banking; inad- without incurring the high costs of deposit insur- equate accounting and auditing standards; poor ance, moral hazard, and excessive regulation-nar- enforcement of prudential regulations; and lack of row banking is a timely and significant issue for early intervention and resolution of problem banks, distressed banking systems. The discussion of nar- as well as other forms of regulatory forbearance. row banking was important not only for countries Banking distress and crises usually have a gesta- that lack lenders of last resort (like Argentina), but tion period during which macroeconomic inconsis- for the entire issue of banking stability. Thus my dis- tencies build up, leading to a resolution that implies cussion here will focus more on narrow banking a shock (capital outflows, devaluation, recession, than on the other issues raised during the confer- and so on) to the system. In many cases creeping ence, since it is a potentially important reform that macroeconomic problems are related to time-incon- is opposed by most economists and almost all sistent exchange rate policy and domestic credit and bankers. fiscal policy. For example, a "soft" fiscal policy cou- Deposit insurance was a natural product of the pled with a pegged exchange rate is a well-known nonexistence of narrow banking systems. Larry recipe for disaster in the financial sector. This disas- Sjaastad clearly traced its linkage to fractional ter can occur with a hard monetary policy or a soft reserve requirements. Latin America's experience monetary policy. Monetary policy can influence the with reserve requirements has been quite varied. duration of the gestation period, but the core of the Some countries have had more than thirty different problem is a fiscal deficit that undermines the credi- required ratios, discriminating by bank size, type of bility of the exchange rate anchor. deposit, and location of regional headquarters. Until Such a combination of policies generates both the tequila effect Argentina, according to Roque high and variable real interest rates and high vari- Fernandez, was on the road to 100 percent reserves ability in the real exchange rate. These two effects are and narrow banking, while Mexico had already pro- deleterious for the asset side of banks' balance sheets. ceeded to zero reserve requirements. Dissension High interest rates lead to a larger share of nonper- over this issue has led to conflicting policy advice to forning assets and encourage bankers to roll over central bankers. bad loans ("evergreening"). The changes in relative Valeriano F. Garcia is principal economist in the Latin America and the Caribbean Technical Department at the World Bank. 97 Postscript prices caused by changes in the real exchange rate discussed in the conference paper by Roque also lower the quality of the financial sector's assets Femandez and Liliana Schumacher and supported portfolio, as Mexico's recent experience demon- by Larry Sjaastad and, to a degree, by Allan Meltzer. strates. A portfolio that is sound with one set of rela- Narrow banking is equivalent to a vaccine: even if tive prices could tum out to be sharply downgraded the virus is there, narrow banking protects the pay- with a different set. ments system. The Chicago plan for narrow banking automati- Perverse Asymmetry: Deposit Insurance, Reserve cally deals with the asymmetry problem. By impos- Requirements, and the Multiplier Effect ing a 100 percent reserve requirement on demand deposits, it resolvesboth the deposit insurance-moral The same policy that increases riskiness on the asset hazard quandary and the asymmetry problem. A side of banks' balance sheets-deposit insurance- narrow banking system imposes symmetry on the also has a perverse effect on their liabilities. Deposit remaining "broad" financial intermediaries' balance insurance causes banks with bad portfolios to pay sheet: any drop in asset prices is immediately mir- higher interest rates to depositors (and charge higher rored in the liability side and borne by investors, not interest rates to their risky clients) than other, less by the intermediary bank. risky banks. The moral hazard generated by deposit Several participants were of the view that the insurance causes the liability side of banks' balance Chicago proposal would only shift risk from banks sheets to increase in volume because, with deposit to nonbank financial intermediaries. In other words, insurance, depositors do not have to worry about the narrow banks rather than reduce risk would merely quality of their banks' assets. Consequently, as the reshuffle it. This contention misses a crucial point: market value of assets falls, the liabilities of the bank- the resolution of a banking crisis is entirely different ing system increase-and with them, the potential from the resolution of a capital markets crisis because for financial disaster. a banking crisis affects a country's payments mecha- Deposit insurance is a second-best solution to the nism. need to ensure confidence in the banking system and A banking crisis immediately threatens the stabil- prevent bank runs. And, as stated by Larry Sjaastad, ity of the payments system, increasing overall risk by deposit insurance is only needed because of frac- exacerbating macroeconomic problems through both tional reserve requirements, which generate a the potential fall in money supply and credit and the deposit multiplier. This multiplier creates euphoria potential fiscal impact. The October 1987 stock magnifying the monetary base increments resulting exchange crash in the United States had no direct fis- from capital inflows and wreaks havoc when the cal cost. Had it occurred in the banking sector, the monetary base decreases as a result of large outflows. consequences would have been quite different. Because deposit insurance creates a serious problem Furthermore, narrow banking is more equitable of moral hazard, it also creates the need for more reg- than conventional banking. In conventional banking ulation, better "policing" (supervision), and stronger the cost of crises is usually passed on to taxpayers enforcement. The sole factor distinguishing banking and, through the inflation tax, can cause perverse from any other ordinary business-such as the bak- income redistributions. In narrow banking the risk is ery business-is its unique feature of fractional transferred to the capital markets and is paid by reserve requirements. The bakery business experi- those who took the risk in the first place. enced only one episode of fractional reserves-when Some analysts have claimed that narrow banking Jesus multiplied the bread. And experience has requires well-developed capital markets. This is not shown that it would be dangerous for bankers to con- true. Consider some of the financial products devel- tinue wielding that kind of godly power. oped by the Islamic banking system that can easily be applied to narrow banking in developing countries. Is Narrow Banking the Answer? Islamic banking has been completely disregarded in the West because of its apparent refusal to accept inter- The perverse asymmetry problem can be tackled in est as an appropriate mechanism for the intertempo- two complementary ways. One is with consistent ral allocation of resources.' Some observers claim that macroeconomic policies that help avoid significant Islamic banking is a contradiction in terms because changes in relative prices, including real interest there cannot be banking without interest, while in rates. This approach evades the virus that causes the reality interest is well-embedded in that system. The sickness in the first place. The second approach is the Islamic system has developed some interesting prod- Chicago plan for narrow banking, an idea that was ucts, particularly its profit-loss sharing scheme. 98 Valeriano F. Garcia The Islamic system has two types of profit-loss in predicting changes in the ratio of cash to deposits sharing schemes: one between a bank's assets and its and in the aggregate ratio of reserves to deposits.3 borrowers and one between a bank's assets and its Some analysts have claimed that reserve require- lenders. Islamic banks place only a small share of ments can protect against the destabilizing effect of assets in the first arrangement. Most credit opera- capital flows if the central bank is willing to raise and tions are done with a mark-up arrangement that lower them in the face of capital inflows or outflows. serves as a disguise to charge ex ante interest rates. Although this proposal has some appeal, it creates The more interesting arrangement is the profit-loss more problems than it solves. Many countries (Brazil sharing scheme between a bank's assets and its lia- and Chile, for example) have attempted such steril- bilities. Here depositors share a percentage of the ization, mainly through open market operations and profit or loss generated by the bank's assets portfolio. reserve requirements. There are two major problems Even if on a very small scale, the existence of this with introducing this sterilization policy in a context type of product in countries like Turkey2 should dis- of fixed exchange rates. First, it increases domestic pel the claim that, to be implemented, narrow banks interest rates in the presence of capital inflows and need well-developed capital markets. A profit-loss reduces them in the presence of outflows, thus pre- sharing scheme between a bank's assets and it liabil- venting arbitrage from working. Sterilizing capital ities eliminates the perverse asymmetry found in inflows also generates huge interest rate losses in the conventional banking, and it may prove especially treasury. In this case the central bank accumulates suitable for agricultural finance. international rcserves at a real interest rate that is higher than it can earn on those reserves. Fixed Exchange Rates, Capital Flows, and the Mexico's recent financial crisis has shown the per- Money Multiplier verse effect of increasing domestic credit within a basically fixed exchange rate regime and then trying Another important advantage of narrow banking is to sterilize the capital outflows that were the natural that it makes the money multiplier equal to one. With outcome of that policy. The sterilization policy floating exchange rates it substantially improves (increasing money supply when facing capital out- control over the monetary stock, and with fixed flows) tried to protect the banking system from very exchange rates it substantially reduces the monetary high interest rates. Ironically, this policy eventually impact of capital flows. contributed to the bankruptcy of many banks and Countries like Argentina, with a fixed exchange threatened the entire banking system, creating huge rate, also benefit from narrow banking even if the quasi-fiscal deficits. money stock is endogenous in the long term. The One rationale for sterilization is that capital flows qualifier "in the long term" is important because are exogenous, determined by the "herd instinct" of large exogenous shocks have affected the Argentine foreign investors and the vagaries of international monetary base in the short term. Those shocks have interest rates. Thus, it is claimed, the domestic mar- then "multiplied" into argen-dollars (and peso ket should be protected from these exogenous deposits). Because the banking system was not nar- shocks. But sterilization would be inappropriate row, the endogeneity process proved more difficult. even if all shocks were exogenous. Moreover, not all When a dollar of capital inflows was, through the capital flows are exogenous, and it is impossible to multiplier, transformed into an inflow of several dol- know what fraction of flows is endogenous and lars, there seemed to be no problem. But problems which is exogenous. arose when a dollar of capital outflows was multi- A large portion of Mexico's 1994 capital outflows plied by the "irrational" (Sjaastad's term) fractional was endogenous. These outflows were caused by reserve system, causing a multiple reduction in the excess domestic credit (Garcia forthcoming), gener- money supply. ated by development banks and other government Narrow banking also permits better control of the expenditures. Adjustment came in the form of cur- money supply in countries that have floating rent account deficits and foreign exchange losses. exchange rates (or managed floating rates). In theory With confidence shaken, further exogenous outflows these countries can control changes in the money followed. multiplier with changes in the monetary base The counterexample is Argentina. Although there (through open market operations). In practice such a was some sterilization (a reduction in reserve require- move is difficult because of the inner lag in monetary ments and some rediscount), a convertibility rule policy (the lag between changes in the monetary base required that the bulk of adjustment come through and changes in the monetary stock) and difficulties interest rate increments. During January-May 1996 the 99 Postscript nominal stock of money was reduced by 20 percent. central bank policy. The controversy on whether Had Argentina chosen to completely sterilize the out- monetary and exchange rate policy should be man- flows, the convertibility plan would have collapsed. aged with rules or with discretion is not new. En his comments Roberto Zahler proposed "active discre- Prudential Regulation tion." He said that the central bank should not only try to fine-tune business cycles, but also that it should New Zealand's experience with market-based regu- have the power and flexibility to target the nominal lation, as presented by Peter Nicholl, shows that mar- monetary base as well as nominal and real exchange ket-based regulation can have many advantages over rates, nominal and real interest rates, international traditional regulation. Although some prudential reserves, and whatever else it deems necessary. regulation is essential to reducing the transactions Notwithstanding what academics have to say costs involved in gathering and processing informa- about the futility of central bankers trying to control tion by bank creditors, market-based regulation has real variables, Zahler asserted that in the real world several advantages over the discretion of official reg- markets are not to be trusted. Active discretion, he ulators. Market-based regulation is automatically said, would avoid major inconsistencies in the sys- enforced, not subject to corruption, and timely. tem that often generate large variability in relative Automatic enforcement guarantees equity-mean- prices. ing there is no exclusive treatment to special banks, no Zahler is correct in saying that good management "strategic" games played by the regulatory body, and is better than bad and inconsistent rules. But good no discretion granted to regulators on how and when and consistent rules are readily available and are bet- to discipline. When there are problem banks, the mar- ter than the best-intentioned management. Good ket determines how to discipline-through deposit rules are better because central bankers (and acade- withdrawal-and when to do it-immediately-in a mics) still know very little about the functioning of predetermined way. This approach should be scary countries' underlying economic model or about the only to bank directors and shareholders, and thus leads, lags, interactions, and variabilities of its should provide them with incentives to monitor the macroeconomic relationships. Although there have bank's financial and economic situation to avoid the been several important cases of bad and inconsistent ultimate punishment by the market. rules, there have been many more cases of bad and Market-based regulation is incorruptible because the inconsistent discretionary management. market has no special auditor or regulator that could One way to avoid asymmetries in the banking sys- be subject to bribes. This form of regulation is, by its tem is to avoid macroeconomic policies that generate very nature, a mechanism-and so is not subject to sharp and unexpected changes in relative prices fraud or malfeasance. (including interest rates). On that objective we all The delays created by regulatory forbearance are one agree; the difference, as noted above, is the of the main problems in resolving banking problems. instruments. Regulators try to avoid spillovers to the whole bank- ing system by engaging in forbearance and conceal- Notes ing "sensitive" information from the mnarket. They tend to wait and see if the problems can be solved 1. In fact, many Islamic banking products implicitly through special agreements or moral suasion. The introduce interest (for example, repurchase agreements). market acts in the opposite way: it acts as soon as 2. The share of Islamic banks in the Turkish bank sys- information is available, with the advantage that this tem is very small. discipline helps prevent problems from going from 3. Most countries that use the fractional reserve system bad to worse. make the system more "irrational" by imposing different reserve requirements according to the type of deposit. Rules or Discretion? Some countries also make it worse by changing required reserves according to bank size and region of operation. Several interesting issues were raised during the Under these conditions the aggregate reserve ratio is not roundtable discussion. Of particular interest is the wholly determined by the central bank, and is influenced subject of rules and discretion in the management of by changes in the public's portfolio choice. 100 Participants and Attendees Argentina Paul Bydalek Roque Fernandez President President Atlantic Rating Central Bank of Argentina Wadico Bucchi Jose Carlos Jaime Professor President University of Sao Paulo SEDESA Alvaro Antonio Zini Jr. Miguel Kiguel Faculty of Economics Subdirector General University of Sao Paulo Central Bank of Argentina Paulo De Tarso Medeiros Gerardo della Paolera Representative Professor Banco do Brasil University of Torcuato di Tella Sergio Werlang Fernando de Santibanies Director of Research and Asset Management President Banco de Bahia Investimentos, SA Banco de Credito Argentino Chile Osvaldo H. Schenone Rolf Luders University of San Andres Professor Liliana Schumacher Universidad Cat6lica de Chile Presidential Adviser Roberto Zahler Central Bank of Argentina President Bolivia Central Bank of Chile Jacques Trigo Loubiere Colombia Superintendent of Banks and Financial Entities Monica Aparicio Smith Superintendency of Finance Subdirector Jorge Patiflo Sarcinelli Banco de la Republica Executive Secretary Francisco Azuero FONDESIF Director Central Bank of Bolivia Fondo de Garantias de Instituciones Financieras Brazil Maria Luisa Chiappe de Villa Marcelo Bessan Baria Su per dent Partner Banking Superintendent KPMG Peat Marwick Superintendency of Banks Celso Boin Jr. Alberto Carrasquilla Bank Analyst Deputy Governor Banco de Investimentos Garantia, SA Banco de la Republica 101 Participants and Attendees Roberto Junguito Jamaica Director Winston Carr Banco de la Republica Deputy Govemor Costa Rica Bank of Jamaica Rodrigo Bolafios Zamora Mexico Executive President Sergio Ghigliazza Central Bank of Costa Rica Di-Pctor Felix Delgado Centro de Estudios Monetarios LatinoAmericanos Coordinator Luis Alberto Giorgio Proyecto de Reforma del Sistema Financiero Subdirector Fernando Herrero Acosta Centro de Estudios Monetarios LatinoAmericanos Minister of Finance Mauricio Naranjo Ministry of Finance Investment Economist Dominican Republic Economic Studies Mirtha Medrano De Rojas Banco de Mexico Director: Financial Department Javier Cdrdenas Rioseco Central Bank of the Dominican Republic Director Luis Manuel Piantini M. Intermediarios Financiero Privados Vice-Govemnor Banco de Mexico Central Bank of the Dominican Republic Nicaragua Ecuador Jose Evenor Taboada Santiago Bayas President General Director of Studies Central Bank of Nicaragua Central Bank of Ecuador OECS Alvaro Guerrero Errol N. Allen President of the Association of Private Banks Deputy Governor Previsora Bank Eastern Caribbean Central Bank Mauro Intriago Peru Superintendent of Banks Henry Barclay Superintendency of Banks Board of Directors Central Bank of Peru Oscar Armando Perez Merino Manuel Vasquez Perales Oscar Armando Perez Merino Superi-ntendent of Banks and Insurance Director of the Financial System Superintendent of Banks and Insurance Banco Central de Reservas Superitendency of Banks and Insurance Germany Switzerland Rolf Schinke Jozef Van' T Dack Professor Senior Economist University of Gottingen Bank for International Settlements Honduras United States Enrique Flores Valeriano Ricardo Caballero Aquino President Professor Commisi6n Nacional de Bancos y Seguro College of Journalism Central Bank of Honduras University of Maryland Fernando Vega M. Leonardo Auernheimer Proprietary Commissioner Professor Commisi6n Nacional de Bancos y Seguro Deptartment of Economics Central Bank of Honduras Texas A&M University 102 Recent World Bank Discussion Papers (continued) No 327 Agricuiltural Reform in Russia A Viewfrom the Farm Level Karen Brooks, Elmira Krylatykh, Zvi Lerman, Aleksandr Petrikov, and Vasilli Uzun No. 328 Insuring Sovereign Debt Against Default David F. Babbel No. 329 Managing Transboutndary Stocks of Small Pelagic Fish. Problems and Options. Max Aguero and Exequiel Gonzalez No 330 China. Issues and Options in Greenhoutse Gas Emissions Control. Edited by Todd M. Johnson, Junfeng Li, Zhongxiao Jiang, and Robert P Taylor No. 331 Case Studies in War-to-Peace Transition. The Demobilization and Reintegration of Ex-Combatants in Ethiopia, Namibia, and Uganda. Nat J Colletta, Markus Kostner, Ingo Wiederhofer, with the assistance of Emilio Mondo, Taimi Sitari, and Tadesse A. Woldu No. 333 Participation in Practice. The Experience of the World Bank and Other Stakeholders. Edited by Jennifer Rietbergen- McCracken No. 334 Managing Price Risk in the Pakistan Wheat Market. Rashid Faruqee and Jonathan R. Coleman No. 335 Policy Optionsfor Reform of Chinese State-Owned Enterprises. Edited by Harry G. Broadman No. 336 Targeted Credit Programs and Rural Poverty in Bangladesh. Shahidur Khandker and Osman H. Chowdhury No. 337 The Role of Family Planning and Targeted Credit Programs in Demographic Change in Bangladesh Shahidur R. Khandker and M. Abdul Latif No. 338 Cost Sharing in the Social Sectors of Suib-Saharan Africa: Impact on the Poor. Arvil Van Adams and Teresa Hartnett No 339 Public and Private Roles in Health: Theory and Financing Patterns Philip Musgrove No. 340 Developing the Nonfarm Sector in Bangladesh: Lessonsfrom Other Asian Countries. Shahid Yusuf and Praveen Kumar No. 341 Beyond Privatization: The Second Wave of Telecommunications Reforms in Mexico. Bjorn Wellenius and Gregory Staple No. 342 Economic Integration and Trade Liberalization in Southern Africa: Is There a Rolefor Souith Africa? Merle Holden No. 343 Financing Private Infrastruicture in Developing Countries. David Ferreira and Karman Khatami No. 344 Transport and the Village: Findingsfrom African Village-Level Travel and Transport Surveys and Related Studies. Ian Barwell No. 345 On the Road to EU Accession: Financial Sector Development in Central Europe. Michael S. Borish, Wei Ding, and Michel Noel No 346 Structural Aspects of Manutfactuiring in Suib-Saharan Africa Findingsfrom a Seven Country Enterprise Survey. Tyler Biggs and Pradeep Srivastava No. 347 Health Reform in Africa: Lessonsfrom Sierra Leone Bruce Siegel, David Peters, and Sheku Kamara No. 348 Did External Barriers Cause the Marginalization of Suib-Saharan Africa in World Trade? Azita Amjadi Ulrich Reincke, and Alexander J. Yeats No. 349 Suirveillance of Agriculltural Price and Trade Policy in Latin America during Major Policy Reforms. Alberto Valdes No 350 Who Benefitsfrom Putblic Education Spending in Malawi Resultsfrom the Recent Education Reform. Florencia Castro-Leal No. 351 From Universal Food Subsidies to a Self-Targeted Program: A Case Stuidy in Tunisian Reform. Laura Tuck and Kathy Lindert No. 352 China's Urban Transport Development Strategy Proceedings of a Symposiuim in Beijing, November 8-10, 1995. Edited by Stephen Stares and Liu Zhi No. 353 Telecommunications Policiesfor Sub-Saharan Africa. Mohammad A. Mustafa, Bruce Laidlaw, and Mark Brand No. 354 Saving across the World: Puzzles and Policies. Klaus Schmidt-Hebbel and Luis Serv6n No 355 Agricultture and German Reunification. Ulrich E. Koester and Karen M. Brooks No. 356 Evalutating Health Projects: Lessonsfrom the Literature. Susan Stout, Alison Evans, Janet Nassim, and Laura Raney, with substantial contributions from Rudolpho Bulatao, Varun Gauri, and Timothy Johnston No. 357 Innovations and Risk Taking: The Engine of Reform in Local Government in Latin America and the Caribbean. Tim Campbell No 358 China's Non-Bank Financial Institutions:Tritst and Investment Companies Anjali Kumar, Nicholas Lardy, William Albrecht, Terry Chuppe, Susan Selwyn, Paula Perttunen, and Tao Zhang No. 359 The Demandfor Oil Produtcts in Developing Coutntries. Dermot Gately and Shane S. Streifel THE WORLD BANK IXSI II Strcet, \ '\\ IVl"imle: 2)2 -477-1394 I tsiiCilt: 24S41 \VM()B\\K .\1 1744)s \(li II.'[ \Vol-l d ViAdC \\ cIl: littil: \ S.\ oddbill lk.01'g l:-uiiuil: l)Of)k4'''W (u">r-ItuS8k.Iorg ISBN 0-8213-3893-5