Research Observer EDITOR Shantayanan Devarajan, World Bank CO-EDITOR Gershon Feder, World Bank EDITORIAL BOARD Susan Collins,Georgetown University Angus Deaton, Princeton University Barry Eichengreen, Universityof California-Berkeley Emmanuel Jimenez, World Bank Benno Ndulu, World Bank Howard Pack, Universityof Pennsylvania Luis Serven, World Bank Sudhir Shetty, World Bank Michael Walton, World Bank The World Bank Research Observer is intended for anyone who has a professionalinterest in development. Observer articles are written to be accessible to nonspecialist readers; con- tributors examine key issues in development economics, survey the literature and the lat- est World Bank research, and debate issues of development policy.Articles are reviewed by an editorialboard drawnfromacross theBank and theinternationalcommunity of econo- mists. Inconsistency with Bank policy is not grounds for rejection. The journal welcomeseditorial comments and responses, which will be considered for pub- lication to the extent that space permits. On occasion the Observer considers unsolicited contributions. Any reader interested in preparing such an article is invited to submit a proposal of not more than two pages to the Editor. Please direct all editorial correspon- dence to the Editor, The World Bank Research Observer, 1818 H Street, NW, Washington,DC 20433. USA. The viewsand interpretationsexpressed in this journal are those of the authors and do not necessarily represent the views and policies of the World Bank or of its Executive Directors or the countries they represent. The World Bank does not guarantee the accuracy of data included in this publication and accepts no responsibility whatsoever for any consequences of their use. When maps are used, the boundaries,denominations, and other information 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. For more information, please visit the Web sites of the Research Observer at www.wbro.oxfordjournals.org, the World Bank a t www.worldbank.org, and Oxford University Press a t www.oxfordjournals.org. Research Observer . Volume 21 Number 2 Fall 2006 Macroeconomic Stability i n Developing Countries: How Much I s Enough? Peter Montiel and Luis Serven Banking and Regulation i n Emerging Markets: The Role of External Discipline Xavier Vives Access to Financial Services: A Review of the Issues and Public Policy Objectives Stijn Claessens What Have We Learned from a Decade of Manufacturing Enterprise Surveys i n Africa? Arne Bigsten and MansSoderbom I s There a Case for Industrial Policy? A Critical Survey HowardPack and KamalSaggi Subscriptions A subscription to The World Bank Research Observer (ISSN 0257-3032) comprises 2 issues. Prices include postage: for subscribersoutside the Americas,issuesare sent air freight. AnnualSubscriptionRate (Volume21,Zissues, 2006) Academiclibraries Print edition and site-wideonline access: US$115/£73/110 Print editiononly: US5109/£69/104 Site-wideonline accessonly: US5104/£66/99 Corporate Print edition andsite-wideonlme access: US$172/£109/ 164 Print editiononly: US$164/£104/156 Site-wideonlme access only: US5155/£98/147 Personal Print editionand individualonline access: US541/£29/44 Pleasenote: £Sterling rates applyin the UK, in Europe,US5 elsewhere. Readers with mailing addressesin non-OECDcountriesandin socialisteconomiesin transitionare eligible to receivecomplimen- tary subscriptionson request by writing to the UKaddressbelow. There may be other subscription rates available; for a complete listing, please visit www.wbro.oxfordjournals.org/ subscriptions. FnUprepaymentin thecorrectcurrencyisrequiredfor allorders.Orders areregardedas h, paymentsarenot refund- and able. Subscriptions are accepted and entered on a complete volume basis. Claims cannot be considered more than four months after publicationor dateoforder,whicheverislater.AUsubscriptionsin Canadaaresubject toGST.Subscriptionsin the EU may be subject toEuropeanVAT.If registered,please supplydetailsto avoid unnecessarycharges.Forsubscriptions that includeonlineversions,a proportionof the subscription price may besubject to UKVAT. Personalrates are applicable only when a subscriptionisforindividualuse and are not availableif deliveryismadeto a corporateaddress. Back issues: The current year and two previous years' issues are available from Oxford University Press. Previous volulnescan beobtained from the PeriodicalsServiceCompany,11MainStreet.Germantown, NY12526. USA. E-mail: psc@periodicals.com.Tel: (518)537-4700. Fax: (518) 537-5899. Contact information: Journals Customer Service Department, Oxford University Press, Great Clarendon Street, OxfordOX2 6DP.UK. Email:jnls.cust.serv@oxfordjournals.org.Tel:+44 (0)1865353907.Fax:+44 (0)1865 353485. In theAmericas,pleasecontact:Journals CustomerServiceDepartment. OxfordUniversityPress,2001EvansRoad, Cary,NC 27513, USA. E-mail: jnlorders@oxfordjournals.org.Tel: (800)852-7323 (toll-freein USAICanada)or (919) 677-0977. Fax: (919) 677-1714. In Japan,pleasecontact:Journals CustomerServiceDepartment,Oxford University Press,1-1-17-5F,Mukogaoka,Bunlyo-ku, Tokyo,113-0023,Japan.E-mail:okudaoup@po.iijnet.or.jp.Tel: (03)3813 1461. Fax: (03) 3818 1522. Postalinformation:The World Bank Research Observer(ISSN0257-3032) is publishedby Oxford University Press for the International Bank for Reconstruction and Development/m WORLD BANK. Send address changes to The World Bank Research Observer,JournalsCustomerServiceDepartment, Oxford UniversityPress, 2001 EvansRoad, Cary,NC 27513- 2009. Communications regarding original articles and editorial management should be addressed to The Editor, The World BankResearchObserver,The WorldBank, 1818 HStreet, NW,Washington, D.C. 20433, USA. Permissions:For information on how to request permissionsto reproduce articles or information from this journal, pleasevisit www.oxfordjournals.org/jnls/permissions. Advertising: Inquiries about advertisingshould be sent to Helen Pearson, Oxford Journals Advertising, PO Box 347, Ab'mgdon OX14 lGJ,UK.E-mail:helen@oxfordads.com.Tel: +44 (0)1235201904. Fax: +44 (0)8704296864. Disclaimer:Statementsof fact and opinion in the articles in The World Bank Research Observerare those of the respective authors and contributors and not of the International Bank for Reconstruction and Developrnent/~mWO RLD BANKor Oxford University Press. Neither Oxford University Press nor the International Bank for Reconstruction and Develop- m e n t / worn~BANKmake any representation, expressor implied,in respectof the accuracy of the material in this jour- ~ ~ nal and cannot accept any legal responsibilityor liability for any errors or omissions that may be made. The reader should make her or hisown evaluation as tothe appropriateness or otherwise of any experimental technique described. Paperused: The World Bank Research Observeris printed on acid-free paper that meets the minimum requirements of ANSIStandard 239.48-1984 (Permanenceof Paper). Indexingand abstracting:The WorldBank Research Observeris indexedandlor abskacted by ABI~INPORM, CABAbstracts, Current ContentslSocialand BehavioralSciences,Journalof EconomicLiteraturelEconLit,PAlSInternational,RePEc(Research in EconomicPapers),SocialServicesCitationIndex,and Wilson BusinessAbstracts. Copyright OTheInternationalBankfor Reconstruction and Developmentlm WORLDBANK AU rightsreserved;nopartofthispublicationmaybereproduced,storedina retrievalsystem,ortransmittedinanyformorby any means,electronic,mechanical,photocopying,recording,or otherwise without prior written permission of the publisher or a licensepermittingrestrictedcopyingissuedin the UK by theCopyright LicensingAgencyLtd, 90 TottenhamCourtRoad, LondonW1P 9HE,orin the USA bythe CopyrightClearanceCenter,222 RosewoodDrive,Danvers,MA01923. Macroeconomic Stability in Developing Countries: How Much I s Enough? Peter Montiel Luis Serv6n Over the 1990s macroeconomic policies improved in most developing countries, but the growth dividend from this improvement fell short of expectations, and a policy agenda focused on stability turned out to be associated with a multiplicity offinancial crises. This article examines the contents and implementation of the macroeconomic reform agenda of the 1990s. It reviews the progress achieved through fiscal, monetary, and exchange rate policies across the developing world and the effectivenessof the changing policy framework in promoting stability and growth. The main lesson is that more often than not slow growth and frequent crises resulted from shortcomings in the reform agenda of the 1990s. These concernlimitations in the depth and scope of the reform agenda, its lack of attention to macroeconomic vulnerabilities, and its inadequate attention to complementary reforms outside the macroeconomicsphere. For developing countries the 1990s were characterized by two major macro- economic developments: improvements in macroeconomic policies and a prolifera- tion of financial crises. Although macroeconomicpolicies as traditionally measured improved in most countries, the growth benefits expected from these better policies failed to materialize-at least to the extent anticipated by many observers-and a series of financial crises had adverse effects on economic growth and poverty in the countries involved. This article examines the relationship between these two devel- opments. It argues that slow growth and multiplecrises were symptomsof deficien- cies in the design and execution of the pro-growth reform strategy adopted in the 1990sof which macroeconomicstability was viewed as the centerpiece.' A useful way to characterize the interpretation of recent growth experience pro- posed here is from the perspective of Rodrik (2004)"growth strategies." He argued that well-established property rights, market-oriented incentives, fiscal solvency, and price stability are first-order economic principles that are necessary conditions for rapid economic growth. He stressed that these conditions can be implemented O The Author 2006. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development 1TEEWORLD BA NK.AU rights reserved.For permissions,pleasee-mail:journals.permissions@oxfordjournals.org. doi:10.1093/wbro/lkl005 AdvanceAccesspublication August 9,2006 21:151-178 through a variety ofinstitutional arrangements and identifiedtwo viewson whether establishing these conditions is also sufficient to accelerate economic growth. One view suggeststhat rapid growth is simply waiting to happen once the conditionsare met; the other contends that more proactive government policies are also required. Thisevaluationof macroeconomicreform's rolein growth takesthe secondperspective. It has four parts. First,improvementsin fiscalsolvency and price stabilitydid not occur in all devel- oping countries. As a consequence,macroeconomic instability continued to impede growth in some countries, and in several cases traditional macroeconomic imbal- ances resultedin crisesduring the 1990s that resembled thoseof the 1980s. Second, what matters for growth is the private sector's perception that fiscal sol- vency and price stability will be sustainable, which requires underlying fiscal and monetary rules and institutions to be reformed. Improved macroeconomic policy realizations were much more widespread than reforms in the rules and institutions governingmacroeconomicpolicyformation.Thelimited progressin reforming mac- roeconomic institutions likely undermined the contribution of macroeconomic policy improvements-evenwhen they could have been sustained ex post-to the stimulation of economicgrowth. Third, the macroeconomic stability sought through fiscal solvency and price stability was undermined by fragility caused by misguided reform policies in the domesticfinancialsystemand the capital account of the balanceof payments.These policies left many stabilizing economies highly vulnerable to adverse shocks and proved to be the Achilles' heelof macroeconomicstabilityin someof the most impor- tant crises of the 1990s. Fourth,and perhaps mostimportant,the growth payoff of macroeconomicstability per se may have been oversold. Fiscal solvency and price stability are conducive for growth because macroeconomic instability hampers investors' ability and willing- ness to undertake investment opportunities-understoodin the broadest sense of the term. But for macroeconomic stability to deliver growth, those opportunities must exist in the first place.In other words, macroeconomicstability may not be the binding constraint that prevents accelerated economic growth. Not only are well- defined property rights and market-orientedincentives also important for satisfac- tory growth, but if the proactive government view mentioned above is correct, a variety of other measures may also be required. In short, macroeconomic stability can help, but by itselfit cannot deliver growth. Unfortunately,gains in macroeco- nomic stability were often not complemented by the necessary growth-enhancing reformsin other parts of the economy. The rest of this article develops these arguments by examining the macroeco- nomic reform agenda of the 1990s. It first reviews progress in implementing the reform agenda during the past decade. It then evaluates the effectiveness of the reformsfrom an economic growth perspectiveand discusses how a policy agenda 152 TheWorld Bank ResearchObserver, vol. 21, no. 2 (Fall 2006) focused on macroeconomic stability was associated with a multiplicity of financial crises. It concludes by summarizing the lessons from the experience of the 1990s. The Facts of the 1990s Macroeconomic instability refers to phenomena that make the domestic macroeco- nomic environment less predictable. Unpredictability hampers resource allocation decisions,investment, and growth.2It can take the form of volatile key macroeco- nomic variables or perceived unsustainabilityin their behavior. This section evalu- ates developing countries' gains in macroeconomic stability during the 1990s, loolung separately at the behavior of macroeconomic outcome variables, policy variables,and exogenousshocks. Macroeconomic Outcomes For developing countries, growth rebounded in the 1990s from the depths of the 1980s,but it stillfellfarshort of thelevels achievedin the late1960s and 1970s and lagged behind the growth of industrial countries (figure1).Low-income countries did much worsethan middle-incomecountriesin the 1990s,showinglittleimprove- ment relative to the 1980s,whereas middle-incomecountries' growth rates in the 1990s wereroughlyon par with thoseof developedeconomies:much higher than in the 1980s but well below thosein the1960s and 1970s. What about the stability of growth outcomes?Developing countries have tradi- tionallybeen characterizedby moremacroeconomicinstabilitythan developedecon- omies,and there is a widespreadperception that globalizationhas madethe situation worse (IADB 1995; De Ferranti and others 2000; Easterly, Islam, and Stiglitz 2001; Rodrilc 2001). However, the volatility of key macroeconomic aggregates actually declined in the 1990s across the developing world. The standard deviation of per capita GDP growth fell from 4 percent in the 1970s and 1980s to about 3 percent in the 1990s, although it still remained significantly above the 1.5 percent seen in developed economies (figure 2).3Although the reduction in volatility of GDP growth was widespread,it was far from universal;of the 77 developingcountriesfor which completedata are availableover1960-2000,more than a third (27 countries)expe- riencedan increasein growth volatilityin the1990s relativeto the 1980s.~ Moreover, the reduction in aggregate output volatility concealed increasing extremeinstability.Largegrowthdisturbancesaccountedfor a largershareof overall instability in the 1990s than in previousdecades becauseof the increased contribu- tion of large negativeshocks(crises),which accountedfor closeto 25 percent of total growth volatilitycomparedwith14 percent in the 1960s and 1970s and 18 percent Peter Montiel and Luis Serven 153 - Figure 1. Median RealGDP Growth,byDecadeand CountryIncome Group (Percent) All (97) Developed (20) Developing(77) Middle-income Low-income (33) (41 ( W1961-70 $1971-80 Ea1981-90 01991-2000 1 Note: The sample comprises 97 countries with a population above 500,000 that have complete data on real GDP growth over 196C2000. The population minimum is set to exclude highly volatile island economies. Of the 77 developing economies, three (Hong Kong. China, Israel, and Singapore) are high-income, non-organisation for Economic Co-operationand Developmenteconomies. Source: World Bank (variousyears):Hnatkovska and Loayza(2004). in the1980s (figure3). Negativeextremeshocksalsoaccountedfor a largershare of the total volatility of gross national income and consumptionin the 1990s than in previous decades. Other key outcome variables commonly used as indicators of macroeconomic stability improved in the 1990s. For example, the median inflation rate across middle-income countries declined from a peak of 16 percent in 1990 to 6 percentin 2000. In low-income countries inflation peaked in 1994-95, after the devaluation of the CFAfranc, and then declined (figure4). Yet over most of the 1990s the gap between developed and developing country median inflation rates was substantial by the standards of the1960s and 1970s. Likewise, the prevalenceofhigh inflationin developingcountries peakedin1991 and then declined sharply. However,the decline took hold only in the mid-1990s, and thus the share of developing countries (among those with complete data) experiencing average inflation above 50 percent over the decade as a whole was unchanged betweenthe 1980s and the early1990s. Finally, current account deficits followed disparate trends in low- and rniddle- income countries.In middle-incomecountries the mediancurrentaccountdeficitto GDP 154 The World BankResearchObserver,vol. 21, no. 2 (Fall2006) Figure 2. StandardDeviation of per CapitaGDPGrowth,byDecade and Country IncomeGroup (Percent) All (97) Developed(20) Developing (77) Ibliddle-income Low-income (33) (41) Note: The sample comprises 9 7 countries with a population above 500,000 that have complete data on real GDP growth over 196@2000. The population minimum is set to exclude highly volatile island economies. Of the 77 developing economies, three (Hong Kong, China, Israel, and Singapore) are high-income, nonqrganisation for EconomicCo-operationand Developmenteconomies. Source: World Bank (various years);Hnatkovslta and Loayza (2004). ratio fell by about 1percentage point from the 1970s and 1980s. In part, however, this apparent improvement reflects the sudden stop of capital inflows to crisis- afflicted emerging market economies. In low-income countries, the deficit rose by about half a point to almost 5 percentin the 1990s (figure5). Policy Stability Conventionalindicators of policy stability also show a broad improvement over the 1990s. Most notably,the overall fiscal deficit fell acrossthe developingworld from a median value of 6-7 percent of GDP in the early 1980s to 2 percent in the early 1990s before rebounding to about 3 percent by the end of the decade.Thefiscalcor- rection wasparticularly pronounced among middle-incomecountries. However,the overallfiscalbalanceis affected by the impact of interest rate changes on publicdebt, whichis beyond the directcontrol of the authorities.Thus the primary balancelikely offers a more accurate measure of fiscal stance. Over the 1990s it shows a clear trend of increasing surpluses, particularly after 1995 (figure 6). By the end of the Peter Montiel and Luis Servin 155 Figure3. Decompositionof GrowthVolatilityinDevelopingCountries,byDecade(MeanPercentage of Total Volatility) 1W Normal BExtrerne BCrisis Boom 1 Note:Total volatility =normal + extreme; extreme =crisis + boom.Extremeshocks are definedas those exceeding two standard deviationsof output growth over the respectivedecade. Source: Authors' calculations based on data from Hnatkovska and Loayza(2004). decade the median developing country showed a primary surplus-although a much more modest one than that of industrial countries. Because of the diversity of monetary arrangements across developing countries and over time,it is more difficult to gauge monetary stability.One rough measure is the resort to seigniorage-that is, using money to finance deficits.Measured by the change in the monetary base relative to GDP, seigniorage collection rose during the late 1980s and early 1990s and then declined in both middle-income and (more modestly)low-income economies (figure 7), a pattern roughly similar to that of the inflationrate. The diversity of exchange rate arrangements across countries also makes it hard to gauge exchange rate policy. One indirect approach looks at the trends in real exchange rates, which are of course endogenous and subject to the influence of a variety of factors,includingthe nominal exchange rate. Shvets (2004)showed that real exchange rates depreciated over the 1990s in most developing economies. At the same time real exchange rate volatility (as measured by the standard deviation of its rate of change) showed a decline from the record high levels of the 1980s. But the declinewasliited tomiddle-incomecountries, and over the1990sdevelop- ing countries as a group exhibited much higher real exchange rate volatility than 156 The World Bank Research Observer, vol. 21,no. 2 (Fall 2006) Figure4. MedianInflationRate,by CountryIncomeGroup,1961-2000(GDP Deflator) L+~evelo~ed(20) +Developing (77)+Middle -income(41) -+IF.- how-income(33) ) ~~~~~ Source: World Bank (variousyears). Figure 5. MedianCurrentAccountBalance,byDecadeand CountryIncomeGroup(Percentof GDP) All (70) Developed (17) Developing(53) Middle-income Low-income(19) (32) 111966-70 1971-80 --- Ed1981-90 1991-24 Source: World Bank (variousyears). Peter Montiel and Luis Servdn 157 Figure 6. MedianPrimaryFiscalBalance,byCountry IncomeGroup,1990-2002 (Percent of GDP) (61) IMDeveloped (20) Developing (41) 1 Note:Data dier in source and coverage from those in figure 7; the figures are not strictly comparable. Source:Fitch Ratings (variousyears). industrial countries (Montieland Serven 2004). This high real exchange rate vola- tility partly reflected the high incidence of exchange rate crashes in the decade, when large devaluations were a frequent phenomenon (figure 8). Their incidence peaked in 1994, with the devaluation of the FA franc, and in 1998, with the East C Asian and Russian financial crises. Over the 1990s as a whole exchange rate crashes were slightlyless frequent than in the 1980s but much more so than in the 1960s and 1970s.~High real exchange rate volatility and frequent exchange rate collapses suggest that over the 1990s progress in achieving robust nominal exchange rate arrangements was limited. Summary Over the 1990s developing countries made notable progresson fiscal consolidation and limiting inflation. Improved fiscal and nominal stability helped attain a modest reduction in output volatility.These achievementswere also facilitated by a some- what morestableexternal en~ironrnent:~the volatilityof the terms of trade declined in all developingregions,in most cases to levels comparable to those of the 1960s, and capitalflow volatilityalsofell,although to a morelimited extent. But thesituationisfarfrom rosy.In termsof outcomevariables,developingcountries remain much more unstable than developed ones. Moreover, extreme volatility 158 The WorldBankResearchObserver, vol. 21, no. 2 (Fall 2006) Figure 7. DevelopingCountries' MedianSeigniorageRevenues,by CountryIncomeGroup, 1966-2001 (Percentof GDP) +Middle-income (34)+Low-income(30) 1 Source: IMP (variousyears);World Bank (variousyears). accounted for a larger share of total volatility in the 1990s than before, which is consistent with evidence suggesting that currency crashes and sudden stops in capital inflowsdid not tend to decline during the 1990s (Montieland Serven 2004). The situation is therefore one of dramatic policy improvementsin some areas, more moderate improvementsin the stability of macroeconomic outcomes,and persistent vulnerability to extreme macroeconomic events. The next section uses these find- ings to interpret the growth performance of developingcountries during the 1990s. Assessing the Experience of the 1990s The previous section has shown that macroeconomic policies and macroeconomic stability in developingcountries improved along severalimportant dimensions dur- ing the 1990s. These improvements were driven largely by the quest for higher growth. Yet as Pritchett (2004a)has argued, the growth payoff fell short of expecta- tions.To examinewhy,this section brieflyreviewsthe analytical linksbetweenmac- roeconomic stabilityand economic growth and then interprets the experienceof the 1990s in the context of that analytical framework. Peter Montiel and LuisServin 159 Figure 8. Share of DevelopingCountriesUndergoingExchangeRateCrises,1963-2002 (Percent of DevelopingCountries) Note: An exchange rate crisis is defined as in Frankel and Rose (1996):a depreciation of the (average) nominal exchange rate that exceeds 25 percent, exceeds the previous year's rate of nominal depreciation by at least 1 0 per- cent, and is at least three years away Gomany previouscrisis. Source: IMF(variousyears). From Stability to Growth Theory suggests that the link between macroeconomic policy stability and growth has three components.First,the directcontribution that policystabilitycan make to growth (byensuring that policyitself does not become an additionalsource of insta- bility) likely depends on the institutional setting, because what matters for invest- ment decisionsis not only whether policy realizations are favorable today but also the perceived likelihood that appropriate policies will be repeatedly implemented in the future. To have a significantimpact on growth, therefore,actual gains in macro- economic stability need to be viewed by the private sector as indicative of a perma- nent change in the macroeconomicpolicy regime. Second, the potential indirect contribution of policy stability to growth-by pro- moting the stability of macroeconomic outcomes-likely dependson the economy's degreeof macroeconomic kagility,that is, the extent to which even relativelyminor shocks can have large effects on the economy. On the one hand, fragility may make it too costly to deploy stabilization policies for the fear of potentially adverse effects, resulting in policy paralysis;on the other hand, fragility can mean that the instability 160 The World Bank Research Observer, vol. 21, no. 2 (FaU 2 0 0 6 ) that policy has to counter may become sosevere that feasible policy adjustments are unable to counter it. Third, as alreadystressed, growth does not depend only on macroeconomic stability. The effectivenessof stability in outcomesin promotingeconomic growth likely depends on a varietyof growthdeterminants,includingmicroeconomicfactorssuchasthede6ni- tionand enforcementof property rightsand the prevalenceof market-orientedincentives that are jointly requiredfor marketsto perform their allocativerole. The rest of this sec- tionevaluatesthereformagendaof the1990sfromthisthree-partanalyt'icalperspective. How Much Progress Was Really Achieved in the 1990s? As documented above,on the whole there were significant achievementsin tenns of stabilityinthe traditionalmacroeconomicpolicysenseduringthelate 1980sand early 1990s.But theseachievementswerenot universal,they werenot alwaysgroundedon solid institutionalfoundations to guarantee their permanence,and they rarely trans- lated intoa moreeffectiveuseof macroeconomicpoliciesasstabilizationinstruments. A useful framework within which to discuss these issues is the fiscal solvency condition, PV(T- G + dM) 2 B(O),which requires the present value (PV)of primary surpluses(T-G) andseignioragerevenue (dM)tobeat leastaslargeasthegovernment's outstanding stock ofnet debt. From a macroeconomicpolicyviewpoint,stabilityrequiresthe authoritiesto take a monetary and fiscal policy stance consistent with maintaining fiscal solvency at low inflation,whileleavingsome scope to mitigatethe impact of real and financial shocla on macroeconomic performance. Obviously, the first requirement imposes con- straints on the magnitudesof both the primary deficit and its moneyfinancing,while thesecond refers to the profilesof monetary and fiscalpolicyover the businesscycle. Most important, these requirements apply not only to current policies but also to future ones, as implied by the PV term in the expression. Indeed, one of the key dilemmas for macroeconomicpolicymakingis precisely how to ensure and convey to the private sector that future policies will abide by the requirements of solvency and low inflationwithout having to surrender the short-run stabilization capability of monetary and fiscal policy, that is, the tradeoff between credibility and flexibility. As discussed later, many of the achievements and disappointments of the 1990s relate to the search for lastingsolutions to this dilemma. Reassessing developmentsduring the 1990s in light of the aboveexpressionleads to six key observations: A comfortable perception of fiscal solvency has yet to be established in most countries. Improvedfiscal balanceshave often been achieved with stopgap measures or in ways inimicalto growth and welfare. Peter Montiel and Luis ServLn In many countriesfiscal policy remains destabilizing. . Lasting nominalstability remains to be crediblyestablished. The transition to robust exchange rate arrangements has been anything but smooth. The reformagenda proved to be incomplete. A comfortable perception offiscal solvency remains to be established in most countries. Despite the trend toward lower fiscal deficits documented earlier, public debt ratios remained high in most developing countries,showing little decline during the 1990s. For developing countries with available data, the median public debt to GDP ratio remained in the 50-60 percent range over the decade. A decline through 1997 was followed by a rising pattern, so that by 2001-02 the median developing country debt ratio exceeded the 1990-2001 level-as did the median industrial country debt ratio. On the whole, for the 46 low- and middle- income countries in the sample debt ratios rose in 24 countries and fell in 22 countries. This persistenceof high debt over the 1990s and its upward drift at the end of the decade have five main causes. First, improvements in fiscal performance,as mea- sured by reductions in primary deficits, were not ~niversal.~Second, in many cases the pressure of weak public finances on debt accumulation was aggravated by attempts at rapid disinflation,which implied a drop in deficit monetization.Without an equally rapid correction of the primary deficit,debt issuance was left as the only availablesource of financing.Empirically,this is confirmed by thefact that over the 1990s disinflationshows a statistically significant association with subsequent rises in debt ratios. Third, in several countries that did achieve a fiscal adjustment over the 1990s, most public debt accumulation reflected the cost of banking system bailouts.(Therealization of other contingent liabilities and the recognitionof hidden ones were also significant sources of debt accumulation in some countries, such as Argentina; see Mussa 2002.) Indeed, some of the banking crises of the 1990s- especially those in East Asia in 1997-ranked at the top of the historicalrecord in terms of fiscal impact.Fourth, where the bulk of publicdebt was denominatedin (or indexed to) foreign currency, large real exchange rate depreciations were another major factor behind the upward trend in debt stocks in the late 1990s.~A fifth factor behind persistently high debt was the high level of real interest rates in many countries, particularlyin the late 1990s,which largely reflected a lack of credibility in their stabilization efforts, documented below. Excessive reliance on short- maturity debt made some countries' overallfiscal outcomes-and thus their rates of public debt accumulation-highly sensitive to changes in domestic interest rates. Thus, in some countries (notably Brazil) high real interest rates contributed to a rapid pileupof publicdebt that furtherweakened perceptionsof solvencyand macro- economicstability. The World BankResearch Observer, vol. 2 1 , no. 2 (Fall 2 0 0 6 ) In terms of the solvency constraint introduced earlier, the bottom line is that through all these channels, increases in the observed value of the primary surplus T-G werenot enough to lower the public debt and establisha comfortableperception of fiscal solvency in many countries. A strong indication that solvency perceptions remained shaky in the 1990s is the fact that default risk premiums,as measured by sovereign borrowing spreadsin international marlcets,remained highly volatile for most emergingmarket economies(figure9).The evidencesuggests that default risk depends not only on debt burdensbut alsoon investors' perceptionsabout the quality of borrowers' policy and institutional framework (Kraay and Nehru 2003). Thus. the volatility of risk premiums likely reflected-among other factors-the market's lack of confidencein borrowers'commitmenttostability. But perceptions of high default risk are not just a symptom of perceived vulnera- bility;they alsoindirectlyunderminedurableoutcome-basedmacroeconomicstability by creating macroeconomicfragility.In particular they hamper countries' ability to conduct stabilizing policy: when default risk is perceived to be high and very sensi- tivetochanges in circumstances,attempts torun deficitsat timesof cyclicalcontrac- tion may be viewed with suspicion and result in large jumps in risk premiums (and thus borrowing costs), discouraging the use of countercyclical fiscal policy (seeCalderon, Duncan, and Schmidt-Hebbel 2003 for empirical confirmation). - - Figure9. EmergingMarketsBondIndexforLatii American and OtherBorrowers(BasisPoints) (-LatinAmerican Non-LatinAmerican 1 -I-- Source: JP Morgan(variousyears). Peter Montid and Luis Servin 163 Moreover, the scope for independent monetary policy can also be severely con- strained by the impact of changes in monetary stance on the cost of public debt through the associatedchanges in the nominal exchange rate and interest rate. Improvedfiscal balances have often been achieved with stopgap measures unlike& to be sustainable or in ways inimical to growth and welfare. Weaknessesinfiscaladjustment were not limited to the fact that increases in debt often offset improvements in primary surpluses. Often the improvements were likely to be perceived as purely temporary-because the measures behind them were transitory, because they reflected accounting transactions that had no effect on solvency, or because they directlycompromised future growth and welfare.In termsof thesolvencyconstraint above, such adjustments often had a significant impact on the current deficit but had littleeffect (orevenan adverseone)on the path of futuredeficits. In some instances, especially during the early part of the 1990s,fiscal adjust- ments reflecteda risein revenuesfrom a temporary boomin tax bases-for example, a consumption boom fueled by a transitory surge in capital inflows in an economy whose tax system was dominated by the value added tax. When the boom ended abruptly, a major fiscal gap opened in the recession. There is evidence that this mechanism had a significant role in some emerging markets in the 1990s (Talvi 1997). Elsewhere a variety of accounting measures improved conventional debt indicators without making substantive progress toward fiscal solvency. Common devices included one-timeasset sales to finance the retirement of publicdebt (which in principleimpliesno change in government net worth) and replacementof explicit debt with contingent liabilities (for example, granting debt guarantees rather than subsidiesto publicfirms).Measuressuch as these resultin improvementsin a bench- mark closely watched by investors and international financial institutions-gross public debt-but have no effect on solvency.In other words, they represent illusory fiscaladjustment (forexample,see Easterly1999 and Easterly and ServCn 2003). More generally,in many fiscal adjustment episodes, the focus on the quantity of adjustment wasnot matched by a comparableemphasison itsquality.The attention given to public spending composition and to its implicationsfor growth and welfare has often been limited.This disinterestsometimesresultedin adjustment at the cost ofbasic social needs-for example, by giving inadequate protectionto criticalsocial expenditures (m2003). More often than not, productive public expenditures (on such items as human capital formation and infrastructure)are compressed in the process of fiscal adjust- ment, mostly because the emphasis on cash deficits and debt discourages projects whose costs are borne upfront but whose returns accrue only over time. Such projects have the same impact on the government's short-term financing needs as pure consumption or any other spending item, even though their impact on sol- vency is quite different because, unlike consumption, they involve creating assets 164 The World Bank Research Observer, vol. 21,no. 2 (Fall 2006) that yield future revenues-be it directly or in the form of augmented tax collection resulting from higher output levels. Conventional fiscal aggregates (such as the primary or theoverallsurplus)closely monitoredby theinternationalfinancialinsti- tutions and investors ignore this distinction, and thus fiscal adjustment tends to have an anti-investment bias amply documented in both developed and developing countries (Blanchard and Giavazzi 2003; Easterly and Serven 2003). To the extent that reduced investment lowers growth and hence future tax bases, such bias can have adverseconsequencesfor growth-or evenfor fiscalsolvencyitself.The experi- ence of Latin America,wheredeclining publicinfrastructure spendingaccounted for the bulk of the fiscal correction achieved by some of the region's major countries in the 1990s,providesa good exampleof this perversedynamic. I n many countriesfiscal policy remains destabilizing. As is wellknown,fiscal policy in developingcountries tends to be procyclical,expandingin boomsand contracting in recessions.Empiricalestimatesshow that a1percentincreasein G P growth tends to D raise the growth rate of public consumption spending, for example, by about 0.5 percentage points in developing countries. The corresponding figure for industrial countries is much smaller (around 0.15),and for the largest of them (the Group of Seven countries), the responseof public consumption is actually negative (Talviand Vegh 2000; Lane 2003).By this measure, fiscal procyclicalityin developing coun- tries peaked in the 1980s and declinedsomewhat over the 1990s-but stillremains much higher than in industrial countries (figure10).Indeed,procyclicalfiscal policy played a keyrole in someof the major crisesof recent years,Argentina being a prime example (Mussa2002;Perry and Serven 2003). Lasting nominal stability remains to be credibly established. As shown earlier, devel- oping countries substantially reduced deficit monetization during the 1990s. But whether price stability can be sustained in many of them remains to be established. As the government's intertemporal budgetconstraint indicates,the rootsof inflation are ultimately fiscal. Thus, while a transitory reduction in deficit monetizationcan be achievedin a variety of ways,unlessdurable increasesin the primary surplus are somehow institutionalized, continuing pressures on the government budget will result in debt accumulation that will in turn create pressures for monetization. Indeed, during the 1990s, many countries' reductions in deficit monetization were not accompanied by lasting solutions to fiscal problems. In some cases (for example, Argentina, Brazil, Ecuador, Mexico, the Russian Federation, and Turkey) reduced inflation rates were achieved by stabilizing exchange rates. While improve- ments in price performance made reductions in money growth rates possible in these cases, sustainability remained questionablein all of them. In most cases con- tinued fiscal pressures were accompanied by real exchange rate appreciations and increases in real interest rates, leading to a pileup of public debt and calling into Peter Montiel and Luis Servkn 165 Figure10. CyclicalBehaviorofPublic Consumption,by CountryIncomeGroup,1980-2000 r+ -- Developing(41) - : Group of Seven +otherdevelopedilifl - - Note:Thefigureshowsthemedianof country-specificcoeficientestimatesobtainedby regressingthe rate of growth of publicconsumptionon the rateof GDPgrowth (plusa constant)over15-year rollingwindows. Source: World Bank(variousyearsj. questionthesustainabilityof thestabilizations.In Argentinaand Ecuador the lackof fiscaldiscipline led to the adoption of hard exchange rate pegs (a currency board in Argentina and dollarization in Ecuador) in the hope that they would somehow harden government budget constraints. Theirfailure to doso shows that such quick f ~ eare not enough to achieve lasting nominal stability without an independent s commitment to responsible fiscal policies. In this way, Brazil,Mexico, and Turkey's exchange rate-based stabilizations that relied on soft pegs eventually resulted in currency crises that gave way to short bursts of accelerated inflation. In view of this experience some countries adopted an alternative institutional arrangement during the 1990s,relying on an independent domestic central bank with a commitment to price stability. Like a fixed nominal exchange rate, such an arrangement worlcs in principle by committing the central bank to a low value of deficit monetization (dM),thereby imposing a hard budget constraint on the fiscal authorities and forcing them to adjust the primarydeficit (T-G) to the requirements of pricestability. For such an arrangement to be effective in promotinglasting price stability, the central bank has to be committed to price stability and able to resist pressureformonetizationfrom thefiscalside (thatis,it has to avoidfiscaldominance and achieve true independence from the finance ministry).But establishinga truly independentand effectivecentral bank has not been a straightforwardmatter either. The creation of independent central banks in Venezuela in 1989 and in Mexico in 166 TheWorld Bank ResearchObserver,vol. 21, no. 2 (Fall2006) 1993, for example, did not prevent the substantial political pressure for credit creation that contributed to currency crisesin thefirst half of the 1990s. How successful have developing countries been in creating a credible commit- ment to nominal stability? One way to infer the private sector's expectations for nominal stability is by observing its behavior, for example, the prevalence of dollar- ization. Since agents can partly protect themselves from nominal instability by denominating their assets in foreign currency, improved confidence in nominal stability should reduce dollarization, even though perceptions of nominal instability are not the only factor behind financial dollari~ation.~However, many developing countries remained heavily dollarized by the end of the 1990s, and the median degree of dollarization of bank deposits among low- and middle-income countries actually increased over the 1990s (IMFvarious years, 2002; Reinhart, Rogoff, and Savastano 2003). The contrast with high-income countries is stark: their much lower degree of depositdollarization showed little change over the same period. Ex post real interest rates may be another indicator: they tend to be high when actual inflation falls short of expectations and when inflation uncertainty is high. Although real interest rates declined in industrial countries during the 1990s, this was not the case in developing countries, where high real interest rates persisted and were higher at theend of thedecadethan at thebeginning (MontielandSemen 2004). As already noted, both dollarization ratios and ex post real interest rates reflect a variety of factors in addition to the perceptions of nominal instability, so this evi- dence is only suggestive. But other indicators point in the same direction. As an extreme example, the currency premium on the Argentine peso was positive throughout the 1990s and became very large at times of turbulence despitethe sup- posedly irrevocable peg to the dollar enshrined in Argentina's Convertibility Law (Schmukler and Serven 2002). The transition to robust exchange rate arrangements has been anything but smooth. Price stability refers not only to stability in the purchasing power of domestic currency over goods and services but also to an appropriate level of purchasing power over foreign exchange. However, recent progress toward robust exchange rate regimes has been uneven in developing countries. Indeed, it probably was an early casualty of thesearch for macroeconomicstability. As already discussed,many countries adopted exchangerate-basedstabilization strategies as a supposedlyquick recipefor disinflation. These not only meant the adoption of single currency pegs but also made such pegs very difficultto adjust, since the credibilityof the entire stabili- zation program was tied up with the stability of the peg. In effect, the defense of the peg sometimes became an end in itself,even when it was evident that it had outlived its usefulness. More flexible exchange rate arrangements-that is, arrangements lacking a pre-announced peg, with or without extensive central bank intemention- have too often been adopted only in the aftermath of currency crises. Peter Mantiel and Luis Servdn The late 1990s showed that hard exchange rate pegs-that is, dollarization and currency boards-are not a speedyshortcut tofiscalorthodoxyand nominalstability in lieu of the slow and painful buildup of credibility required when countries rely on an independent monetary policy. In particular, the Argentine episode showed the threat to stability posed by inflexibleexchange rates, which made adjustment to real disturbances exceedingly difficult. These shackles eventually undermine the sus- tainability of such rigid arrangements. Though less well lmown, the experience of the CEA franc during thefirst half of the 1990s is another exampleof this situation. The reform agenda proved to be incomplete. The preceding observationssuggestthat, as far as fiscal solvency and price stability are concerned, the reform agenda of the 1990s leftmuch to be desired. But the agenda was also deficient in its very design, becausethe macroeconomicstability that fiscal solvency and price stability are sup- posed to deliver was undermined by leaving in p l a c m r worse yet, creating- important sourcesof macroeconomicfragility. Aparticulararea of fragilityin whichthe policy-basedstabilityagenda wasincom- plete is financial sector soundness. While research shows that an efficient domestic financialsystem is important for growth, the experienceof the 1990s strongly sug- gests that a sound one is indispensablefor macroeconomicstability. The macroeco- nomicreformagenda of the early1990s was incompletein that the central roleof the financial system for macroeconomic stability was often ignored-even though it should have been clear in light of the Southern Cone crises of the early 1980s. Thus to the standard policy-orientedprescriptionsfor stability-a solventfiscalstance, low and stable money growth, and robust exchange rate policies that neverthelessallow adjustment to shocks-it isnecessaryto add policiesthat fostera sound financialsys- tem.Indeed,in the wakeof the crisesof the1990s the IMT redefineditscorecompeten- cies to includefiscal,monetary,exchangerate,and financialsector policies. Stability in this particular sense-that is, ensuring a sound domestic financial system-was clearly not widelyachieved by developing countriesduring the1990s. As a result, an important source of macroeconomic fragility was not only left in place but may have even been magnified, for reasons to be explained. Inadequate attention to financialsector soundness often resulted in a domestic economicenvi- ronment in which institutional problems involving moral hazard were rife, render- ing both public and private balance sheets highly vulnerable to changes in the environment (interestrate and exchange rate changes) and posing a major obstacle to outcome-based stability in several major countries.1°The proliferation of fman- cia1 crises in the 1990s reflects in part this missing piece of the reform agenda. Indeed, the incidence of systemic banking crises was even higher in the 1990s than in the 1980s,particularlyin thesecond half of the decade (Bordoand others 2001). But the frequency and the severity of crises were also affected by an important change in theeconomicenvironment-namely,increasedcapitalmobility.Thiswas 168 The World Bank Research Observer. vol. 2 1 , no. 2 (Fall 2006) another key source of fragility, making economies vulnerable to sudden shifts in capitalflows.In fact, the combinationof unsound policiesin thefinancialsector and open capital accounts helps explain many characteristicsof the crises of the 1990s. First, many of these crises were twin crises, simultaneously involving currency and banking collapses, often characterized by banking problems preceding a cur- rency crash, which then fed back into a full-blown financial crisis (ICaminskyand Reinhart 1999).There is evidence that twin crises are usually much more costly in terms of output than standard banking-only or currency-only crises (Bordo and others 2001). Second, many of these crises proved hard to foresee on the basis of standard macroeconomic imbalances. The hardest--especially the Mexican and East Asian crises-occurred where the main vulnerabilities concerned financial, rather than macroeconomic,variables and took the form of balance of payments runs similar to traditional bank runs.'' Third,many of thesecrises weresurprisingly severe.The deepest onesinvolved seriousproblemsin the financialsector (EastAsia, Ecuador,Mexico, and Turkey),in private sector balance sheets (Argentina and East Asia), and with fiscal insolvency (Argentina and Ecuador). Where none of these problems was present and events took the form of a simple currency crash (Brazil), crisis-inducedeconomic contraction was not as severe. The Growth Payoff While the improvements in macroeconomicpolicies were limited-as the preceding discussion has shown-growth rates haveindeed risen relative to the 1980s in many developingcountries.The achievement is only a modest one, however, since growth in the 1980s was generally low, and for most countriesgrowth rates over the 1990s remained well below those over the 1960s and 1970s. Indeed, of the 77 developing countries with complete data, only 28 achieved growth rates over the 1990s that exceededthoseoverthe1970sand only 24 achieved growthratesthatexceededthose over the1960s. But is this growth payoff commensuratewith the progresson macro- economic stability,or is it disappointing?There are severalreasonsto believe that the growthpayoff wasindeedcommensuratewith what was actuallyachievedby reform. First,asargued above,the growth payoff from macroeconomicstabilitydependson its perceived permanence.But, asdiscussedin the previous section,often progress on macroeconomic stability was based on policy changes that were not perceived to be durableor that failedto reformthe institutionsmakingmacroeconomicpolicy.In this sense,the growth payoff expected from the stability that was actually achieved may have been overstated.Moreover,a vicious circle may have taken hold in some coun- tries,with the social consensusthat made the policiespossible-and that is needed to makethem sustainable-faltering in the absenceof a fairly prompt growth payoff. Second,the search for macroeconomicstability-narrowlydefined-may in some cases have actually been inimicalto growth. As already noted, a preoccupationwith Peter Montiel and Luis ServJn 169 reducing inflation induced some countries to adopt exchange rate regimes that ulti- mately conflictedwith outcome-basedstability.In other cases, as shown previously, a single-mindedpursuit of macroeconomicstabilitymay have come at theexpenseof growth-enhancing policies (forexample,an adequate provision of public goods)and social investments that might have both increased the growth payoff and made stability more durable. From this perspective, some economies may well have been overstabilizedin both microeconomicand macroeconomicsenses. From a microeco- nomic perspective the presumed stability gains from further fiscal adjustments may not have justifiedthecostsof forgoingkeysocialand productiveexpenditures. Froma macroeconomic perspective the narrow focus on stability may have precluded more progress toward countercyclical policies. The contrast between the significantfiscal adjustment achieved by most developing countries and the persistenceof outcome- based instabilitysuggeststhat thisfactor may have been important. Third, aside from whether the search for macroeconomic stability worked at cross-purposes with the search for higher growth, the incomplete macroeconomic reform agenda failed to bring about the reduction in macroeconomic fragility required to fully translate policy-based stability into outcome-based stability. Although overall macroeconomic volatility decreased among developing countries, extreme volatilityactually rose during the 1990s,reflecting largely a spate of crises during the decade. Moreover, the adverse impacts of extreme volatility on growth appear to exceed those of normal volatility.12Thus,the growth payoff of the macro- economic policy improvementsachievedin the 1990s was limited not only by their weak institutional underpinnings but also by theextreme outcome-basedinstability that emergedduring the decademainly as a result of thefragilitiesoverlooked by an incompletereformagenda. Fourth, as argued earlier, while macroeconomic stability may facilitate growth when other forces are driving the growth momentum-that is, when macroeco- nomic instabilityis the binding constraint on growth-macroeconomic stability is not enough to drive the growth process when other essential ingredients are lacking-that is, when other constraints are binding. These constraints involve the various policies and institutions that shape the opportunities and incentives to engagein growth-enhancing activities(Pritchett 2004b).Theyincludesecure prop- erty rights and market-oriented microeconomic incentives as well as proactive government interventionsto overcomeinformationalexternalitiesand coordination failures (Rodrik 2004). The importance of these complementary factors may not have been sufficientlyappreciated earlyin thedecade. In sum, there is little reason to expect a simple direct association between macro- economic stability and growth, even if stability as measured by commonly used macroeconomic policy indicators is achieved. From this perspective, the limited growth payoff that emerged from the gains in macroeconomic stability achieved during the 1990s may not be all thatsurprising. 170 The World BankResearch Observer, vol. 21, no.2 (fiU2006) Summary: Lessons from the 1990s This article does not take a position on why the implementation of reform in devel- oping countries during the 1990s proved to be flawed in the ways emphasized. Perhaps, as others have suggested, the problem was the absence of a clear order for reform priorities in a context where developing countries have limited political and administrative capacity for reform (Hausmann, Rodrik, and Velasco 2005). This may account for many countries' inability to put in place theinstitutional underpin- nings of macroeconomic stability in cases where macroeconomic instability was indeed the binding constraint on growth, and for other countries' failure to implement microeconomicreformswith a larger potentialgrowth payoff when mac- roeconomic instability was not the binding constraint. This section instead draws somelessonsfrom the reformexperienceasimplemented. This experience offers several lessons for the future. A central one is that the old verities concerning the importance of macroeconomic stability still hold true. While macroeconomic policy realizations are not all that matter for promoting economic growth, they clearly do matter. Perceived fiscal insolvency, high and unstable infla- tion, and severely overvalued real exchange rates remain reliable ingredients for extreme instability and slow growth. But the 1990s also showed that in addition to macroeconomic policy realizations, three other ingredients are critical for growth: (a) the institutional framework within which fiscal, monetary, and exchange rate policies are formulated; (b) the degree to which macroeconomic fragilities are avoided; and (c)the extent to which complementarypro-growthreforms are imple- mented.Theseelementsare reviewed briefly below. Institutions for Macroeconomic Policy Formulation The institutional context in which traditional macroeconomic policies are formu- lated is critical when resolving the tradeoffbetween policy credibility and flexibility. Both are requiredfor the durable outcome-basedstability that ultimatelymatters for economicgrowth. Procyclical fiscal policies arise in developing countries because without strong budgetaryinstitutions a "tragedyofthe commons" phenomenon setsin during pros- perous times, when government revenues are high: because no claimant on the government's budgetary resources has an incentive to internalizethe need for fiscal solvency, political imperatives cause the government to spend all its resources (and even toborrow)during booms,leavinglittlemargin of solvencytodraw on tofinance fiscaldeficits when times are bad. A mechanism that makes it politically possible to ensure prudent fiscalresponsestofavorableshocksis requiredin this context. The speciGcmechanism best suited to the job depends on countrycircumstances.It may require, for example, reforming budgetary institutions to centralize budgetary Peter Montiel and Luis Servin 171 authority in the finance ministry rather than in line ministries or the parliament or implementing fiscal rules that forceclaimants to government resources to respect the government's intertemporal budget constraint. These transparent fiscal rules may be embodiedin thecountry'sconstitutionorsubjecttochangeonlybylegislativesuperma- jorities,withpenaltiesstipulatedfornoncompliance(Perry2003).Alternativeproposals have focused on independent fiscal policy councils to set annual deficit limits, modeled alongthelinesofindependentcentralbanks.However,suchinstitutionalarrangements need to balance credibility and flexibility.Simpler rules may be more transparent and thus more easilyverfiable,but they need to be designedto allowsufficient flexibilityfor fiscal policy to react to a changing economic environment. Overly rigid rules are unlikely to be sustainable or credible--as shown by the increasingpressures to revise theEuropeanStabilityPact becauseof itsneglectof the macroeconomiccycle. With respect to monetary policy and exchange rate regimes, the evidence indi- cates that low and stableinflationis conducive toeconomicgrowth, and theorysug- gests that it is most important in this regard for the private sector to be convinced that low and stable inflation is a permanent feature of the economic environment. Asin the case of fiscal credibility,an appropriateinstitutional underpinning for price stability is required to generate such a perception.However,as shown earlier, a key lesson of the 1990s was that purely monetary arrangements cannot achievesuch a perception,because they are not sufficient todisciplinefiscal policy. Fiscalcredibility is a necessaryconditionfor monetary credibility,and not even the most rigid mone- tary arrangements (a currency board or de jure dollarization) guarantee hard government budget constraints. In short, no institutional shortcuts of a purely monetary nature can achieve credible price stability. Instead, the task is to choose monetary arrangements that can best complement reformed fiscal institutions in achievinga desirabletradeoff betweencredibility and flexibility. Again, the optimal institutionalresponses to this challengeare likely to be country specific.Theexperienceof the1990s,however,suggeststhat a monetaryarrangement oftensuitableforthistaskfeaturesan independentcentralbankthat operatesafloating exchange rate and commits to a publicly announced inflation target. This arrange- ment has the important advantages of flexibility (since the central bank is not con- strainedin howit attainsitsinflationtarget)andcommitment(sincethecentralbank's prestigeispubliclyput on theline byannouncing such a target).Mostimportant,float- ing exchange rates and inflation targets allow the domestic authorities to establish anti-inflationary credibility the hard way-that is, by establishing a track record- rather thanby attemptingtoimport it through someformof exchange ratepeg.13 Robustness: The Scope of the Macroeconomic Reform Agenda The proliferationof crises during the 1990s has made it clear that beyond an appro- priate institutional setting for formulating fiscal, monetary, and exchange rate 172 TheWorld Bank Research Observer,vol. 2 1, no. 2 (FaU 2006) policies the reform process in developingcountries also needsto attend to robustness issues to achieve the stable macroeconomicenvironment sought through fiscal and monetary reforms.Thisincludes,in particular, policiesdirectedtoward the domestic financial system and capital account, which have been shown to have important implications for macroeconomic fragility and thus for outcome-based macroeco- nomicstability. As has been widely recognized, an appropriate institutional framework in the domesticfinancialsectorinvolves(a)clearand secure propertyrights; (b)an accessi- ble, efficient,and impartial legal system to enforce contracts; (c) appropriate legal protectionforcreditors;(d)well-specifiedaccounting and disclosurestandards; (e)a regulatory system that screens entrants while encouraging competition, imposing adequate capital requirements, and preventing excessively risky lending; and (f) a supervisory system that can effectively monitor the lending practices of domestic financial institutions. The key lesson is that the pace of liberalization for domestic financial systems that have not already been liberalized should be modulated to reflect the quality of the institutional framework governing the domestic financial sector and that improvingthe qualityof thisframeworlrdeserves high priorityin the macroeconomicreformagenda. In contrast to the consensus that has emerged on domestic financial reform, managing a county's integration into international financial marlrets remains a controversial part of the institutional agenda. Despite strong theoretical arguments concerning the gains from capital account openness, the empirical evidence on whether it has in fact been conducive to enhanced growth and reduced consump- tion volatility is inconclusive(ongrowth,see Edison and others 2002; on consumption volatility,see Kose,Prasad, and Terrones2003).Thedesireto avoid macroeconomic &agilitymakes a strong case for institutional arrangements for the capital account that at least prevents maturity mismatches in a country's external balance sheet, sincesuch mismatchescan leave the country vulnerable to creditor runs analogous to bank runs. These runs played a key role in the East Asian crisis (Rodrik and Velasco 1999). The question is how to preclude them. One difficulty is that short maturities are attractive to creditors as a means of monitoring borrowers and con- trolling their behavior precisely when asymmetric information and moral hazard problems are serious.Under these circumstances,therefore,short-maturity borrow- ing will arise endogenously because it will be substantially less costly to borrowers than long-term loans. Mismatches may reflect not only an inadequate borrowing strategy but also the reluctance of investorstolend long-term in theface of a macro- economic financial framework deemed suspect. The problem is, of course, that voluntary short-maturity loans between private partiesfail to take into account the socialcosts associated with the risk of creditorruns. Again,there are alternative ways of meetingthis challenge.One way todeal with maturity mismatches is for the public sector to accumulate large stocks of foreign Peter Montiel and Luis Serven 173 currency (foreignexchange reserves) to offsetliquid liabilitiesincurred by the pri- vate sector. This approach is being pursued by many countries, but it is expensive. Holding large volumes of low-yielding,short-term assets instead of (illiquid)long- term investment entailsseriousopportunity costs and leaves in place incentives that give rise to short-term borrowing. Under this strategy the costs of insuring against creditor runs are ultimately borne by taxpayers. An alternative is to discouragethe privatesector Gomincurring short-term external liabilitiesin thefirst place through restrictionson short-term capital inflows or to make those liabilitieseffectively less liquid at times of crisis through restrictionson short-term capital outflows.Because both of these policies tend to increase the cost of short-term loans, they effectively operate by internalizing the systemiccosts associated with the risk of creditor runs. The available evidence suggests that inflow restrictions such as unremunerated reserve requirements (for example, the Chilean encaje) tend to have no significant effect on the overall volume of inflows but do affect their composition,reducing the share of short-term flows in the total. This is because a uniform reserve requirement is more onerousforshort-termtransactionsthan for others (seeMontiel and Reinhart 1999 for review of the cross-countryevidence on the effectivenessof inflow restric- tions). By contrast, evidence on the effects of restrictions on outflows is much less conclusive. Overall, it appears that restrictions on short-term capital inflows may have a role in achieving outcome-based macroeconomic stability in developing countries. In addition to maturity mismatches,external borrowing may also create fragility in the form of currency mismatches. Developing countries may suffer from "originalsinv-- inabilitytodenominateexternalborrowingintheirowncurrencies- an which causes exchange rate risk to be borne by domestic residents rather than by foreign residents (Eichengreenand Hausmann 1999).In the short run the key is to promote the efficient distribution of this exchange rate risk within the domestic economy by ensuring-throughregulatory means-that it is appropriately priced and therefore borne by those best able to bear it (typicallyagents holding foreign currency assets-including exporters---or those with a high degree of risk toler- ance).In the longer term a larger role in ameliorating the problem of currency mis- matches would be assumed by institutional changes that promote crediblenominal stability,thus mitigating exchange rate risk. This perspective is consistent with the experienceof a few emergingeconomies-such as Chile,Mexico,Poland, and South Africa-for which domesticcurrency-denominatedexternal borrowing is becoming a reality,evenif only marginallyat first. ComplementaritiesAmong Pro-Growth Policies But most of the burden of jumpstarting growth in developing countries must fall on pro-growth policiesoutside the macroeconomicarena. Aside from the fundamental 174 The World Bank Research Observer, vol. 21. no. 2 (Fa112006) policies,securing property rights, and establishingmarket-orientedincentives,such policies may include, for example, an open international trade regime, national innovation policies, well-functioning factor markets, and an investor-friendlylegal and regulatory environment. In some cases these policies actually facilitate reforms aimed at macroeconomicstability-for example,disinflation or the correction of a real misalignment are easier and less costly to achieve with well-functioning labor and financial markets. The key lesson is that policies of this type are mutually complementary with policies that focus on creating and preservingmacroeconomic stability.An unstable macroeconomicenvironment tends to undermine the growth benefitsof such policies. Nonetheless,the 1990s showed that macroeconomicstabil- ity alone is not enough: policies outside the macroeconomicarena are indispensable for harvesting the fruits of macroeconomicstability in the form of sustained high rates of economic growth. Notes Peter Montiel is a Professor of Economics at Williams College;his email address is peter.j.montiel@ williams.edu. Luis Servknis Research Manager for Macroeconomics and Growth in the Development Research Groupof the WorldBank; his email address is lserven@worldbank.org. Thisarticle was origi- nally prepared as a background paper for the volume "Lessons from the 1990s." The authors are indebted to Roberto Zagha for numerous comments and discussionsthroughout the preparation of the article. They also thank, without implication, Indermit Gill, Jim Hanson, Ricardo Hausmann, RaviKanbur, Phil Keefer, Guillermo Perry, Lant Pritchett, S. Ramachandran, and three anonymous reviewers for valuable comments on previous versions. Patricia Macchi provided excellent research assistance. I.Easterly (2001)also viewed the crises of the 1990s as a symptom of, rather than an explanation for, the slow growth of the 1990s. 2. In recent years, there has been renewed interest, sparked by Ramey and Ramey (1995),in the adverse effects that real and nominal instability can have on economic growth, as documented by a growing empiricalliterature on the subject.For a recent evaluation, see Hnatkovska and Loayza (2004). 3. The decline in developing country volatility over the 1990s is also documented by De Ferranti and others (2000), Rodrik (2001), and Hnatkovslta and Loayza (2004). The same result holds if volatilityis measuredby a robuststatistic such as the interquart'derange instead of the standard deviation. The declinein volatilitywas statistically significant:formal tests strongly reject the hypothesis that the cross-country distribution of growth volatility did not change between the 1980s and 1990s and the hypothesis that the changes in volatilityacross the two decadesare centered at zero. 4. The declinein aggregate volatilityalso extends to other variables more directly related to individ- uals' welfare,such as income and consumption growth, although to varying extents. In particular, the volatility of private consumption growth declined relative to the previous decade but mainly in low- income countries; in middle-incomecountries consumption volatility remained virtually unchanged from the record highs of the 1980s. 5. In a smaller country sample (whose t i e coverage ends in 1997),Bordo and others (2001) also found that the frequency of currency crashes declined in the 1990s compared with the preceding 15 years. 6. The evidencefor increased stabilityof the external environment is presented in the working paper version of this article (Montieland SemCn2004). which also includes additional figures. PeterMontiel andLuisSemen 175 7. In hdia, for example,continuing large primary deficits(averagingclose to 4 percent of GDPin the late 1990s)were the main factor behind persistently high debt ratios. Moredramatically,fiscalvulner- abilities had a prominent role in some of the major recent financial crises: the Russian Federation in 1998, Ecuador in 1999, and Argentina in 2002. In Argentina, the expansionary fiscal stance taken during the 1995-97 boom left authorities virtually no room to adjust to the global real and financial slowdown after the Russian crisis of 1998 and to the real appreciation of the peso under the hard dollar peg (Perry and Semen 2003). 8.In Argentina and Uruguay,for example,the 2002 exchangerate collapsemore than doubledthe debt to GDP ratio-from 50 percent of GDP to more than 140 percent in Argentina andfrom 40 percent to more than 80 percent in Uruguay. Acrossemerging marketsdebt dollarizationremained pervasive: the median ratio of foreign currency debt to total public debt rose over the late 1990s to exceed 55 percent by 2001. 9. The degree of real dollarization and the perceived stability of the real exchange rate also matter, as do fmancialsystem regulations and the availability of other assets that shelter investors from nomi- nal instability, such as instruments indexed to domestic inflation, as in Chile, or short-term interest rates, as in Brazil(IMF 2002; De la Torre and Schmukler 2003). Thus the interpretation in the text should be taken as suggestiverather than as conclusive. 10. Ironically,under these circumstances incipient progressalong conventional dimensionsof mac- roeconomic stability,such as disinflation,may even have made financial crises more likely. For exam- ple, the use of the exchange rate as a nominal anchor may have encouraged agents to ignore exchange rate risk and in the case of hard pegs,as in Argentina, may have made it more difficultfor regulatorsto induce financial institutions to factor such risk into their portfolioallocations without raising fears of a possibleabandonment of the peg. 11. The recent analytical literature on crises continues to stress weak fundamentals as a prerequi- sitefor the occurrence of crisesbut emphasizesthe key role of ingredients such as self-fulfillingexpecta- tions and multiple equilibria in triggering them. These views assign an increasingly important role to financial system imperfectionsin full-blown balance of payments crises (Krugman1999). 12. There are good reasons why crisis volatility (due to large adverse shocks)should entail greater growth wsts than normal volatility.On the one hand, with a given set of risk management mechanisms, large shocks may be more difficultto absorb than small ones. These threshold effectsof volatilityhave been found to be empirically relevant to investment (ServCn 2003). On the other hand, asymmetries built into the economy mean that negative shocks have qualitatively different consequences than positive ones. A clear example is buffer stoclrs-for example, bank liquidity or international reserves. Largeadverseshocks(ora succession of small negative ones)can exhaust them and triggeran adjustment mechanism verydifferentfrom theone associatedwith positivedisturbances.The same applies to firms' net worth: onceit becomes negative,adjustment proceedsthrough bankruptcies,with the corresponding destruction of productive assets. Hnatkovska and Loayza (2004) provided empirical evidence that crisis-type volatilityis significantlymore adversefor growth than normal volatility. 13. Such arrangements are currently maintained by Brazil,Chile,Colombia,the Republic of Korea, Mexico, Peru. South Africa,and Thailand. The longest running of these arrangements, in Chile, was remarkably successful in maintaining price stability throughout the 1990s, while avoiding severe episodes of real exchange rate volatility. More recent converts to this type of nominal institutional arrangement have also been quitesuccessfulsince its (admittedlyrecent)adoption. References Blanchard, O., and F. Giavazzi. 2003. "Improving the Stability and Growth Pact Through Proper Accounting of Public Investment." Discussion Paper 4220. Centre for Economic and Policy Research,London. The World Bank Research Observer, vol. 21, no. 2 (FaU2006) Bordo, M., B. Eichengreen, D. Kliigebiel, and M. S. Martinez-Peria. 2001. "Is the Crisis Problem GrowingMoreSevere?"Economic Policy: A European Forum 16(32):51-82. Calderbn,C.. R.Duncan, and K. Schmidt-Hebbel.2003. "TheRole of Credibility in the CyclicalProper- ties of Macroeconomic Policies in Emerging Economies." Working Papers Central Bank of Chile, CentralBankof Chile,Santiago. De Ferranti, D., G. Perry, I. GiIl, and L. ServCn. 2000. Securing Our Future. Washington, D.C.: World Bank. De la Torre, A., and S. Schrnukler.2003. Coping with Risk Through Mismatches: Domestic and Intema- tional Financial Contracts for Emerging Economies. Washington,D.C: WorldBank. Easterly,W. 1999. "When Is FiscalAdjustment an Illusion?"Economic Policy 14(28):55-76. . 2001. "The Lost Decade:Developing Countries' Stagnation in Spite of Policy Reform."Journal ofEconomic Growth 6(2):135-57. Easterly, W., and L. Servkn, eds. 2003. The Limits of Stabilization: InJrast~ucture,Public Deficits, and Growth in Latin America. Palo Alto,Calif.:StanfordUniversityPress. Easterly, W., R. Islam, and J. Stiglitz. 2001. "Volatility and Macroeconomic Paradigms for Rich and Poor."InJ.DrCze,ed., Advancesin MacroeconomicTheory. New York:Palgrave. Edison,H., R. Levine,L. Ricci,and T.Sloken.2002. "International FinancialIntegrationand Economic Growth."WorlungPaper 9164. NationalBureau of EconomicResearch,Cambridge.Mass. Eichengreen, Barry, and Ricardo Hausmann. 1999. "Exchange Rates and Financial Fragility." WorkingPaper 7418. NationalBureauof EconomicResearch,Cambridge,Mass. Fitch Ratings.Various years. Sovereign Comparator. New York. Frankel, Jeffrey A., and Andrew K. Rose. 1996. "Currency Crises in Emerging Markets: Empirical Indicators."WorkingPaper 5437. NationalBureauof EconomicResearch,Cambridge. Mass. Hausrnann, R., D. Rodrii, and A. Velasco. 2005. "Growth Diagnostics." Harvard University, JohnF. Kennedy Schoolof Government,Cambridge,Mass. Hnatkovska,V., and N. Loayza. 2004. "Volatilityand Growth." Policy Research WorlungPaper 3184. WorldBank,Washington,D.C. IADB jlnter-American DevelopmentBank).1995. Overcoming Volatility in Latin America. Washington,D.C. IMF(InternationalMonetaryFund).2002. Financial Stability in Dollarized Economies. Washington,D.C. 2003. Fiscal Adjustment in IMF-Supported Programs. Washington.D.C. .Variousyears.International Financial Statistics Database.Washington,D.C. JPMorgan.Variousyears. Emerging Market Bond Index (EMBI). New York. Kaminsky, G., and C. Reinhart. 1999. "The Twin Crises: the Causes of Banking and Balance of PaymentsCrises." American Economic Review 89(3):473-500. Kose, A., E. Prasad, and M. Terrones. 2003. "Financial Integration and Macroeconomic Volatility." IMFStaff Paper03/50. International MonetaryFund,Washington,D.C. IOraay,A., and V. Nehru. 2003. "When Is ExternalDebt Sustainable?"Policy ResearchWorkingPaper 3200. WorldBank,Washington,D.C. Krugman,P. 1999. "BalanceSheets,the Transfer Problemand Financial Crises." In P. Isard,A. Razin, and A. Rose,eds., International Finance and Financial Crises. London:Kluwer. Lane, P. 2003. "The Cyclical Behavior of Fiscal Policy: Evidence from the OECD." Journal of Public Economics 87(12):2661-75. Montiel, P., and C. Reinhart. 1999. "Do Capital Controls and Macroeconomics Policies Influence the Volumeand Compositionof Capital Flows?Evidencefromthe 1990s."Journal oflnternational Money and Finance 18(4):619-3 5. Peter Montiel and Luis Serven 177 Montiel, P., and L. ServCn. 2004. "Macroeconomic Stability in Developing Countries: How Much Is Enough?"PolicyResearch Worling Paper 3456. World Bank,Washington, D.C. Mussa,M. 2002. Argentina and the Fund: FromTriumph to Tragedy. Washington. D.C.: Institute for Inter- nationalEconomics. Perry, G. 2003. "Can FiscalRules Help Reduce Macroeconomic Volatility?"Policy Research Working Paper 3080. WorldBank, Washington, D.C. Perry, G., and L. Semen. 2003. "The Anatomy of a Multiple Crisis: Why Was Argentina Special and What Can WeLearnfromIt?"PolicyResearchWorling Paper 3081. WorldBank,Washington,D.C. Pritchett, L. 2004a. "The Grist and the Millfor the Lessonsof the 1990s." In G. Nanltani and R. Zagha, eds.,The Growth Experience:What Have W e Learned From the 1 9 9 0 s Washington, D.C.: World Bank. . 2004b. "PolicyReforms and Growth Performance:What Have We Learned?"In G. Nankani and R. Zagha, eds., The Growth Experience: What Have W e Learned From the 1 9 9 0 ~ 2Washington, D.C.: World Bank. Ramey, G., and V. Ramey. 1995. "Cross-Country Evidence on the Link between Volatility and Growth." American Economic Review 85(5):1138-51. Reinhart, C., K. Rogoff, and M. Savastano. 2003. "Addicted to Dollars." Working Paper 10015. NationalBureau of Economic Research,Cambridge,Mass. Rodrik,D. 2001."WhyIsThereSoMuchEconomicInsecurityinLatin America?"CEPALReview 73:7-30. , 2004. "Growth Strategies." Harvard University, John F. Kennedy School of Government, Cambridge,Mass. Rodrik,D., and A. Velasco.1999. "Short-TermCapitalFlows." WorkingPaper 7364. NationalBureau of Economic Research,Cambridge,Mass. Schmukler, S., and L. ServCn. 2002. "PricingCurrency Risk under Currency Boards." Journal of Devel- opment Economics 69(2):367-9 1. ServCn,L. 2003. "RealExchangeRateUncertainty and PrivateInvestmentin LDCs."Review of Economics and Statistics 85(1):212-18. Shvets,0.2004. "Something Special about the 1990s?"In G. Nanltani and R. Zagha,eds., The Growth Experience: What Have W e Learned From the 1990s?Washington. D.C.: World Bank. Talvi,E.1997. "Exchange-RateBasedStabilizationwith EndogenousFiscalResponse."Journal of Devel- opment Economics 54(1):59-75. Talvi, E., and C. Vegh. 2000. "Tax Base Variability and Pro-Cyclical Fiscal Policy." Working Paper 7499. NationalBureau of EconomicResearch,Cambridge,Mass. World Bank.Variousyears.World Development Indicators Database. Washington, D.C. The World Bank ResearchObserver, vol. 21, no. 2 (Pan2006) Banking and Regulation i n Emerging Markets: The Role of External Discipline Xavier Vives This article reviews the main issues of regulating and supervising banks in emerging mar- kets with a view toward evaluating the long-run options. Particular attention is paid to Latin America and East Asia. These economies face a severe policy commitment problem that leads to excessive bailouts andpotential devaluation of claims of foreigninvestors. This exacerbates moral hazard and makes a case for importing external discipline Cforexample, acquiring foreign short-term debt). However, external discipline may come at the cost of excessive liquidation of entrepreneurial projects. The article reviews the tradeoffs imposed by external discipline and examines various proposed arrangements, such as narrow bank- ing, foreign banks and foreign regulation, and the potential role for an international agency or international lender of last resort. Liberalization and integration of financial markets have been associated with an increase in capital movementsand with the financialcrises. In particular, surges in foreign short-term debt have been blamed for crisis episodesin emerging economies in Asia (Thailand,Indonesia,and the Republicof Korea)and LatinAmerica (Mexico, Brazil, Ecuador, and Argentina), as well as in the periphery of Europe (Turkey). These criseshave proved costly in terms of output. Several policy responseshave been suggested. Among them have been the reduc- tion of short-term debt, the development of stock markets, the improved regulation and supervisionof domesticfmancialsystem,enhanced transparency requirements and market discipline,and the establishmentof an internationallender of last resort. Acatalogof "solutions"has been proposedto takecare of the problemsof banlungin emerging market economiesincludingmoving to a narrow bank system, building a currency union, and leaving banking in the hands of foreign banks and offshore institutions. This article identifies policy responses tailored to the needs of emerging market and developing economies. The question is whether the regulatory policies and O The Author 2006. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development/ nreWOR D BA K. Allrights reserved.For permissions,pleasee-mail: journals.permissions@oxfordjournals.org. L N doi:10.1093/wbro/kl002 Advance Access publication August 5,2006 21:179-206 practices of developed economies can be recommended essentially without change or whether a different policy mix is needed. A basic theme is that more acute asym- metric information problems and a weak institutional structure in emergingmarket economies call for policy prescriptions that differ not only from those of developed economiesbut also acrossemerging marlieteconomies. Attention is focused on a particular consequenceof the weak institutional struc- ture in emergingcountries:the lacli of capacityfor policy commitment. This lacliof capacity for commitment may be due to the short horizons of public officials in the faceof,for example,political instability. The outcome is that the government of an emerging marliet economy may bail out the private sector, encouraging excessive risk taking, or devaluethe claimsof foreigninvestors,discouragingtheir investment in the first place. Indeed, a major problem in emerging markets is the implicit or explicit guarantee of a bailout in the event of a banking crisis, as experiences in Argentina, Mexico, and Thailand show, or the use of inflation to devalue domestic currency4enominated claims. ' The result is that domestic regulation may not be enough in countries that face a commitment problem,and those countries may have to import discipline.However, some ways of importingdiscipline,such as increasing the role of short-term foreign debt, have costs. The article examines the tradeoffs imposed by different ways of importingdisciplineand classifiescountries accordingto the desirabilityof doing so. It analyzes the catalog of solutions to the problems of financialsystems in emerging market economies and the potential role of an international agency such as the InternationalMonetaryFund (IMF). Banking i n Emerging Market Economies What malies banking regulation different in emerging market economies?Why do these countriesrequiredifferent regulatory and supervisoryarrangements? The Role of Banks and Fragility Banks provide transaction and payment system services,insurance, and risk shar- ing (transformingilliquid assets into liquid liabilities).Acentral function of banks is the financing and monitoring of entrepreneurial projects, which are illiquid and opaque because of asymmetric information problems such as moral hazard and adverse se~ection.~Some entrepreneurial projects cannot obtain marliet financing because no credible information on them can be conveyed to the public domain. A bank can accumulate relationship-specificskills to monitor those projects and be able tofinancethem.In this way the banlcingsystem helps overcomeproblems asso- ciated with asymmetricinformationin an economy. 180 The World Bank Research Observer, vol. 2 1 , no. 2 (Fall 2 0 0 6 ) Asymmetricinformation problems are bound to be more acute in emerging mar- ket and developingeconomies.The productionof informationis more problematicin emerging market economiesbecause of institutional factors.Indeed,emergingmar- ket economiesfare poorly on the indicators of rule of law, the protectionof property rights, and accounting standards, pointingto aggravated moral hazard and adverse selection problems. Furthermore, the production of information, which typically involves a fixed cost, is discouraged by the normally small size of the emerging market. Onedirectconsequenceof the enhanced asymmetricinformationproblemsis that the financialsystemis less developedbecause the cost of setting up well-functioning markets is higher. Arms-lengthfinance just does not work. Aderivedconsequenceis that the role of the banking system, in particular the monitoring of entrepreneurial projects, is much more crucial. Indeed, for most companies in an emerging market economy the only possible source of finance, except for earnings, is bank loans. Banks and their monitoring capacity are therefore at the center of economic devel- opment, and their potential fragility may dramatically worsen downturns. The cri- ses of Mexico, East Asia, and Russia provide examples, as well as the more recent crisis in Argentina. Why are banks fragile?The essence of banks is that they create liquidity, which leaves them vulnerable to runs. Banks protectentrepreneurs that need financefrom the liquidity needs of depositors and investors. There are different versions of the story, but this is the cornerstone of modern banking theory (Diamond and Dybvig 1983; Holmstrom and Tirole1997, 1998; Diamond and Rajan 200I ) .Firms may ~ be unable to obtain fundingbecause of asymmetricinformation,as they do not have enough pledgeable income (the fraction of their return that can be committed to be paid to outsiders).Banks come to the rescue,for example,by creating liquidityhold- ing collateral and committing to make payments (Holmstrom and Tirole 1997, 1998). In short, the standard deposit contract and loan provision to opaque entre- preneurial projectsarecomplementaryand central to the function of a bank. Short-term debt-a deposit redeemable at par-leaves banks exposed to failure when returns are low. However, this possibility has desirable incentive properties because it can create an incentive to exert effort for self-interested bank managers who are, put simply, interested mostly in the continuation of their jobs. This is rea- sonable when the private benefits of control loom large, as may well be the case in emerging market economies with a weak institutional structure. In general, short- term debt has a disciplining effectin the presence of moral hazard. Indeed, in the extreme, the repayment oflong-termdebt may not be enforceable, and payment to the creditor may be ensured only by the threat of liquidation in an interim period (Boltonand Scharfstein1990; Hart 1995).~ How does the theory relate to the trend in the banking industry of developed economies moving from the traditional business of taking deposits and granting Xavier Vives 181 loans to the provisionofservices to investors (investmentfunds, advice, and insur- ance) and firms (consulting,insurance, mergers and acquisitions, underwriting of equity and debt issues, and risk management)?Banking in industrial countries is in a process of transformation (more advanced in the United States than in Europe), where the financial margin makes way for feeand commissionrevenue. Indeed, the share of assets held by banks relative to nonbank intermediaries is declining in developedeconomies(Allenand Santomero 2001).In contrast, in emergingmarket economiesthe traditional roleof banks remains central. In summary, in emerging market economies the traditional function of banks is all the more important, because financial markets are less developed and asymmet- ric informationproblems are more acute. Fragility, Regulation, and the Safety Net The inherent fragility of the banking system, with asymmetric information at its root, leads to the failure of institutions, panic, and systemic crises that can have a major impacton the economy.The great depressionof the 1930s is a good example, and more recent episodes of financial crises in the United States, Scandinavia, Mexico, East Asia, and Russia remind us of the potential for economic disruption. The failure of a bank has adverse consequences on nonfinancial firms precisely becauseindividualbank-firmrelationshipsare valuable(Petersenand Rajan1994). In fact,even a contraction of bank capital may result in a credit crunch, with severe disruption to the private sector. This is especially evident in an emerging market economy,which ismoredependenton the intermediaryservicesof banks. At the base of the fragility of banking is the coordination problem of depositors, who may decide to call back their short-term deposits and make a sound bank fail. Theliterature has two viewsof crises: the multiple-equilibriumpanic view (Diamond and Dybvig1983)and the information-basedview (Gorton1985,1988;Jacklinand Bhattachrya 1988).According to the first,runs are triggeredby events unrelated to the fundamentals, whereas accordingto the second,runs are triggered by bad news about the assets of the bank. Recently, those views have been reconciled by intro- ducing asymmetricinformation and linking the probabilityof a run to the strength of fundamentals (Morris and Shin 2000; Rochet and Vives 2004; Goldstein and Pawner 2005).' Thus, a solvent bank may be subjected to a purelyspeculative panic,with deposi- tors withdrawing funds and the bank being forced to quickly liquidate assets at a high cost. The cause of the problem is the dependence of banks on short-term debt (or the standard deposit contract). In addition, there is the danger of systemic risk owing to contagion from the failure of one entity, which may give rise to a strong negativeexternality both for the financialsector and for the real sector of the economy. For example,the failureof oneentity may, through interbank market commitments, 182 The World BankResearch Observer. vol. 21, no.2 (Fall 2006) lead to the failureof others (Allenand Gale 2001).Similarly,large variations in the price of assets, such as an abrupt fall in the stock market or the failure of a main intermediary, may generate a domino effect and systemic crises affecting the pay- ment system. In general, competitive banking will be excessivelyfragile, and the lender of last resort facilities and prudential regulation (discussed in the following section) will come to the rescue. The aim of regulation has been to provide the banking and financial systems with stability to avoid the negative effects associated with failing institutions and systemic crises. Other aims have been to protect the small investor and to promote the competitivenessof the system. The Lender of Last Resort and the Policy Commitment Problem In industrial countries the lender of last resort and deposit insurance are basic to the stability of the banking system. There is a tendency, however, to protect banks and depositors above the levels required by the deposit insurance, in particular, under the too-big-to-failpolicy. One reason is the potentially systemicconsequencesof the failure of a large institution,but more often help is a reflectionof a time-inconsistency problem in the presence of a moral hazard problem.Awell-intentionedlenderof last resort-the central bad<-willfind it optimalex post to help whenever thissalvages the value of projects, whereas bankers, anticipating the help,will tend to exert sub- optimal effort,creating a moral hazard situation because the central bank is unable to observe the banker's levelof effort in monitoringprojects. The time-inconsistency problem faced by a central bank arises because ex ante the central bank may want to commit to closing the bank if the returns are low (sig- naling a solvency problem),whereas helpingthe bank if the returns are only moder- atelylow (signalinga liquidityproblem).Such a commitmentprovidesincentivesfor bank managers to monitor the projects they finance. In this way the central bank may implement the second-best solution in a competitive banking system.6How- ever, ex post, costly liquidation of the projects may not be optimal, so the central bank may hesitate to carry out its threat. The commitment problem is compounded by the interest of a bank manager in the continuation of the bank. Building a central bank with a "tough" reputation can alleviate the time-inconsistencyproblem. This commitment problem because of intertemporal inconsistencyis aggravated in emergingmarket economieswhere institutions are weak and sufferfrom a lack of credibility and independence.It is difficultfor central banks to build a reputation for discipliningbanks because the central banker's effective horizon is short because of political instability. For example, in Argentina in the 1980s the average term in office for a central bank governor was less than a year although the legal term was four years (see Cukierman1992, chapter 19). A related problem is the lack of legal protectionforbanksupervisorswho attempttoimposediscipline(againasin Argentina). Xavier Vives 183 Then, even if the perceived problem is serious,the bank may be allowed to continue or even be granted help (WorldBank 1998).Aweak institutionalstructure allowing the regulated to unduly influence the regulators (regulatorycapture) also explains why failuredoes notlead to a change of management. "Crony capitalism,"wherethe governmenthelpsfirmsthat are consideredfriends,is an extremeform of capture. The consequenceof the intertemporalinconsistencyand regulatorycapture is that the central bank of an emerging market economy that has a commitment problem will have incentivesto induceinflationto reducethe real valueof nominaldebt com- mitmentswhen banks or entrepreneurs are in trouble.This willavoid projectliquida- tion, but it destroys incentives to exert effort and, in turn, devalues foreign investmentsin domestic currency. The outcome is a lack of foreign investment. As argued below,foreign-denominatedshort-term debt may be crucial for the access of an emergingmarketeconomy to international capitalmarketsbecauseit protectsfor- eign investorsfrom the devaluationof their claims by the actionsof the government. Prudential Supervision The too-big-to-failpolicy,deposit insurance,and,in the extreme, blanket protection and bailouts introduce distortions into the decisions of financial entities. They reduce the incentive of depositors to monitor bank performance and, coupled with the bank's limited liability,they give rise to excessive risk taking. Bailoutseliminate the disciplining effect of closures and exacerbate risk taking and inadequate moni- toring by bank managers. The need for regulation is particularly acute when charter values7 are low (and therefore incentivesto take risk are high),and the social cost of failure is high (and therefore banking failure has a large impact). With either high disclosure require- ments or risk-based insurance, banks pay for taking more risk, and capital require- ments have a chance to be a sufficientinstrument for controlling risk taking (table1; see Diamond and Rajan 2000for the role of bank capitalin controllingexcessfragil- ity).Otherwise,capitalrequirementsmay need to be complementedwith restrictions on the investmentsof banks to check risk taking. Capital requirements together with supervision and market discipline are the three pillars of bank regulatory reform. The general trend in bank regulation is to check risk taking with capital requirements and appropriatesupervision. Both risk- based (deposit)insurance and disclosure requirements have been proposed to limit risk-taking behavior (top and the bottom rows of table 1).Developed economies have tended to move in that direction.This movement has been accompanied by a reform of the 1988 Base1Accord on capital requirements to better adjust them for risk (Base1II). Base1 I1contemplates that banks can adopt either a "standardized" approach to capital requirements in which external rating agencies set the risk weight for differenttypes of loans (saycorporate, banks, and sovereignclaims)or an 184 The World Bank ResearchObserver, vol. 21, no. 2 (Fall 2006) Table 1. PossibleBankingRegimes,the Incentives toTakeRisk,andthe NecessaryRegulatory Instruments When Charter Valuesare Low and the SocialCost of Failure Is High Risk-taking incentives Bankinq reqimes Liability (rates) Asset (investment) Regulation Free banking, obsemable Mediumlow Absent Capital requirements risklhigh disclosure Free banking, unobservable Mediumhigh Maximal Capital requirements and asset restrictions riskllow disclosure Risk-insensitiveinsurance High Maximal Capitalrequirements and asset restrictions Risk-basedinsurance Low Absent Capital requirements Source: Based on Matutes and Vives (2000),Cordella and Yeyati (2002),and Hehann, Murdock,and Stiglitz (2000). internal rating-based approach in which banks estimate the probability of default (and also the loss, given default,in an advanced version of the method).The idea is to calibrate the capital requirement so that it covers the value at risk (expectedand unexpected)fromthe loan under some assumptions. It must be noted,however,that transparency has its limitations.Whereasit isfea- sible to introduce the disclosure requirements of the market positionsof banks, it is more difficult to assess the risk level of the illiquid loan portfolio of a bank. Further- more, more disclosure may induce information-based runs of investors, generating instability. Regulation in an Emerging Market Economy If it is feasible to introduce risk-based insurance and disclosure requirements that eliminatemoral hazard, capital requirements (risk-basedthemselves)may be a suffi- cient instrument to check risk taking and improve welfare (table1).However, the characteristics of an emerging market economy cast doubt on the feasibilityof such a strategy. First, an emerging market economy is likely to face considerable uncertainty in terms of high economic volatility, high direct and indirect exposure to exchange rate risk,high maturity and currency mismatch,and high nondiversifiablerisk in a typical loan portfolio.Two sourcesof increased risk are the high proportionof debt in foreign currency and of debt of short maturity (the reasons are explained later). The risk of high (andvariable)inflationis,in a first instance,at the baseof the useof thoseinstru- ments. Furthermore, higher levels of risk are hidden behiid the (false)security of a pegged exchange rate. For example,a bank feels protected because it has matched a dollar liability with a dollar-denominatedloan without realizing that if the borrower earns income in pesos, a collapse of the peso will provoke a default. The currency Xavier Vives 185 match has hidden credit risk.If the bank doesnot match a dollar deposit with a dollar- denominatedcredit,it becomesexposed directlyto exchange rate risk. Second, as noted, financial marltets are less developed, and the monitoring role of intermediaries is enhanced in emerging market economies. The production of infor- mation on private sector activities and the general contracting environment are problematic. The background of these problems is the lack of economies of scale in the production of information and severe moral hazard and adverse selection prob- lems. This implies,at the same time, that marltets are thin and that the generation of information and contract enforcement roles rely relatively more on financial inter- mediaries. A further consequence of the thinness of financial marltets is that banks have a high exposure to publicdebt (forexample, government bonds) and are there- fore vulnerable to inflationary strategies of the government and may be less able to match long-term investments by issuing appropriate liabilities(maturity mismatch). Short-term debt leaves banlts and firmsexposed to sharp increasesin interest rates in response to a currency devaluation. At the same time a bank may have a harder time diversifyingits portfolio because default probabilities may have a high correla- tion across projects. An obvious case is the collapseof the exchange rate. All of these factors point to a riskier environment for banks. A currency crisis leads to a financial crisis and to strong effects in the real sector. Adepreciation of the currency leads to a deterioration of balance sheets for firms and banks and to a decline in the net worth of the private sector. Because of asymmetric information problems, this decline in net worth willlead to a credit crunch, and badts with weak balance sheets will cutback on lending, exacerbating moral hazard and adverse selection problems (see Bernanke and Gertler 1989 for the general mechanism and Mishlcin 1999a, 1999bfor an application to the Tequila crisis in Mexico). A weak banking sector can also lead to a currency crisis. Kaminslty and Reinhart (1999) found that banking sector trouble typically precedes a currency crisis and that the currency crisis aggravates the banking crisis in a self-reinforcing manner. Finally, underdeveloped financial markets and no sound contracting environment increase the social cost of failure and the liquidation of projects. This means that the real effects of financial crises are multiplied. Third, an emerging marltet economy will tend to have a weak supervisory struc- ture. The reasons are rooted in the same factors that keep financial markets under- developed: the difficulty producing information and enforcement problems aggravated by the lack ofprotection for supervisors. Supervisors are either more eas- ily corrupted,becauseof the lack of resources and lowsalaries,or morevulnerable to retribution if they do not acquiesce to the demands of lobbies,because of the lack of effective legal protection. Symmetrically, some banks may see how expropriatory regulatory decisions are made, perhaps because of their weaker political position, and this willinduce a high rate of discounting. The consequence will be low charter 186 The World Bank ResearchObserver, vol. 21, no. 2 (Fa112006) values as part ofthe bank profitisdiluted,and this willhappen even with a relatively low levelof competition.The consequencewill be enhanced incentivesto take risk. The characteristics of an emerging market economy of high uncertainty, increased likelihood and incidence of financial and currency crises, predominant financialroleofbanks, and weaksupervisory structure lead to the policy conclusion that the regulatory strategy needs to be adapted to these conditionsand to protect the fundamental role that banks play. Indeed, these characteristics make it much more difficultto follow the industrial country regulatory strategy in an emerging market economy. First, because information problems are more acute and the production of information is more problematic, it is more difficult to move toward a disclosurestrategy.Second,risk-baseddepositinsurance can work only when insur- ance can be priced according to objective indicators of bank risk. Those indicators may be moredifficult to obtain in an emerging market economy (they are dificult to obtain even in a developed economy). This makes the move toward a risk-based insurance strategy more difficult. Furthermore, the application of Base1I1criteria to emerging market economies may be problematic. As Powell (2001) noted, these economieswill have dficulty implementingthe internal rating approach, especially because the new standards have not been calibrated for the environment in these countries, and so they will tend to adopt the standardized approach. But with the limited number of rated institutions in emerging market economies, this will mean little change from the current situation. Yet the problem of building a better link betweenrisk and capitalis,if anything, more acute there. The corollary is that the regulation of banking and financial markets must be adapted for emerging market economies. Reliance on transparency and disclosure requirements as well as risk-based insurance and capital requirements is limited. Capital requirements in particular will need to be adapted to the conditions of emerging market economies (for example, public debt is risky because of inflation) and, most likely, complemented by other restrictions on the activity of financial institutions. At the same time,becauseof the high cost of the liquidationof projects and the social cost of failure in emerging market economies, competitive pressures and market disciplineshould not be set at the same level as in developed economies. A broader consequence of the weak institutional structure of emerging market economies is that the policy commitment problem becomes central for attracting foreign capital. This problem and how external discipline can help overcome it are discussed in the followingsection. Argentinacoped with a policy commitmentproblemby importingexternaldiscipline through its currency board's "hard peg" (a fixed exchange rate backed by foreign reserves) and adopted a market discipline model (high disclosure levels, subordinated debt,limiteddepositinsurance,andrisk-basedcapitalrequirements).Argentinawasfol- lowing a modem industrial country strategy anchored in the currency board (it was moving toward the top line of table 1,free banking with disclosure). The obvious Xavier Vives 187 questionis whether thismodel isfeasibleonce the currencyboard has collapsedand the anchorhasdisappeared.Thebanlcingandcurrencycriseswerederivedkomthe nonsus- tainabilityof thecurrencyboardin a recessionarycontextandwiththe underlyingprob- lemsin Argentina,such as the lack of the credibility of institutionsand the protectionof propertyrights. The crisis was more likean informedrun (suchas that after the Tequila crisis) than a crisis derived kom the moral hazard of bank managers or a coordination failureof theexpectationsof depositors.The financialsystem was reasonablywellregu- lated and thecurrencyboard,whencredible,provideddiscipline,therebylimitingexces- sivebailouts(seeCalvo,Izquierdo,and Talvi2003forfurtherdetailsof thecrisis). The Policy Commitment Problem and External Discipline The governmentof anemergingmarketeconomymaydevaluethe claimsof foreigners in domestic currency to protect the domesticprivatesector.This lack of policycom- mitment capacity derived from a weak institutional structure is a central problem for an emerging market economy that needs access to the international capital market. In a similar vein Tirole (2002) argued that the central market failure in external borrowing for an emerging market economy is the lack of contracting capacity betweenits government and foreign investors. It has already been shown how short-term debt has a disciplining effect when there is a moral hazard problem,whether it isfrom the side of theentrepreneurseek- ing credit tofinancea project or the banker monitoring a loan.This givesscopefor a central bank in a competitive banking system to provide help in a range of returns while maintaining incentives by denying help when returns fall below the optimal critical threshold. The problem arises, however, that while ex ante it is optimal to commit not to help when returns are low, ex post, once effort decisions have been made, it is optimal to help avoid costly liquidation. A central bank with no policy commitment capacity will have incentives to induce inflation to reduce the real value of nominal debt commitments when banks or entrepreneurs are in trouble. This will avoid liquidation but destroy incentives to exert effort and discourage foreignlenders,who may see their claimsdevalued. But to implement the second-best solution,a central bank must be able to commit not to help when returns are low. External discipline may come to the rescue (Vives 2002).An extremeform of importingexternal discipline,examinedherefor illustra- tive purposes,is "dollarization." Dollarization Dollarizationrepresents a commitment to a limited use of the lender of last resort facili- ties. Dollarization means that banking contracts are in "real" (dollar) terms. In a 188 The World Bank Research Observer. vol. 21,no. 2 (Fall 2006) dollarized regime help for the banking system (bailouts)must be arranged in advance, through stabilizationfunds or tax schemes, or precontractedin the international mar- ket. For example, when Argentina adopted a currency board in 1991-2001, it also established a contingent liquidity facility with international banks. Banlcs were also required to meet new liquidity requirements and to hold excess reserves, because the Convertibility Law of 1991 and the Charter of 1992 severely res cted the central banks' lender oflast resort activity (Calomirisand Powell 2000).Most currency boards have established limited lender of last resort facilities. A stabilization fund can provide liquiditywhen needed,but itcan be diverted(ashappenedin Mexicoand Thailand). Dollarization represents a commitment because it is costly to reverse. A currency board or a hard peg, as Argentina's recent experience suggests, represents a lesser commitment. Typically,the currency board is established by law (as in Argentina) and therefore can be dissolvedby another law, which does, however, raise the cost of getting rid of the arrangement. What are the costsand benefitsof dollarizationin a smallopen economy?In a com- petitive banking environment, projects are liquidated when the returns of the bank tri cannot cover the promised payment to depositors. However,in this competitive bank- ing solution, there is typicallyexcessiveliquidation.The liquidationthreshold imposed by a competitive banking environment is stricter than the second-best threshold because of risk sharing (acompetitivebad