. 3 Preventing Financial Crises in Developing Countries

INANCIAL CRISES OCCUR WHEN FINANCIAL SYSTEMS BECOME ILLIQUID or insolvent. Such crises have recurred throughout the . A collapse in confidence, usually after a period of market euphoria, marks such crises—examples include the Dutch mania crisis of 1637–38, the Indian cotton futures market Fcrash of 1866, and the Great Depression of 1929. When foreign lenders are involved, cross-border payments problems arise as well. The East Asian crisis belongs to the class of twin financial crises, in- volving both banking and currency problems. According to modern eco- nomic theory, information asymmetries and financial market failures are central in explaining macroeconomic fluctuations and financial crises.1 Because lenders know less than borrowers about the use of their funds and cannot compel borrowers to act in the lenders’ best interests, lenders can panic and withdraw their funds when they perceive increased risks, in the absence of adequate public regulation and safeguards. That can trigger

121 GLOBAL ECONOMIC PROSPECTS

much wider financial crises, with spiraling countries (including the sovereign debt cri- real-sector effects. The costs can be severe. sis of the 1980s in Latin America) stemmed Such crises can bring down the financial from macroeconomic mismanagement, system, cause asset prices to collapse, and including excessive public deficits and over- bankrupt sound as well as unsound banks borrowing abroad. As evident in the light and . The East Asian crisis is of recent events in Russia, reforms and expected to cause output in Indonesia, the policies to avoid such sovereign debt crises Republic of Korea, and Thailand to drop are important and still relevant in develop- 12–24 percent in 1998 (from previous ing countries. The focus of this chapter, trend levels), throwing millions into unem- however, is on the type of crisis which ployment and poverty. involves private-to-private capital flows, and the role of domestic and international ver the past 100 years industrial financial systems in intermediating such O flows. The international setting is impor- countries have reduced the incidence tant because these crises (East Asia in 1997, and severity of systemic crises through Mexico in 1994, and Chile in 1982) are public policy and institutional closely connected to rapidly rising cross- reforms—although they have not border private capital flows. These flows eliminated crises entirely. have grown massively in the past decade, but without the improvements in institu- Over the past 100 years industrial tions and public regulation needed for their countries have reduced the incidence and safe management. severity of systemic crises through public The analysis of financial crises and the policy and institutional reforms. They have appropriate policies needed to prevent not eliminated them entirely, however, as them highlights the way various factors the in the United interact and amplify risks. These include States in the 1980s, the banking crises in inadequate macroeconomic policies, surges Nordic countries in the early 1990s, and the in capital flows, fragility of domestic finan- unfolding financial sector problems in Japan cial systems, weak , illustrate. In developing countries there is and ill-prepared financial and capital often a mismatch between public policies account liberalization. Policymakers need and the institutional structures (which are to be concerned about these interactions in slow to change) intended to prevent finan- the sequencing and timing of policy and cial crises, and their integration with world institutional reforms. financial markets. Thus the number of such This chapter’s key messages: crises remains large and their costs have • The number and costs of financial been growing. Reducing their incidence calls crises have risen in developing coun- for policy and institutional reforms in both tries since the 1980s, partly because rel- national and international settings. atively small economies are more Until the surge in private capital flows exposed to the risks of international in the 1990s, most crises in developing capital flow reversals. Many recent

122 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

crises are in fact both currency and and shift the composition to -term banking crises, including the East Asian and volatile inflows. Flexible exchange crisis (1997) and the Mexican peso cri- rates help regain autonomy for mone- sis (1994). Developing countries have tary policy, improve risk perceptions, recently been exposed to a wave of cap- and reduce incentives for excessive bor- ital inflows but have little experience rowing, but they are not always enough with the institutional and prudential to avoid crises and may result in volatile safeguards needed to minimize associ- and misaligned real exchange rates. ated risks. The easier availability of Countercyclical fiscal policy is impor- cheap international capital in good tant, but it too has tradeoffs (fewer times encourages excessive private risk schools and roads, for instance, to taking, which can turn into a major accommodate more shopping malls and problem when favorable financial sen- office towers). What is needed is a mul- timent erodes. The institutions needed tidimensional approach, often with to minimize the risks of such crises take more flexible exchange rates, greater a long time to develop, while the politi- reliance on fiscal policy, and better and cal constraints on prompt policy tighter domestic actions to avert them are often severe. (and, where necessary, restrictions on However, the building of such required capital flows) to reduce excessive capi- institutions and safeguards needs to tal inflows, domestic lending booms, proceed vigorously in all countries, so and risks of financial crises. that the potential benefits of globaliza- • Financial sector liberalization, which tion can be realized with fewer risks. can significantly boost the risk of crisis • Poor macroeconomic policies leave a (particularly in conjunction with open country vulnerable to , capital accounts), should proceed care- and prudent policies are the first line of fully and in step with the capacity to defense. In the presence of large capital design and enforce tighter regulation inflows and weak financial systems, and supervision. At the same time, however, the room for maneuver in set- however, efforts to improve prudential ting appropriate macroeconomic poli- safeguards and banking operations will cies to control excessive private bor- need to be accelerated. There is strong rowing and risk taking is constrained evidence of a higher probability of cri- because of the presence of numerous sis following liberalization without tradeoffs and their distributional conse- stepped-up prudential safeguards (even quences. Fixed or pegged exchange in industrial countries). Regulations rates help anchor expectations and that increase safety and stability are reduce uncertainty. But they may also needed. Banking and provide unintended incentives to the reforms, oriented toward better risk private sector to overborrow (as in management, are critical in any strat- Thailand), while sterilizing capital egy to prevent financial crises. Public inflows may be costly and ineffective policy and institutional reforms that

123 GLOBAL ECONOMIC PROSPECTS

clamp down on connected bank lend- • Changes are needed in the architecture ing and improve corporate governance of the international financial system in are equally essential to support the view of the excessive volatility (euphoria safety of the financial system. and panics), strong contagion effects, • Capital account liberalization should and increased scope for in also proceed cautiously, in an orderly international financial markets. The and progressive manner, given the large most pressing issue is to develop better risks of financial crises—heightened by mechanisms to facilitate private-to-pri- international capital market failures— vate debt workouts—including, under in developing countries. Benefits of some conditions, “standstills” on exter- capital account liberalization and nal debt—and help resume capital flows increased capital flows have to be and increase international liquidity to weighed against the likelihood of crises countries in crisis. Although there are and their costs. Clearly the benefits some compelling arguments in favor of from foreign direct investment (FDI) a , appropriate bur- and longer-term capital inflows out- den-sharing, rules for intervention, and weigh the costs associated with the moral hazard remain difficult and unre- increased likelihood of financial crisis, solved problems. Improved regulation and developing countries should pur- by creditor-country authorities and bet- sue a policy of openness. But for more ter risk management of bank lending to volatile debt portfolio and interbank emerging markets should also help short-term debt flows and the related reduce the probability of crisis. More policy of full capital account convert- timely and reliable information is desir- ibility, there are higher associated risks able, but complete transparency and of financial crisis and greater uncer- better information alone will not pre- tainty about the benefits. Tighter pru- vent crises. Still, better use of warning dential regulations on banks, and, indicators may help governments take where the domestic regulatory and pru- corrective actions early enough to dential safeguards are weak, restric- reduce the extent and cost of crises. The tions on more volatile short-term issues are undergoing debate and con- inflows that minimize distortions and sideration in different forums. are as market-oriented as possible (through taxes, for instance), may reduce the risk of financial crisis. For Costs and causes of countries that are reintroducing such financial crises restrictions on capital inflows, these inancial crises have become more fre- actions will need to be managed care- quent in developing countries since the fully so as not lead to a loss of confi- Fstart of the 1980s (figure 3-1). They have dence; their reintroduction for capital taken three main forms: currency crises, outflows during a crisis may pose diffi- banking crises, or both. Currency crises are cult problems (not considered here). usually attacks on the domestic currency that

124 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Increased incidence of financial crises since the 1980s Figure 3-1 Incidence of financial crises, 1970Ð97

Number of crises 15 22 Banking crisis

10

5

0 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1997

Source: Caprio and Klingebiel 1996a; Frankel and Rose 1996; and Kaminsky and Reinhart 1997.

end with a large fall in its value, although experienced similar problems (Kaminsky they can include speculative attacks that are and Reinhart 1997). While these crises have successfully warded off by the authorities.2 been associated with large volumes of pri- Banking crises refer to bank runs or other vate-to-private capital inflows, many other events that lead to closure, merger, takeover, currency or twin crises in developing coun- or large-scale assistance by the government to tries, including most recently in Russia, are one or more financial institutions. of the traditional type where excessive pub- Sometimes, both currency and banking lic borrowing plays a central role. crises occur around the same time—the so- Financial crises can entail large costs called twin crises. The 1997 financial crisis (in lost output and welfare) and distribu- in East Asia is the most recent example— tional effects, which are substantially mag- with Indonesia, the Republic of Korea, nified in a twin crisis. Banking crises exac- Malaysia, and Thailand all experiencing erbate the negative impacts on the economy currency turbulence along with serious through a reduction in the volume of loans, banking sector problems. Earlier examples the misallocation of financial resources, include the Southern Cone countries— and the ensuing contraction in credit and Argentina (1981), Uruguay (1982), and cutbacks in investment (box 3-1). Chile (1982). More recently, Mexico The greater frequency and cost of cur- (1994), Argentina (1995), and the Czech rency and twin crises have been associated Republic (1997), as well as Finland, Nor- with surges in international capital in- way, and Sweden in 1991 and 1992 have flows—especially private-to-private flows—

125 GLOBAL ECONOMIC PROSPECTS

Box 3-1 Costs of financial crises

World Bank study found for a sample of 14 math (Lindgren and others 1996). Such crises can also banking crises a 5.2 percent average decline in result in significant resolution costs, stretched over many output growth after crisis (World Bank 1997b). years. A study in Latin America found that at least 4 to 5 A Another study found in emerging markets an years are required to resolve banking crises (Rojas-Suárez average cost in lost output (over to trend output) of 14.6 and Weisbrod 1996). The direct fiscal or quasi-fiscal out- percent of gross domestic product (GDP) per crisis (IMF lays for bank restructuring vary between industrial coun- 1998b). Yet another study found that both output growth tries and emerging markets and between individual coun- and efficiency fall after a banking crisis, with exchange tries from 1.5 percent of GDP for U.S. commercial banks rate volatility and currency crisis common in their after- in 1989 to 45 percent for Kuwait in 1995 (box figure

Costs of crises can be huge… Cost estimates of bank restructuring

Percent of GDP

50

45

40

35

30

25

20

15

10

5

0 Industrial countries Emerging markets Emerging markets Emerging markets Argentina (1980Ð82) Chile (1983Ð85) Uruguay (1981Ð84) Kuwait (1992) a (1980 to 1990s) (1980s) (1980s) (1990s)b

Notes: The midpoint cost estimate for each country-episode was selected. The sample includes seven industrial and 34 emerging countries. a. Excludes Argentina, Chile, and Uruguay. b. Excludes Kuwait. Source: World Bank staff estimates based on Caprio and Klingebiel 1996a; Lindgren and others 1996; Rojas-Suárez and Weisbrod 1996; Alexan- der and others 1997.

to developing countries and the growing roughly $250 billion in 1996. Long-term integration of these economies with world private capital flows went from less than 1 financial markets (see below and Kaminsky percent of developing countries’ GDP in and Reinhart 1997). 1990 to a peak of 3.7 percent in 1993, and about 2.8 percent in 1996 (figure 3-2). The Private capital flows have surged surge in private capital flows is unprece- Private capital flows to developing coun- dented (at least since the end of the First tries rose from about $42 billion in 1990 to World War),3 being twice as high as the pre-

126 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

below). In general, restructuring costs are higher in devel- These costs are also much higher for twin crises oping countries—where they range between 3 percent and reaching 18 percent of GDP in 26 emerging markets, and 25 percent in Africa, between 1.8 percent and 13.2 per- for developing countries than for industrial countries. cent in Asia, between 0.3 percent and 41.2 percent in Moreover, the average recovery time back to trend growth Latin America, and between 1 percent and 15 percent in rates is longer for such crises (2.6 years, compared with Europe and Central Asia—than in industrial countries, 1.5 years for currency crises and 1.9 years for banking where they average less than 6 percent. crises). Calculations for selected individual countries (for Costs of currency crises are higher for emerging mar- this report) find that the cumulative loss of output (rela- kets than for industrial countries, and even higher in cases tive to trend) ranges from a low of 0.2 percent of GDP of currency crashes (see box figure below). (Mexico 1976) to a high of 30.6 percent (Chile 1971).

…and are greater in developing countries Cumulative loss of output per crisis for industrial and emerging economies

Percent of GDP

20

18

16 Industrial Emerging 14

12

10

8

6

4

2

0 Currency crises Currency crashes Banking crises Currency and banking crises

Note: Only crises with output losses are represented. Source: International Monetary Fund 1998b.

vious peak in private capital flows during in Thailand and 50 percent in Malaysia; the oil-boom years (1978–82). Further set- and in 1993–96, they reached 43 percent in ting them apart, much of recent capital Hungary and 35 percent in the Czech flows has been private-to-private, rather Republic. By contrast, the peak for Organi- than private-to-public, flows. Some surges sation for Economic Co-operation and in capital inflows have been particularly Development (OECD) countries in 1989 massive: in 1989–96 cumulative private was 2 percent of GDP per year (on a capital inflows reached 55 percent of GDP weighted average basis).

127 GLOBAL ECONOMIC PROSPECTS

High volatility of private after the 1994 Mexican crisis. Short-term capital flows bank loans are even more volatile, as wit- Private capital flows have also been volatile nessed in the East Asian crisis. and subject to large reversals. This is seen In the context of the increasing integra- in the decline during the of the tion of developing economies with world 1980s, and in the reversal after Mexico’s financial markets, the fundamental causes crisis in 1994, and after East Asia’s crisis in of twin crises of the type seen in East Asia 1997. FDI has been more stable and rose lie both in domestic policies and institu- steadily throughout the various crises in tions and in international capital market developing countries (figure 3-2). Thus the failures. The analysis in the rest of the recent rapid increases in FDI flows might chapter focuses on these causes of financial be construed as being of the “jet-airplane” crises and appropriate policies to prevent variety, bringing benefits with fewer risks.4 them.6 The discussion highlights the inter- Non-FDI flows show far greater volatility, action of these factors, especially the inter- with sudden reversals.5 Analysis (see below) action of international capital markets with of non-FDI flows (portfolio equity, bonds domestic financial vulnerabilities, which and other debt securities, and bank loans) amplify the risks of a crisis: shows that medium-term bank loans have • Macroeconomic policies may either declined and have been replaced by portfo- exacerbate financial risks or fail to pre- lio flows, which show greater volatility vent boom-bust cycles, often a cause of than FDI, and sudden reversals, as evident financial crises.

Private capital flows are volatile, but FDI has been rising steadily Figure 3-2 Net private capital flows to developing countries, 1975Ð96

Percent of GDP 4.0

3.5

3.0

2.5 Total flows 2.0

1.5

Non-FDI 1.0

0.5 FDI 0

1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1996

Note: Weighted average. Source: World Bank 1998c.

128 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

• Inadequate prudential regulation and acting decisively to prevent a crisis, even premature liberalization of domestic when there are warning signals of vulnera- financial systems (along with poor cor- bility (often for many variables at the same porate governance) may create condi- time) and a crisis is known to be brewing— tions for excessive risk taking by as in Thailand and Mexico (box 3-2). lenders and borrowers. • These two factors, coupled with short- term private-to-private capital inflows Macroeconomic policies to surges (as in East Asia) and premature manage capital flows and capital account liberalization, can cre- reduce financial risks ate even greater risks, and increase the acroeconomic policies designed to likelihood of financial crises. avoid large external and internal • Reliance on capital inflows exposes Mimbalances are a first line of defense in the developing countries to external panics prevention of financial crises. Crises are that may cause sudden and massive often a result of boom-bust cycles, with sig- reversals in capital inflows, deep illiq- nificant interaction between macroeco- uidity, and strong contagion effects. nomic factors and weaknesses in financial Minimizing these risks and dealing more and corporate sectors. For instance, an eco- effectively with such financial crises nomic boom may result when weak regula- would require a better architecture of tion and government guarantees of finan- the international financial system. cial liabilities lead financial institutions to Moreover, political economy con- engage in excessively risky lending (Krug- straints may also prevent governments from man 1998; Corsetti, Pesenti, and Roubini

Box 3-2 Political economy and financial crises

reventive measures to avert a crisis are obviously sitive to inflationary signs, while policymakers in indus- preferable to waiting for one to occur. Policymak- trial countries have a strong collective memory of the ers, however, often respond to other pressures. effects of the Great Depression. Finally, policy conflicts PGovernments may fail to act because politicians abound, and the process of policymaking within countries give greater weight to short-term costs and less to long- is sometimes flawed (in Korea and Thailand, for example, term gains. Bad information and analyses may also play a with limited information sharing between the role. Interest groups likely to lose out from policies that and the Ministry of Finance). would minimize the risks of crises lobby to protect their Some of the most effective banking sector reforms interests. For example, measures to correct banking sys- have taken place only in the wake of major crises, as in tem fragility hurt bank owners, managers, shareholders, Chile in the 1980s and Argentina after the 1994–95 Mexi- and well-connected firms (as in Indonesia) almost immedi- can crisis. Once the domestic financial system is in deep ately, while benefits are long-term and diffused. Countries trouble, with large external borrowing requirements, the that have not experienced financial crisis lack a realistic conflicts in policy may no longer be manageable. Bailing notion of their costs. The lessons of crises influence the out domestic banks only results in more pressure on the behavior of policymakers. The hyper-inflation experience external situation. A “soft landing” scenario may no of the 1920s has left German policymakers extremely sen- longer be practical.

129 GLOBAL ECONOMIC PROSPECTS

1998). Macroeconomic policies can either rates can also result in big losses of compet- lessen or aggravate these risks. Surges in itiveness and misalignments when capital capital inflows can create the conditions for flows surge to high levels. boom-bust cycles and compound macro- economic and financial management prob- Fixed exchange rates and lems—especially in small, open developing sterilization of inflows economies with fragile financial systems Policymakers use fixed or quasi-fixed (McKinnon and Pill 1997; Corbo and Her- exchange rates to reduce uncertainty about nandez 1996).7 Moreover, real exchange exchange rates, avoid nominal appreciation rate movements affect resource allocation, and maintain external competitiveness, and particularly between tradable (export and provide a nominal anchor to preserve domes- import-competing) and nontradable (for tic stability. instance, real estate) sectors. Loss of com- When private capital inflows surge, how- petitiveness for tradables and booms in ever, fixed or pegged exchange rates may nontradables can contribute to strains in become untenable and costly because of the the domestic financial system while aggra- implications for domestic macroeconomic vating external imbalances. goals (such as reducing unemployment), the inflationary pressures they generate, and the incentives they create for private agents to terilization of capital inflows may S overborrow. The pursuit of a pegged nominal work in the short term, but it is contributed to the East Asian increasingly costly over time. crisis, especially in Thailand. The typical initial response to a surge in Government’s room for macroeco- capital flows is to increase official reserves to nomic policy maneuver is often restricted maintain the exchange rate and guard against by important tradeoffs and ineffective sudden reversals. An analysis of 27 episodes instruments. Fiscal policy may be too blunt of capital inflows shows that, on average to offset the effects of volatile capital over the period of surge, one-third of the cap- inflows, while reducing public spending ital account surplus was absorbed by reserve may conflict with other goals. Tighter mon- accumulation. This ratio rises to 50 percent etary policies and sterilization may even when the capital inflow is at the lower (0–3 increase capital inflows, particularly vola- percent of GDP) or the higher range (more tile short-term flows, while an exchange- than 9 percent of GDP; box 3-3).8 rate peg eliminates the effectiveness of Under a currency board arrangement, an and increases incentives extreme form of fixed exchange rate, a coun- for private borrowing abroad. A shift to try formally gives up its autonomy in mone- flexible exchange rates increases the lati- tary policy and strongly anchors its currency tude in monetary policy maneuver but by to a fixed rate (box 3-4). With a pegged itself may be insufficient to control over- exchange rate, however, the authorities tend borrowing and may lead to greater to retain some autonomy in monetary policy exchange rate volatility. Flexible exchange in the pursuit of domestic objectives. During

130 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Box 3-3 Capital inflows and reserve accumulation

onetary authorities frequently intervene to increase foreign reserves when capital inflows A sizeable share of capital inflows begin to surge, in order to preserve stability of goes to reserve accumulation Mthe exchange rate (when the domestic currency Reserve accumulation and capital inflows is implicitly or explicitly anchored to an exchange rate

peg) and to reduce market uncertainty. When capital Percent

began to flow into Morocco in 1990, its foreign reserves 60 were low, and the authorities absorbed 75 percent of the

incoming capital in the first three years. Similarly, in the 50 Czech Republic in 1993, the authorities built up reserves

of roughly 70 percent of capital surpluses for the follow- 40 ing three years. In 27 inflow episodes in 21 developing

countries, reserve accumulation absorbed an average of 30 32 percent of the change in the capital account surplus,

with the extent of reserve accumulation depending on the 20 size of flows. At low levels of capital inflows, reserve accumulation 10 absorbs close to half of the capital account surplus (box figure). This may be due to the buildup of reserves for 0 trade purposes during the initial phase of capital inflows 0Ð3 3Ð6 6Ð9 >9 and determination of authorities to defend the exchange Capital inflow range (percent of GDP) rate. At intermediate levels of inflow (3–9 percent of GDP) reserve accumulation falls to about 20 percent of the capi- Note: Average reserve accumulation as percentage of capital account surplus during inflow surge period. tal account surplus. When inflows are large (exceeding 9 Source: World Bank staff calculations based on data from IMF percent of GDP), the authorities once again intervene International Financial Statistics. aggressively, possibly because of increased perceived risks of reversals. The rate of reserve accumulation is also clearly related to the composition of capital inflows. The to the accumulated reserves and the income earned on larger the non-FDI component of the increase in capital these reserves. This estimated cost for some East Asian inflows, the higher the reserve accumulation. countries (Malaysia and the Philippines) reached about Accumulation of reserves has a social cost (different 0.1 percent of GDP a year for many years. But this cost from the cost of sterilization) measured by the difference may be significantly higher: it was 0.16 percent of GDP a between the cost of servicing the capital inflow equivalent year for the Czech Republic and 0.12 percent for Peru.

the early stages of a surge in capital inflows, and increasing financial system fragility. Eco- the authorities buy foreign exchange (which nomic agents would lose confidence in the immediately expands the supply of domestic authorities’ ability to maintain the peg, and high-powered money) and simultaneously expectations of a would increase, sell domestic bonds or increase reserve possibly leading to an attack on the currency. requirements to sterilize the effects of the The fundamental problem with steril- inflows on domestic . Without ization is the “inconsistent trio” or “open such sterilization, capital inflows would economy trilemma”: any two, but not all expand the domestic monetary base, creating three, features of macroeconomic policy—a a temporary economic and lending boom fixed exchange rate, full capital mobility,

131 GLOBAL ECONOMIC PROSPECTS

Box 3-4 Currency boards—when do they work?

nder a currency board, a country gives up its Speculative attacks on the peg can happen under discretionary power over monetary policy, com- either arrangement, and the required response to such mitting itself to issue no money that is not attacks is to raise domestic interest rates and squeeze Ubacked by reserves and to tolerate the interest domestic credit high enough to stop the attacks. But if the rates that result. The hope is that the very strong com- costs to domestic output (primarily in nontradable sec- mitment to maintaining the value of the currency tors) are severe—as they tend to be if interest rates remain reduces its susceptibility to attack, helping to sustain a high for a long period—chances are the peg will be aban- fixed exchange rate and creating greater confidence. But doned. By ruling out this possibility a currency board cre- does it work? ates greater credibility for the arrangement. Currency A currency board is different from a pegged boards appear to work best for only two groups of exchange-rate system primarily because the authorities economies: small, open ones with large tradable sectors have chosen—through legislation (Argentina and Esto- (Hong Kong [China] and Estonia); and economies that nia) or other arrangements (an automatic link arrange- have been extremely unstable, where a currency board ment, as in Hong Kong [China])—to subordinate domes- would restore badly needed credibility to domestic mone- tic policy and objectives to policies to maintain a fixed tary policy (Argentina and Bulgaria). Other requirements exchange rate. Under a pegged exchange rate, the mone- for successful currency board arrangements include tight tary authority commits to the currency peg as a mecha- fiscal policies, with substantial fiscal surpluses and flexi- nism to maintain low inflation, but can abandon the peg bility in fiscal policy; labor market flexibility; successful in the event of a large shock to output (Obstfeld and high interest rate defense against previous attacks (with- Rogoff 1996). The authority weighs the costs of main- out large residual costs to the economy); and enough ini- taining the peg (lower output, higher unemployment) tial reserves to make the system credible. against the costs of abandoning it (loss of credibility, higher inflation). Source: ADB and World Bank 1998; Obstfeld and Rogoff 1996.

and monetary policy independence—are Sterilization may work in the short feasible (Mundell 1963; Wyplosz 1998; term, but it is increasingly costly over time. Obstfeld and Taylor 1998).9 Sterilization If inflows persist, this strategy becomes even presupposes that independent domestic harder to maintain because of rising fiscal monetary policies can be pursued effec- costs, reflecting the fact that interest rates tively (to control domestic money supply) on domestic bonds exceed the interest that under conditions of international capital central banks earn on foreign deposits mobility. But when exchange rates are fixed abroad.11 Moreover, sterilization leads to or pegged and there is a large degree of cap- higher domestic interest rates, which attract ital mobility (that is, when a country’s further inflows of capital (figure 3-3). Short- financial assets issued in its currency are term capital flows—which tend to be the reasonably substitutable (in private port- most sensitive to interest rate differentials— folios) for other internationally accepted increase, raising the vulnerability to liquid- assets), sterilization policies may be ineffec- ity crises (Montiel and Reinhart 1997). tive, because any contraction or expansion Pegged nominal exchange rates can cre- of the domestic assets of the central bank ate unintended incentives to domestic resi- will give rise to an offsetting capital inflow dents to overborrow, thereby fueling surges or outflow (Montiel 1993).10 in capital inflows. Maintaining the peg (as

132 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Switching to a flexible exchange rate also Sterilization means higher returns autonomy to monetary policy and interest rates and more provides incentives that, in a world of short-term capital inflows greater capital mobility, may reduce the Figure 3-3 Interest rates and sterilization policies, Indonesia likelihood of crises (Goldstein 1995; and Chile Corbo and Hernandez 1996). Flexible exchange rates—whether managed floats, Percent 30 exchange-rate bands (usually with a

25 crawling peg; Williamson 1996), or fully

20 floating exchange rates—offer several ben-

15 efits. Through nominal and real apprecia- tion, exchange rates take the brunt of the 10 adjustment to large capital flows and 5 allow greater independence for domestic 0 Indonesia Chile monetary policy (that is, more effective

Pre-inflow (1988/89Ð1989/90) application of sterilization policies) and 13 Capital inflows and heavy sterilization 1990/91Ð1992) lower inflation. Unintended incentives to

Capital inflows and partial sterilization (1993Ð94) overborrow are avoided because market participants are unsure about the future Source: Asian Development Bank and World Bank 1998. direction of exchange rates. By minimizing the impact of capital inflows on the exter- nal component of high-powered money,14 long as it is credible), as in Thailand prior flexible exchange rates limit the effects of to the recent crisis and in Chile in the late capital flows on the (potentially fragile) 1970s, effectively guarantees against any domestic banking system (Goldstein exchange rate risk to domestic borrowers 1995). Malaysia and Chile, for example, acquiring foreign liabilities.12 It lowers the managed surges in capital flows better cost of borrowing by socializing the than most other countries because of exchange rate risk and allowing private wider targets for exchange rates and capi- borrowing without currency hedging. Nor- tal controls (Corbo and Hernandez 1996; mally, a prudent borrower facing exchange Goldstein 1995). rate risks (and without a natural hedge, While shifting to a floating exchange such as exports) would be expected to rate may limit some of the boom-bust partly or fully hedge those risks in forward effects of capital flows, it may create other exchange markets, thereby lowering incen- problems (Gavin and Hausmann 1996) in tives to borrow abroad. the process of reaching equilibrium. Exchange rates and interest rates may Shifting to flexible exchange rates become more volatile. A large appreciation Placing restrictions on capital mobility can will worsen external competitiveness (espe- return autonomy to monetary policy cially if trade reforms require a deprecia- under a fixed exchange-rate regime. tion), with potentially severe consequences

133 GLOBAL ECONOMIC PROSPECTS

for the sustainability of capital flows in highly open and small economies.15 Further, Crises are more frequent when exchange rates appreciate, they feed under flexible exchange expectations of a lasting boom, reduce rate regimes Figure 3-4 Frequency of crises under domestic interest rates, boost the demand flexible and fixed exchange rate regimes for credit, lower the costs of—and raise Percent demand for—foreign borrowing (in domes- 60 Crises under flexible regimes Crises under fixed regimes tic currency), and raise the returns on 50 domestic assets (stock markets, for instance) 40 to foreign , thereby encouraging 30 more capital inflows. All these elements can 20 continue to support a boom-bust cycle. Indeed, it is the underlying real appreci- 10 0 ation (the price mechanism that can operate 1975Ð81 1982Ð89 1990Ð96

either through nominal exchange rate Note: Ratio of number of crises during period to changes or through domestic nontradable number of countries with flexible and fixed exchange rates, respectively. prices) that puts the boom-bust cycle in Source: World Bank staff estimates based on data from International Monetary Fund 1998b, and IMF place (Corbo and Hernandez 1996). Shift- Exchange Arrangements and Exchange Restrictions, ing to a flexible exchange rate does not pre- various issues. clude a crisis (Khatkhate 1998; IMF 1997a). Indeed, crises are as likely to occur under flexible exchange rates as under fixed other shortcomings with flexible exchange exchange rates, especially if other condi- rate regimes, notably the loss of a nominal tions, such as adequate prudential and reg- anchor and lower inflation gains. ulatory safeguards on the financial sector are not in place (figure 3-4). Countercyclical fiscal policy Under flexible exchange rate regimes, Given large and potentially destabilizing the monetary authorities may attempt to capital flows, a tightening of fiscal policy sterilize a surge in capital inflows or they can help curb borrowing from abroad and may opt not to. If they choose to sterilize reduce appreciation of the real exchange flows, domestic inflation is moderated, rate, but only if higher public savings are domestic interest rates do not fall rapidly, not offset by lower private savings.17 In and capital flows continue, but the fiscal practice, few countries take significant costs may be high. If they choose not to ster- countercyclical fiscal action to temper a ilize, interest rates fall more sharply, reduc- capital inflow boom (Schadler et al. 1993). ing incentives for foreign borrowing, but That places too great a burden on mone- inflation may rise. Often, however, domestic tary policy to restrain aggregate demand, interest rates will remain persistently high in which leads to accumulation of short-term developing countries.16 The incentive for liabilities and increases vulnerability. increased capital inflows thus remains, con- Three factors make it difficult to take tributing to vulnerability. There are also the required fiscal adjustment measures:

134 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

first, the state is to some extent held allocation, privatize financial institutions, hostage to private capital inflows; second, and allow entry and from new the fiscal process is inflexible relative to the private institutions. A growing body of evi- volatility of capital flows; and third, dence shows the importance of strong finan- demand is not met for many essential pub- cial systems and financial deepening for lic goods and services—often those (such as long-run growth and development (King human resources and physical infrastruc- and Levine 1993; Levine 1997; Levine and ture) that might be essential to increase the Zervos 1998). Demirgüç-Kunt and Detra- longer-term efficiency of the economy and giache (1998) find that moving from finan- the absorption capacity of resources from cial repression to liberalization of domestic abroad. The result is that fiscal policy is financial systems results in faster long-run often procyclical, which makes the situa- growth of almost 1 percent per year. tion even worse (as happened in East Asia recently). Domestic financial liberalization Financial liberalization also requires Resorting to other instruments more—not less—and effective prudential If the surge in capital inflows is large, the regulation to ensure the safety and sound- standard tools of macroeconomic policy— ness of financial systems. It also requires shifting to flexible exchange rates, avoid- better corporate governance structures and ing strong sterilization efforts, and imple- arm’s-length relationships between banks menting strong countercyclical fiscal and corporations. These arrangements take adjustment—may prove ineffective or time to build, and without them, financial impractical. Other instruments may be liberalization associated with surges in cap- needed. Indeed, since fragile financial sec- ital inflows often leads to financial crises. tors are a prime vulnerability of develop- These crises are not limited to developing ing economies, policies should support the countries—Scandinavian countries that conduct of prudent macroeconomic poli- undertook financial liberalization at the cies by improving and tightening the pru- end of the 1980s and early 1990s also expe- dential regulatory framework of the finan- rienced these problems.18 cial system (and implementing other A study of the relationship between measures related to external financial banking crises and financial liberalization by liberalization). Demirgüç-Kunt and Detragiache (1998) for 53 countries between 1980 and 1995 found that crises are more likely in liberalized Financial liberalization, financial systems19 (figure 3-5). Several domestic banking reforms, mechanisms link deregulation and liberaliza- and corporate governance tion to crisis (Goldstein and Turner 1996): n the past two decades developing coun- • Increased competition among financial tries have been encouraged to liberalize institutions (from existing banks, the Itheir domestic financial sectors—lift con- entry of new banks, development of trols on domestic interest rates and credit nonbanks, and expansion of capital

135 GLOBAL ECONOMIC PROSPECTS

markets) may lower bank margins, • Rapid credit expansion due to reduced profitability, and franchise values or reserve requirements and a larger effective capital base (Asian Develop- money multiplier released pent-up ment Bank [ADB] and World Bank demand for credit or easier access to 1998). Empirical evidence of this effect foreign resources, and expanded bank on franchise values is found by lending to boom-bust prone activities Demirgüç-Kunt and Detragiache (Caprio, Atiyas, and Hanson 1994). (1998; figure 3-6). The decline in bank Bank credit managers trained in a con- margins and profits may be an objec- trolled environment may not have the tive of financial liberalization, but if skills needed for a riskier environment. excessive competition leads to sharp • Freeing deposit rates with weak declines in franchise values, it may banks, in developing countries and reduce the incentives for prudent bank- even more in transition economies, ing and lead to excessive risk taking by leads to higher deposit and lending bank managers. Sheng (1996) finds rates to reflect the higher risk. This these factors to be responsible for bank tends to attract riskier investors and failures in Argentina, Chile, Kenya, increases the overall portfolio risk of Spain, and Uruguay. banks. An increase in systemic risk • Higher real interest rates often emerge further pushes up interest rates (Mas- following liberalization (Galbis 1993). If sad 1994). firms are operating with high debt- • Many episodes of banking crises are equity ratios, a hike in interest rates can associated with the entry of bank lead to distress borrowing and an inelas- owners bent on engaging in risky and tic demand for credit, which perpetuate questionable activities (as in Chile in high interest rates.20 A bidding up of the 1970s and many transition deposit rates may also weaken banks. economies in recent years).

Banking crises occur more in liberalized financial systems Figure 3-5 Interest rate liberalization and probability of crisis

Probability of occurrence

0.25 With liberalization Without liberalization 0.20

0.15

0.10

0.05

0 Chile India Malaysia (1985) Paraguay (1995) Portugal (1986) Turkey (1991) Note: Countries are classified as crisis cases if the predicted probability is greater than 0.05, which is actual frequency of crisis. Source: Demirgüç-Kunt and Detragiache 1998.

136 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

for example through higher reserve and cap- More competition may ital requirements.22 Developing countries lower bank margins and might temporarily impose limits on credit franchise values growth to avoid the risks associated with Figure 3-6 Financial liberalization and bank franchise value credit booms, especially during rapid trans- formation of the banking system, when the Correlation coefficient supervisory system is insufficiently devel- 0.25 oped.23 Alternately, countries may wait to 0.20 lift constraints or decide to impose more 0.15 stringent and explicit limits and restrictions 0.10 on risky lending activities (and concentra- 0.05 tion of risk), such as real estate, securities, 0 and foreign exchange exposure. -0.05 Finally, careful sequencing of domestic -0.10 and external liberalization is called for. -0.15 Restrictions on the capital account, espe- -0.20 cially on the more volatile capital flows, Return on equity Capital Liquidity should be lifted only after the domestic Note: Pearson correlation coefficient between dummy financial sector has been strengthened. for financial liberalization and variable (based on Bank-level variables which are averaged by country for the period 1988–95). Source: Demirgüç-Kunt and Detragiache 1998. Supporting the financial system and improving corporate governance In view of the benefits and risks from How well the financial system functions domestic financial liberalization, the tran- also depends on the legal framework to sition needs to be carefully managed.21 enforce contracts and protect property Supervisory capacity has to be devel- rights and the state of corporate gover- oped quickly and should precede liberaliza- nance. While these measures are not dis- tion. New bank owners and managers need cussed in detail here, improvements in this to meet the criteria for prudent professional area are nevertheless of vital importance. bankers. Similarly, bank managers, loan When transparency is lacking and corporate officers, and other professional staff need governance is weak, both banking systems to be properly trained. Entry of foreign and corporate sectors are more fragile.24 banks may help achieve this objective. But Cozy relations among banks, government, lifting restrictions on domestic and foreign and corporations weaken market discipline, entry to increase competition and innova- encourage connected lending, increase the tion needs to be monitored to avoid large scope for moral hazard, and foster ineffi- declines in the franchise values of banks cient outcomes. Other signs of weaknesses and excessive risk taking. are loose financial accounting and disclo- Authorities should be vigilant in curb- sure and high leveraging, which facilitates ing lending booms following liberalization, excessive risk taking. Concentration of

137 GLOBAL ECONOMIC PROSPECTS

power in family dominated and politically ing a credit culture in which trust and connected companies, with weak protection expectations of repayment and transac- of minority shareholders, is also common in tion costs are reduced. developing countries. • Restricting connected lending practices. These factors have contributed to weak Such policies would also support the devel- performance and banking sector distress in opment of capital markets and alternatives East Asia, their effects intensified by (equity and long-term debt) to short-term increased access to foreign resources and debt finance and reduce the extent of lever- domestic financial liberalization. Debt-to- age and vulnerability of firms to shocks.26 equity ratios rose significantly in many countries, and economic efficiency and Strengthening domestic banks profitability declined. through better regulation and market incentives Banking reform, strongly oriented toward Banking reform, strongly oriented risk management, is a key ingredient of any toward risk management, is a key long-term strategy to minimize the risks and costs of financial crises. An efficient ingredient of any long-term strategy banking sector with effective supervision to minimize the risks and costs of and regulation helps reduce the distortions financial crises. that increase vulnerability to potential crises.27 The central aim should be to The policy prescriptions to support reduce information asymmetries and development of the financial system and develop a risk management culture in the improve corporate governance are straight- banking sector. Internal systems of risk forward:25 management have to be developed and • Developing accounting, auditing, and strengthened, and best practice techniques disclosure standards to increase the used. Bank supervisors tend to prefer ensur- flow of information, and enhance effi- ing the adequacy of a bank’s internal con- ciency by improving the quality of trols to directly assessing financial condi- investment, reducing misallocations, tions.28 This is important in developing correcting mistakes rapidly, and countries, since the risks facing the banking strengthening business risk assessment sector are especially great because of prob- and the accountability of managers to lems in the state of development and com- shareholders. petitiveness of domestic financial markets, • Setting up the legal infrastructure— and governance, contract bankruptcy laws, debt workout proce- enforcement and bankruptcy, sophistica- dures, enforcement of and tion of bankers and their regulators, extent guarantees—to write and enforce con- of political connections between institu- tracts confidently and to protect and tions and governments, and susceptibility balance the interests of creditors, share- of the economy to domestic and interna- holders, and managers, thereby creat- tional economic shocks.29

138 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Developing country regulation should • More transparency and adequate infor- take account of the strengths and weak- mation on the soundness of banks. nesses of financial systems and of regula- • Public financial safety nets that boost tors. Regulations, controls, and restrictions confidence in the financial system but that produce inefficiencies and distortions also limit induced distortions, such as should be abandoned. However, the vul- explicit or implicit government guaran- nerabilities and frequent failures of finan- tees, that encourage excessive risk taking. cial systems indicate that some restraints • Effective regulatory and supervisory are needed (World Bank 1998b, chapter 6). oversight for controlling risk and limit- For instance, mild restraints on deposit ing the adverse impact of official safety interest aimed at creating franchise value nets. for banks may induce outcomes that are • Transparent ownership structure that more efficient than financial repression enhances competitive behavior, and (where interest rates are kept at low nega- limits on connected lending. tive real levels, inducing inefficient deepen- The Basle committee’s core principles ing and misallocation of resources) or can also be extended to include accounting immediate financial liberalization (Hell- and information disclosure, loan classifica- man, Murdock, and Stiglitz 1996, 1997; tion, and bankruptcy regimes. International World Bank 1998b). Financial restraint accounting and auditing standards are also features played a significant role in available.32 International financial institu- improving stability in East Asian countries tions can help countries adopt and imple- in the past; while some of the other fea- ment these regulations.33 tures such as market incentives have been Recommending that developing coun- implemented with some success in Chile, tries build a sound and healthy financial New Zealand, and the system according to these principles is not (Nicholl 1997; Caprio and Klingebiel sufficient, however. Building such systems 1996b; Goldstein and Turner 1996). takes a long time, and it is hard to deter- Banking regulation and supervision. mine a minimal requirement for the quality Weak regulation and supervision are the of the banking sector. Also, adjustments are most widely recognized sources of vulnera- needed to take account of specific features bility in developing countries’ banking sys- in developing countries. tems.30 Most industrial countries subscribe Incentives and market discipline. Rely- to “Core Principles for Effective Supervi- ing heavily on regulation and supervision sion” of the Basle Committee in the design to control excessive risk taking is of ques- of banking regulation and supervision to tionable efficacy, particularly for develop- reduce vulnerability of the financial sys- ing countries (Caprio and Klingebiel tem.31 In addition to macroeconomic sta- 1996b; Caprio 1997; Goldstein and Turner bility, the building blocks include: 1996). It takes too long to develop supervi- • Higher standards of competence and sory capacity and skills. Moreover, super- integrity of bank management, as well visors are often unable to detect risky as effective management controls. behavior and take action against troubled

139 GLOBAL ECONOMIC PROSPECTS

banks, because the kinds of behavior tend These structural features have implica- to change over time and supervisors are tions for the institutional framework of the not prepared for them. They may also be banking sector in developing countries, prevented by policymakers from taking such as a need for higher capital-adequacy action. ratios than the Basle international standard These factors argue for relying more on (see Goldstein and Turner 1996; and Hono- market-imposed discipline and improving han 1997) and for more stringent limita- incentives for prudent banking. In addition tions on the concentration of risks (such as to raising capital adequacy ratios, this strat- loans for real estate or securities).37 egy could include:34 increasing the financial The role of government, guarantees, and personal liability of managers and and moral hazard. Government often plays directors (going to unlimited liability), or a pervasive role in the banking sector in introducing mutual liability; requiring a tier developing countries. This generates serious of uninsured subordinated debt for individ- conflicts, especially moral hazard problems, ual banks (to increase the incentives for pri- that are a major underlying factor in risky vate monitoring of banks); and requiring lending. Necessary reforms include: banks to regularly publish key information, • Above all, severely limiting the govern- such as credit ratings. Clear exit policies ment’s role in directly running and and resolution mechanisms should also be managing banks. This can mean priva- spelled out, covering automatic or struc- tization of banks, which has to be han- tured early intervention and graduated dled carefully, especially with respect to responses by the authorities as bank capital reaches some predetermined thresholds (Caprio 1997). Volatility is linked to Volatility and the need for more strin- banking crises gent regulations. Developing countries Figure 3-7 Bank crises and volatility, share structural characteristics that subject 1980Ð94 them to greater volatility. These include an Average coefficient of variation 0.9 unstable macroeconomic environment, con- Systemic banking cases 0.8 centrated economic activity and exports, Borderline cases Noncrisis cases 0.7 and susceptibility to greater shocks—terms of trade, weather, interest rates, and policy 0.6 volatility (figure 3-7).35 0.5 Vulnerability to external shocks and, 0.4 especially, to changes in international inter- 0.3 est rates, has been shown to be the most 0.2 36 important factor in banking crises. A 0.1

reversal of macroeconomic conditions in 0 Gross Consumer Terms capital-exporting countries leads to higher domestic product price index of trade

interest rates, curtailed capital inflows, and Source: Caprio and Klingebiel 1996b. slower growth of bank lending.

140 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

pricing (and its effect on franchise val- Benefits and associated ues), treatment of risks, capital ade- risks of capital account quacy, and management. liberalization • Reducing the government’s direct apital account (or external) liberaliza- involvement in allocating credit and in tion and financial integration with providing guarantees to commercial Cworld capital markets can potentially bring enterprises so as to enhance market-ori- large benefits, and both have been advo- ented banking behavior. Because of cated for developing countries for that rea- moral hazard, implicit or explicit gov- son.38 Letting domestic agents trade finan- ernment guarantees can lead to exces- cial assets with foreign economic agents sive risk taking. Banks tend to raise may increase access to capital and lower its money at safe rates and lend at premium cost. Productivity improvements, risk diver- rates to finance speculative investments sification, and consumption-smoothing are beyond prudent levels (McKinnon and other potential benefits. Pill 1997; Krugman 1998). Many developing countries have liber- • Setting up a formal deposit insurance alized capital accounts in the past decade scheme to deal with the negative exter- (box 3-5), but recent experience suggests nalities that individual failures may have that such liberalization and increased finan- on the rest of the banking system. Insured cial integration can sharply raise the risks depositors, however, have little incentive of financial crisis (box 3-6). to monitor banks, and regulators may engage in regulatory forbearance and delay action against troubled banks. Many developing countries have Indeed, Demirgüc-Kunt and Detragiache liberalized capital accounts in the past (1997) find that deposit insurance has a decade, but recent experience suggests significant positive effect on the likeli- this can sharply raise the risks of hood of a banking crisis. Thus, supervi- financial crisis. sion, minimum capital requirements, and mandatory issues of subordinated debt In fact, the duality of benefits and risks would help reduce moral hazard and of international capital mobility is induce banks to reduce their risks. In inescapable in a world of asymmetric infor- addition, there should be limits on the mation (Obstfeld 1998), where lenders do amounts insured, and co-insurance not know as much as borrowers about the should be required (that is, covering less uses of their money and are therefore prone than 100 percent of deposits), as well as to panic. Thus, the benefits of capital account charging risk-weighted deposit insurance liberalization and increased capital flows premiums. Policymakers should also con- have to be weighed against the likelihood of sider mutual liability for banks, clear pro- such crises and their costs. Recent discussions cedures for closing insolvent banks, and at international forums have heightened the possibly private provision and manage- recognition of the issues, especially in rela- ment of the insurance program. tion to volatile short-term flows.

141 GLOBAL ECONOMIC PROSPECTS

Box 3-5 How far has capital account liberalization progressed?

he OECD’s Code of Liberalization of Capital Most industrial and developing countries still had Movements of 1961 (extended to include all capi- some type of capital controls at the end of 1997, tal account transactions by 1989) and the - mainly on direct investment (143 countries), real T pean Union’s 1988 Second Directive on Liberal- estate transactions (128), and capital market securities ization of Capital Movements were milestones in the (127). In addition, most countries implement provi- liberalization of industrial countries’ capital accounts. It is sions specific to commercial banks and other credit only since the late 1980s and early 1990s that most indus- institutions (152). trial countries have accelerated the pace of capital-account Only a few industrial countries (Luxembourg and liberalization. The number of industrial countries with the ) and developing countries (Armenia, neither separate exchange rates nor restrictions on pay- Djibouti, El Salvador, Panama, and Peru) report no ments for capital transactions increased from 3 in 1975 to capital controls, and a few report just one type of con- 9 in 1985 and 21 in 1995. The number increased in devel- trol (Canada, Denmark, Mauritius, Uganda, and oping countries as well, from 20 in 1975 to 31 in 1995. Paraguay).

Most countries still have some form of capital controls Controls on capital-account transactions, year-end 1997

Total Developing countries Industrial countries

Number of IMF member countries 184 157 27 Controls Capital-market securities 127 112 15 Money-market instruments 111 102 9 Collective investment securities 102 97 5 Derivatives and other instruments 82 77 5 Commercial credits 110 107 3 Financial credits 114 112 2 Guarantees, sureties, and financial backup facilities 88 86 2 Direct investment 143 126 17 Liquidation of direct investment 54 54 0 Real estate transactions 128 115 13 Personal capital movements 64 61 3 Provisions specific to: Commercial banks and other credit institutions 152 137 15 Institutional investors 68 54 14

Sources: Quirk and others 1995; Mathieson and Rojas-Suárez 1993; International Monetary Fund 1996, 1997b, and 1998c.

Evidence suggests that for FDI and sim- tion between these two categories of ilar long-term foreign capital flows, the inflows is not watertight (see further benefits are significant and the risks low. below), in practice, the effects are clearly The benefits of capital account openness to differentiated. There are larger benefits and short-term debt and other volatile non-FDI fewer risks for FDI-type flows, which tend flows are less certain; the greater volatility to be more resilient in times of crises and to of such flows is strongly associated with carry important benefits beyond finance, financial crises. While the clear demarca- than for short-term flows. Thus, developing

142 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

countries should tailor their openness to The analysis below highlights the evi- their capital inflow needs and their ability dence for the benefits and risks of capital to bear the risks. flows, differentiated mainly by FDI and In addition to foreign direct invest- non-FDI flows. While it is simplified, it ment and trade credits, capital flows can illustrates the major issues and their policy range from pure debt such as short- and implications, which in practice have to take medium-term bank loans, to long-term account of the variety of financial instru- bonds, very long-term debt (century and ments, the mechanisms of their intermedia- perpetual bonds), quasi-equity (such as tion, and the use of the associated resources. convertible bonds), and portfolio equity flows. The extent of use of these financial Benefits of capital account instruments by developing countries liberalization and capital flows reflects investors’ preferences in terms of The theoretical benefits from capital risk sharing between the parties in the account liberalization include increased source and destination countries, currency access to capital and faster productivity exposure and maturity risk to the develop- growth, risk diversification, and consump- ing country firm, and extent of diversity of tion smoothing.39 sources of finance using the instrument Capital accumulation and growth. Ben- (table 3-1). efits include increased investment and more Derivatives can also reduce the cost efficient allocation of resources, which and risk developing country firms face in result from taking advantage of differences accessing international capital markets. among countries in the productivity of cap- For example, through interest rate swaps, ital and opportunities for risk diversifica- borrowers can assume the kind of liability tion. Incomplete risk markets discourage they prefer (fixed or floating rate) at a investors from undertaking risky projects, lower interest rate than through regular many of which have high potential returns. borrowing. Currency-swaps enable bor- By allowing more risk diversification, more rowers to match the currency composition of these projects will be undertaken, lead- of assets and liabilities. ing to higher expected returns. The

In capital markets, the risk depends on the type of borrowing Table 3-1 Financial instruments and their risks

Instrument Risk sharing Currency exposure Maturity risk Diversity of sources

Borrowing facilities Low High High Low Syndicated bank loans Low High Moderate Low Straight bonds Low High Moderate/low High Leasing Moderate High Moderate Low Limited recourse financing Moderate Moderate/low Moderate Low Quasi-equity instruments Moderate Moderate Moderate High Portfolio equity investment High Low Low High

Source: World Bank.

143 GLOBAL ECONOMIC PROSPECTS

Box 3-6 Are capital flows the main culprit?

ragile domestic financial systems are often the entry into the OECD, Korea liberalized the ability of its root cause of a financial crisis, and while capital banks to borrow (short-term) abroad (instead of tighten- inflows are also blamed, they are but one element. ing safeguards), there was a massive surge in such F Risks carried by capital inflows and excessive bor- inflows; their reversal subsequently precipitated the crisis. rowing are important. In the Republic of Korea, exces- The problems may have been aggravated by Korea’s sive domestic financial risk taking—including low equity retaining tight controls on FDI inflows, preventing pre- and heavy bank borrowing—was a long-standing prac- cisely the type of flows that it should have encouraged— tice. What may have tipped the balance in the 1997 cri- more stable, longer-term flows that would have brought sis, however, was capital flows: when in the context of its equity, technology, and better risk management.

expected results are higher capital accumu- ated with just $0.60 of additional invest- lation and productivity growth. Benefits ment, however. One reason is that a signifi- vary by type of flow. cant share of such inflows goes into reserve For FDI and similar long-term, rela- accumulation and results in a net social tively stable, flows the benefits are well loss, rather than a gain (see box 3-3).41 Pru- documented. FDI flows accounted for 5–6 dent behavior also implies that short-term percent of aggregate investment in develop- financial resources should not be used to ing countries in the 1990s, above the 1–2 finance long-term investment projects. Tak- percent of the previous 15 years (World ing this into account, and using the sample Bank 1997a). FDI also tends to “crowd in” average capital-output ratio of 2.5 and elas- more domestic investment: every $1 of FDI ticity of output with respect to capital stock in developing countries is associated with of 0.4, a 1 percentage point increase of $0.50–$1.30 of additional domestic invest- non-FDI capital inflows in GDP would be ment. While it is difficult to establish expected to generate additional growth of causality, increased FDI flows are generally only about 0.10 percent of GDP in gross associated with faster aggregate long-run terms and less in net (GNP) terms.42 Some growth (and total factor productivity benefits on productivity could also be growth), with each percentage point expected, especially in low-savings coun- increase of FDI in gross domestic product tries where non-FDI inflows might make (GDP) associated with a 0.3–0.4 percentage possible highly productive investments. point faster growth in per capita GDP Again, while there are direction of causality (Wacziarg 1998). issues and results should be interpreted For non-FDI—particularly short-term with caution, simple correlation on a sam- debt and more volatile flows—the benefits ple of countries showed little evidence of a are less certain.40 Among 18 countries that significant positive association between received significant private capital flows in non-FDI inflows and productivity growth; the late 1980s and early 1990s, the surge in in the one case that showed statistically sig- such capital flows was associated with nificant association—the subsample of low- increased investment, as expected; each $1 savings countries—the association was of non-FDI inflows appears to be associ- strongly negative (figure 3-8).

144 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Benefits depend on the policy environ- there significant losses in terms of slower ment. The opposing signs and different val- growth when countries have capital controls, ues of correlation shown in figure 3-8 especially on short-term and portfolio flows? imply that the impact of non-FDI capital Based on the experience of 20 OECD coun- flows and capital account openness may tries in 1950–89, Alesina, Grilli, and Milesi- depend more on the economic and policy Ferretti (1994) find no negative impact of environment in individual countries than capital controls on GDP growth. Using a on the extent of capital account openness simulation model, Razin and Yuen (1994) and flows themselves. This is the case for show that the long-run effects of liberalizing FDI inflows (World Bank 1997b) and offi- capital flows are very modest. Rodrik (1998) cial aid flows (Dollar 1998), for both the uses a GDP per- capita growth equation and size and direction of impacts on productiv- a simple index of capital account openness ity depend on the policy environment. with a sample of almost 100 industrial and Some indirect evidence about the bene- developing countries for 1975–89 and finds fits of non-FDI inflows can also be inferred that capital account convertibility has no sig- by looking at the counterfactual case: are nificant effect on growth once other effects are taken into account (figure 3-9). Car- rasquilla (1998) finds similar results for There is little connection 1985–95 for 19 Latin American countries between non-FDI inflows using more direct measures of capital con- and productivity growth trols (figure 3-10). Figure 3-8 Correlation between capital Risk sharing and consumption smooth- inflows and total factor productivity ing. In developing countries characterized growth in low- and high-savings countries by a concentration of exports and economic activity, allowing domestic banks to diver- Bivariate correlation sify their portfolio helps reduce their vulner- 0.6 Non-FDI ability to external (terms of trade) and inter- FDI 0.4 nal output shocks. The scope for gains from open capital accounts may also arise from 0.2 risk sharing and asset diversification. The 0 evidence for such gains is based on simula- tion models whose results are mixed. Obst- -0.2 feld (1995) estimates that the potential gains

-0.4 may be very significant, while Tesar (1995) finds them small. Levine and Zervos (1998) -0.6 find no evidence of significant effects on

-0.8 growth of international risk sharing through Low-savings High-savings All increased integration of stock markets. Note: Sample of 18 countries receiving substantial Another potential source of welfare capital inflows in the late 1980s and early 1990s. Source: World Bank staff estimates. improvement from capital flows is increased opportunities for consumption

145 GLOBAL ECONOMIC PROSPECTS

There is no evidence… …that capital controls slow Figure 3-9 and capital growth account liberalization in 100 countries, Figure 3-10 GDP growth and capital 1975Ð89 controls in Latin America, 1985Ð95

GDP growth GDP growth

0.06 7 ◆ ◆

6 0.04 ◆ ◆ ◆ 5 ◆ ◆ ◆ ◆ ◆ ◆ 0.02 ◆ ◆ ◆ 4 ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆◆◆ ◆ ◆ ◆ ◆ ◆ ◆◆ ◆◆◆ ◆◆ ◆ ◆◆ ◆ ◆ ◆ 3 0 ◆◆◆ ◆ ◆ ◆◆ ◆ ◆◆ ◆ ◆◆ ◆◆ ◆ ◆ ◆ ◆◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆◆ ◆ ◆ ◆ ◆ ◆ ◆ ◆ 2 ◆ ◆ ◆ ◆ ◆ ◆ ◆ -0.02 ◆ 1 ◆ ◆◆ ◆ ◆ -0.04 0

◆ -1 ◆ ◆ -0.06 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 -2 0 0.5 1.0 1.5 2.0 Capital account liberalization index Index of capital controls Note: The figure is a partial scatter plot (controlling for per capita income, secondary education, quality of governmental institutions, and regional dummies Source: Carrasquilla 1998. for East Asia, Latin America and the Caribbean, and Sub-Saharan Africa). Source: Rodrik 1998.

have reduced the volatility of consumption relative to that of income, on average they smoothing in the presence of high income are associated with increased volatility. volatility.43 A country that is isolated finan- Results for a sample of 17 countries that cially would have to accommodate any gained significantly greater access to pri- external shock through changes in con- vate capital flows show that volatility dur- sumption and investment. In contrast, a ing the inflow surge remained higher for country that is well integrated with world consumption than for income, with the dif- financial markets can lend and borrow and ference increasing in 10 countries. Thus, thus maintain consumption and investment the gains from consumption smoothing close to desirable levels—even when appear uncertain and limited. national income is fluctuating. The gains may be larger for developing countries with Risks associated with capital more income volatility.44 The general obser- account openness vation, however, that capital flows tend to The risks associated with capital account lib- be procyclical in developing countries indi- eralization and capital flows for a develop- cates that consumption smoothing is not ing country depend on the ability of policy- significant.45 More detailed evidence also making institutions as well as the financial suggests that while capital inflows may and corporate sectors to adjust to shocks

146 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

and absorb risk, as well as on its own volatil- there has been a negative correlation ity. Various financial assets traded interna- between capital flows and the lifting of cap- tionally differ in their volatility and implica- ital controls (IMF 1997a). At the same tions for increased vulnerability to crisis,46 time, currency crises have increased, which but three arguments link financial integra- may indicate causality between capital tion and increased risks of financial crises. account liberalization and currency crises The first is that openness to capital (Wyplosz 1998). Most empirical analyses, flows may increase the risk of currency however, have failed to find statistical evi- crises if surges and reversals of capital dence linking the volume of capital inflows flows (and crises) occur independently of a to crises (Sachs and others 1996). An country’s policies and actions. 47 When exception is Radelet and Sachs (1998), who international interest rates rise, interna- find some evidence of a relationship tional investors are likely to cut back their between crises and capital account deficits financing to developing countries. At the (but not current account deficits). Frankel same time, the capacity of developing coun- and Rose (1996) also find that higher FDI try banks, firms, and governments to ser- flows (relative to debt) are associated with vice debt is reduced. There is strong empiri- a lower probability of crises. cal evidence that international interest rates There is additional indirect evidence are a major determinant of non-FDI capital linking capital flows with crises for a sample flows,48 and are a big factor in the proba- of 27 capital inflow surges in 21 countries bility of crises (Frankel and Rose 1996; (table 3-2). In 1996 these countries Kaminsky and Reinhart 1997). Foreign accounted for 69 percent of private flows to interest rates and a volatile external envi- developing countries (or 83 percent, when ronment have also been found to be signifi- cant determinants of banking crises and, therefore, indirectly of currency crises (see The risks associated with capital discussion of domestic financial sector). account liberalization depend on a A second argument against financial developing country’s ability to adjust integration is that international capital to shocks and absorb risk, and on the market failures can aggravate domestic volatility of flows. financial weaknesses and have contagion effects. A third is that integration, while not China is excluded). The mean ratio of total the root cause of financial crises in emerg- private-to-private capital flows to GDP over ing markets, may contribute to crises whose the inflow periods ranges from 2.2 percent origin is domestic—especially given weak to 11.8 percent. The composition of these financial systems and inappropriate macro- inflows varies considerably, with the mean economic policies.49 ratio of FDI to non-FDI private-to-private While there is little direct evidence of flows ranging from a negative 0.1 to 3.4. In the role of capital account liberalization or about two-thirds of the cases, there was a capital inflows in financial crises, there is banking crisis, currency crisis, or twin crises some indirect evidence. Since the 1980s in the wake of the surge.

147 GLOBAL ECONOMIC PROSPECTS

Volatility of non-FDI and portfolio flows, the risks are small because these flows. The risks associated with capital flows respond more to longer-term consid- account liberalization hinge on the volatil- erations than to short-term international ity of capital flows and the risks of reversal interest rates, and because they interact less during bad times, when access to additional with domestic financial markets. The risks financing is especially important. For FDI of large reversals are even lower because

Indirect evidence links capital inflow surges with crises Table 3-2 Surges in private-to-private net capital inflows and financial crises

Mean ratio of Mean ratio of annual capital FDI to non-FDI Country Inflow flows to GDP capital inflows Crisis following inflow episode

Argentina 1991–94 2.5 1.0 1994–95 banking crisis following Mexican devaluation Brazil 1992–96 3.1 0.2 1995 banking crisis Chile 1978–81 11.1 0.1 1982–83 currency and banking crisis Chile 1989–96 5.1 0.7 No crisis Colombia 1992–96 4.4 1.2 No crisis Costa Rica 1986–95 5.5 1.0 No crisis Czech Republic 1993–96 8.3 0.6 1997 currency crisis Estonia 1993–96 5.4 3.4 1997 near-crisis Hungary 1993–95 11.8 1.1 1995 crisis India 1994–96 2.5 0.3 No crisis Indonesia 1994–96 3.7 1.1 1997 crisis Korea, Rep. of 1991–96 2.5 –0.1 1997 crisis Malaysia 1982–86 3.1 (a) 1985–88 banking crisis Malaysia 1991–96 9.8 2.5 1997 crisis Mexico 1979–81 2.5 0.7 1982 crisis Mexico 1989–94 4.5 0.6 1994/95 financial crisis Morocco 1990–96 3.2 0.6 No crisis Pakistan 1992–96 3.5 0.4 No crisis Peru 1988–96 6.9 0.4 No crisis Philippines 1989–96 4.5 0.5 1997 crisis Philippines 1978–82 3.0 0.0 1981 banking crisis 1983–84 currency crisis Sri Lanka 1991–95 5.3 0.3 No crisis Thailand 1978–84 3.0 0.3 1983 banking crisis 1984 currency crisis Thailand 1988–96 9.4 0.2 1997 crisis Tunisia 1992–96 3.6 2.5 No crisis Venezuela 1992–93 2.2 0.0 1993–94 banking crisis 1995 currency crisis Venezuela 1976–79 3.9 –0.1 1980 banking crisis

Note: The inflow episodes were selected based on the length (minimum of two years) and the volume of total private-to-private capital flows as a percentage of GDP (minimum ratio of 2 percent). Total private-to-private capital flows = total private capital flows – public and publicly guaranteed private creditors. Total private capital flows = total flows – official flows – net use of IMF credit. Total flows = foreign direct investment + portfolio investment + other investment + net errors and omissions. Total non-FDI private-to-private capital flows = total private-to-private capital flows – foreign direct investment. a. The mean ratio is very high and negative, reflecting a very low negative denominator (non-FDI private-to-private capital flows). Source: IMF 1996 and 1997; World Bank 1998a; Kaminsky and Reinhart 1997.

148 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

FDI inflows are usually invested in longer- public) is volatile and intensifies the sever- term assets (plant and machinery, and ser- ity of financial crises. Non-FDI and debt vices, for instance) that cannot be liqui- portfolio flows increase in the years just dated quickly. For non-FDI flows, there are before a crisis, then reverse sharply after clearly more risks of volatility and rever- the crisis occurs. These features magnify sals, but they may differ according to vari- boom-bust cycles and, hence, the severity of ous categories.50 financial crises in small, financially open Volatility of non-FDI flows and stabil- developing countries. ity of FDI flows for three countries— Volatility of short-term interbank Argentina, Mexico, and Hungary—suggest flows. Interbank borrowing also tends to be different characteristics and behavior, par- highly volatile. The reversal in flows from ticularly in times of downturns (figures 3- Bank for International Settlements (BIS) 11 to 3-13). FDI is far less volatile and less reporting banks in Korea and Thailand was subject to reversals.51 It even continues to dramatic in the second half of 1997 increase in downturns. Non-FDI private-to- (figure 3-14). The in both private flows, in contrast, are much more countries largely reflects this reversal, par- volatile. Portfolio equity most closely ticularly in short-term interbank credit resembles FDI, but is more volatile. Debt lines. In contrast, FDI flows held up, at portfolio investment (including private-to- least in the first half of 1998 (when they

Portfolio equity flows… …closely resemble FDI… Figure 3-11 Net capital flows to Figure 3-12 Net capital flows to Mexico, Argentina, 1990-96 1990Ð96

Millions of U.S. dollars Millions of U.S. dollars

25,000 30,000 ◆ Portfolio debt 20,000 25,000 ◆ ◆ 15,000 FDI 20,000 Non-FDI 10,000

15,000 5,000

10,000 0 ◆ ◆

Portfolio equity ◆ -5,000 5,000 ◆ ◆ ◆ Portfolio debt FDI -10,000 0 ◆ Portfolio -15,000 equity ◆ -5,000 ◆ -20,000 Non-FDI ◆ -10,000 -25,000 ◆

1990 1991 1992 1993 1994 1995 1996 1990 1991 1992 1993 1994 1995 1996

Source: International Monetary Fund, World Bank. Source: International Monetary Fund, World Bank.

149 GLOBAL ECONOMIC PROSPECTS

What are the implications? …but debt investment is The larger risks and uncertain benefits of volatile portfolio and short-term flows for coun- Figure 3-13 Net capital flows to Hungary, 1990-96 tries with weak institutional capability and financial systems suggest proceeding care- Millions of U.S. dollars 5,000 fully with capital account convertibility. FDI

4,000 Because the risks stem largely from the dis- Portfolio debt tortions and externalities associated with 3,000 ◆ ◆ international borrowing and from the 2,000 ◆ wedge between social and private rates of 1,000 Portfolio equity return, and social and private risk, policy 0 should attack distortions at or close to their

-1,000 source. Since the capacity to implement Non-FDI ◆ ◆ ◆ such policies and their effectiveness may -2,000 not be perfect, this approach must be prag- -3,000 ◆ matic and take account of developing coun- -4,000 tries’ specific conditions. 1990 1991 1992 1993 1994 1995 1996 The first step is to eliminate tax incen- Source: International Monetary Fund, World Bank. tives and other distortions that encourage short-term capital inflows. Another is to use prudential regulations on currency and were higher than in 1997), even though the maturity positions by banks. The Basle Core prospects for the following period will also Principles for Effective Banking Supervision depend on the global situation. recommend only that banking supervisors

The reversal in short-term credit in the second half of 1997 was dramatic Figure 3-14 Rate of change of total debt outstanding by BIS-reporting banks to banks and nonbanks

Percent 35 Rep. of Korea Thailand 25

15

5

-5

-15

-25

-35 Banks Nonbanks Banks Nonbanks Dec. 1995 Ð June 1996 June 1996 Ð Dec. 1996 Dec. 1996 Ð June 1997 June 1997 Ð Dec. 1997

Source: Bank for International Settlements.

150 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

ensure that bank managers set appropriate This often implies maintaining or reinforc- limits and implement adequate internal con- ing capital account restrictions. For coun- trols on foreign currency exposure. But tries that are reintroducing such restric- developing countries should also introduce tions, this may mean loss of credibility, so specific limits on currency and maturity such actions have to be managed in a way mismatches (for example, requiring mini- mum liquid foreign currency assets to cover short-term foreign currency liabilities), and The larger risks and uncertain benefits prudential regulations limiting the aggregate of portfolio and short-term flows for open currency positions of banks, including countries with weak institutional derivatives. But because even well-managed firms and financial institutions have run capability and financial systems suggest into severe losses through the use of such proceeding carefully with capital instruments, there is also a need for better account convertibility. supervision of these regulations and of risk management procedures. Countries may that does not lead to even greater loss of also introduce more stringent liquidity confidence. The imposition of capital requirements in terms of foreign assets rela- account restrictions, as part of a preventive tive to foreign liabilities than for domestic package to minimize the risks of financial currency liabilities. crisis, is concerned mainly with capital Prudential regulations of banks and inflows. Their reintroduction for capital financial institutions does not resolve the outflows during a crisis poses many diffi- risk of excessive exposure by the corporate cult problems, not considered here. sector. Banks may satisfy foreign currency Restrictions on capital flows should exposure requirements by borrowing in for- minimize distortions and be as market- eign currency and lending in foreign cur- oriented as possible. One way is explicit rency to domestic firms. If domestic corpo- taxes or reserve requirements on foreign rations do not have foreign exchange cover, exchange liabilities according to holding the currency risk for banks is transformed period. In Chile and Colombia implicit taxes into a . Thus, additional measures have substantially shifted the composition of are needed for domestic corporations. These such flows and discouraged short-term flows may include requiring disclosure of short- without having much impact on the volume term and unhedged borrowing, reducing the of flows (box 3-7; World Bank 1997b; Mon- tax deductibility of such borrowing, and the tiel and Reinhart 1997). Restrictions on cap- rating of firms raising funds abroad and list- ital flows have to reflect specific factors, ing on the domestic .52 such as administrative capability, and have When the domestic regulatory and to balance the need to be comprehensive in supervisory system for banks is weak, con- order to minimize distortions and evasions trols over corporations are ineffective, and with the need to discriminate between capi- access lender of last resort is uncertain, tal inflow categories, according to the bene- restrictions on capital flows may be useful. fits and risks associated with such flows.

151 GLOBAL ECONOMIC PROSPECTS

The international financial with the outbreak of the East Asian crisis system and its global spread. The Group of 22 he international environment plays an countries (G-22) established three working important part in financial crises in groups on enhancing transparency and Temerging markets. Volatility in interna- accountability; strengthening financial sys- tional interest rates and economic growth tems; and managing international financial in industrial countries affect the allocation crises. These working groups have now of assets to emerging markets and create finalized and submitted their reports,54 and risks of booms and reversals in capital discussions of these proposals in official flows. Other characteristics, such as volatil- forums began in early October 1998. The ity and sudden shifts in market sentiment G-7 countries have since agreed on a num- associated with euphoria, panics, herd ber of specific initiatives to strengthen the behavior, and contagion are also influen- international financial system (Group of tial. These failures in international financial Seven 1998). These include, in the immedi- markets have implications for international ate context, an enhanced IMF facility to financial institutions. provide a precautionary line of credit and a Proposals for reforming the interna- World Bank emergency facility to provide tional financial system architecture have support to countries at times of crisis for the been under discussion since the Mexican protection of vulnerable groups and finan- crisis. A working group, under the auspices cial sector restructuring; and, in the longer- of the Group of 10 (G-10) industrial coun- term, agreement on other principles to tries, drafted the Resolution of Sovereign strengthen the global financial system, Liquidity Crises, which focuses on sovereign including greater transparency, enhanced bonds.53 Discussions have gained urgency surveillance, orderly and progressive capital

Box 3-7 Restrictions on capital flows in Chile and Colombia

hile introduced restrictions on capital inflows in Chile has since lowered the reserve requirement to zero. 1991 through unremunerated reserve require- It is difficult to gauge the effects these restrictions ments (World Bank 1997b). These reserves, have on the volume of flows, as a change in flows could Cwhich have to be maintained for one year regard- also be caused by other macroeconomic and financial less of loan maturity, constitute an implicit tax on foreign developments. The restrictions in Chile and Colombia can borrowing that varies inversely with the holding period. In be thought of as an implicit tax that significantly increased 1995, reserve requirements were extended to all types of the interest differential between domestic and foreign foreign financial investments, including American deposi- short-term interest rates. Econometric studies that use this tory receipts. Colombia introduced capital controls in approach to estimate their effects suggest that they sub- 1993 through unremunerated reserve requirements on stantially changed in the term structure of external bor- direct external borrowing with a maturity of less than 18 rowing—discouraging short-term inflows—and encour- months. These were subsequently tightened, requiring aged equity investment in Chile and Colombia (Cardenas reserves for all loans with maturities of less than five years. and Barrera 1997; Quirk and others 1995).

152 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

account liberalization, orderly resolution of A third argument is based on the risks future crises, and the need for good prac- caused by contagion and the potential spread tices in social policy to protect the most vul- of panic among international investors (the nerable. Other announcements include the Asian crisis provided yet another striking need to pursue further proposals for example of this). When a vulnerable currency strengthening prudential regulations in is attacked, the attack may spread to other industrial countries to promote safe and countries’ currencies, even when their funda- sustainable capital flows, strengthening mentals are sound. A lender of last resort financial systems in emerging markets, and would provide reserves to emerging markets improvements in other related areas. This threatened by speculative attacks and thus section considers five main issues that are prevent a currency collapse. still evolving and remain subject to some A final justification for a lender of last debate.55 resort is on social welfare grounds. While market participants should bear the conse- An international lender of last quences of their actions and incur the costs resort? of a crisis, some costs are borne by groups Arguments have been advanced for an inter- not responsible for the crisis, particularly national of lender of last resort, but such the more vulnerable. arguments also raise unresolved issues. The G-7 adopted the principle of estab- The traditional argument concerns the lishing a precautionary bilateral and multi- possibility of systemic risk. If a country lateral line of credit to countries that are at fails to serve bank debt—whether sovereign risk and pursuing strong IMF-approved or private—it may undermine the liquidity policies—to be drawn only in the event of a and even the solvency of banking systems in creditor countries. This risk was clearly present during the debt crisis of the 1980s, BIS banks’ exposure to emerging when BIS reporting banks’ direct exposure markets is much less than in the 1980s to major emerging markets exceeded their Table 3-3 Commercial (BIS-reporting) banks’ capital, but it was much weaker in the 1997 exposure to emerging markets East Asian crisis (table 3-3). This argument (debt as percentage of banks’ capital) has lost some of its force with the greater Major East Latin emerging risk diversification by banks and use of Asia-5a America-5b marketsc 56 non–bank-based financial instruments. All BIS-reporting banks A second argument is based on the End 1982 19.1 58.1 101.1 June 1997 18.8 14.2 50.0 absence of an effective national lender of German banks 17.0 13.7 last resort (Mishkin 1998), whether Japanese banks 39.2 5.2 U.S. banks 6.8 14.5 because the country has chosen a currency a. Indonesia, Republic of Korea, Malaysia, Philippines, and board or because of the intrinsic difficulty Thailand. in a small, highly open economy of an b. Argentina, Brazil, Chile, Mexico, and Venezuela. c. Major emerging markets: East Asia-5, Latin America-5, internal resolution of the liquidity problems China, Colombia, Czech Republic, Hungary. Source: Bank for International Settlements and Organisation 57 of the domestic financial system. for Economic Co-Operation and Development

153 GLOBAL ECONOMIC PROSPECTS

liquidity need. This would have potentially elusive. It has been argued that the Mexico important benefits in helping to avert crises bailout may have contributed to excessive by reducing perceptions of uncertainty risk taking in Asia, but the very large gener- about international support and securing alized decline in spreads on lending during country policy improvements. There are the period preceding the East Asian crisis also potential caveats to the effectiveness of across all emerging markets may have also this proposal, the problems being the same owed significantly to a generalized climate as those that apply to a lender of last resort. of euphoria. Still, it would be hard con- clude that moral hazard has not been play- Bailouts, moral hazard, and risk ing a significant role in influencing investor and burden sharing and borrower behavior in recent times, Whether through a formal lender of last especially in the case of Russia before the resort or ad hoc rescue packages, bailouts immediate runup to the crisis (when create moral hazard. Three types are possi- spreads were still moderate). The abrupt ble: the first type relates to expectations by cut-off in capital flows and sharply higher spreads to all emerging markets as a risk- class following Russia’s collapse may also Whether through a formal lender of be partly ascribed to the realization that last resort or ad hoc rescue packages, bailouts were no longer certain. bailouts create moral hazard. A supervisory role would be required for an international lender of last resort to developing country governments of a minimize moral hazard. This implies using bailout, which can reduce incentives to conditionalities for prudent macroeco- implement better policies. In most circum- nomic management, implementing institu- stances, however, the economic, social, and tional reforms to reduce risks of crisis, and political costs of a financial crisis are too supporting measures that reduce incentives high for such moral hazard to operate. In for (and introduce restrictions on) excessive fact, governments may delay calling on risk taking.58 Imposing this supervisory role international financial institutions— on sovereign governments poses many chal- despite the fact that a prompt response lenges, however (Obstfeld 1998). would reduce the costs of a crisis. A second Bailouts also require dealing with risk type of moral hazard can arise because and burden sharing issues, which means international creditors expect to be pro- adopting clear rules to make sure that pri- tected if a crisis occurs. A third type can vate operators bear some of the costs of their arise because banks and private corpora- risky behavior. For domestic debtors, guar- tions undertaking risky activities expect to antees may be justified only for commercial be bailed out under workouts of foreign banks in order to protect the payments sys- debts, leading to the domestic socialization tem. These guarantees have to be paired of these debts. with significant debt-reduction concessions Hard evidence about the extent of by private creditors (Goldstein 1998), which moral hazard in international lending is must bear some of the costs of a crisis and

154 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

not be the only ones bailed out through the to cope with the country’s financial crisis in intervention of the lender of last resort.59 1997. The size of rescue packages has Another alternative is to promote increased dramatically in recent years. debtor-creditor negotiations to reach restruc- Large amounts are thought to be necessary turing agreements allowing rollovers, exten- to quiet down markets as they panic. But sion of maturities, and reduction of debt. If such “rescue creep” has risks. No reason- clear and predictable, such workouts can able amount of public money can stop a help reduce lending distortions and induce justified speculative attack. By themselves, better pricing of risk.61 The main implemen- larger packages worsen moral hazard prob- tation issue is collective action by creditors. lems and may lead to excessively tough Every creditor has an incentive to try to get conditions, defeating the end objectives. out first or to “free-ride” on others’ accep- In the final analysis, recourse to a lender of last resort depends on resolving a series of issues: the political concerns asso- By themselves, larger packages worsen ciated with the need to supervise sovereign moral hazard problems and may lead to governments, the tradeoff between the excessively tough conditions, defeating short-run benefits of avoidance or reduc- the end objectives. tion in the severity of crisis and the long- run risks from moral hazard, and the avail- tance of workout arrangements. Negotia- ability of alternatives to official new tions are difficult to initiate, protracted, and lending. In the present international archi- hard to enforce because of information tecture, the mandate and corresponding asymmetries and transactions costs (Eichen- resources to play this role are lacking. green and Portes 1995). The collective Given such limits, better national risk man- action problem is much more challenging in agement in private and public spheres will a crisis that involves mostly private-to-pri- remain a key. vate debt (as in East Asia) than in one involving public debt (as in the crisis of the Complements to new official 1980s).62 Further complicating the process lending, and involvement of the is the much greater number of creditors and private sector in crisis debtors than in the past and the centrality of prevention and resolution exchange risk, as recent debt workouts in A first cushion against a reversal in capital Korea and Indonesia show. flows is adequate international reserves, a Three types of contract clauses in debt common but costly policy. Another possi- instruments can be used to improve credi- bility is to enter into private market tor coordination: collective representation, arrangements that guarantee liquidity up to such as in bondholder councils; qualified a predetermined limit. Argentina has such a majority voting; and sharing clauses that contingent repo facility with international discourage dissident creditors from engag- banks.60 Indonesia had standby credit ing in disruptive legal proceedings (Eichen- options, but the amounts were far too small green and Portes 1995, Goldstein 1998).63

155 GLOBAL ECONOMIC PROSPECTS

Debtor and creditor country govern- credibility and thereby greatly reduce the ments are the central players in orderly benefits. This seems to have been the case in workouts of private debts, as well as sover- East Asia in 1997. A second consideration is eign debt (Aggarwal 1998). The debtor the need to prevent debtor governments with country usually has primary responsibility weak reputations from making excessive use for setting negotiations, especially given the of standstills; debt standstills should be pos- adjustment policies they will have to imple- sible only under exceptional circumstances ment. This should not lead to provision of and in extreme distress. The third is the need government guarantees and the socializa- to get a standstill in place at the earliest pos- tion of private debts, however, which hap- sible date, so that all (or at least most) credi- pens frequently (including in the recent tors share in the costs of restructuring. Korean and Indonesian agreements) and exacerbates moral hazard. Creditor govern- Improved regulation and ments play a crucial role by forcing their stepped-up supervision on bank financial institutions to the negotiating lending in creditor countries table, to see that the private sector bears Asymmetric informational problems are some of the costs of risk taking. Interna- more acute in cross-border lending and can tional financial institutions, which have lead to less discriminating and more risky restrictions on lending into arrears, need to lending. This was the case in the runup to formalize the conditions for exceptions the debt crisis in the 1980s, and seems to (which are frequent in practice) if they are have occurred in the East Asia crisis. Wit- to negotiate with private creditors in pro- ness the dramatic drop in spreads for viding additional liquidity. Korean and Thai private borrowing, with The external debt workout must also spreads between bank and corporate bor- be properly sequenced with domestic debt rowers nearly equalized by 1996–97 (fig- restructuring. External creditors should not ures 3-15 and 3-16). receive undue precedence or seniority. Improved prudential regulation and The most critical aspect of a debt stepped-up supervision of internationally workout, however, is the temporary sus- active banks in creditor countries can help pension of debt payments, which helps stop reduce these risks.64 One proposal is to the decline in the currency and buys time to require a higher risk weight (than the 20 per- put in place a credible adjustment program cent under the Basle rule) for lending to and to organize debtor-creditor negotia- emerging markets,65 based on the assessment tions. By allowing an orderly debt restruc- of the country’s financial system. This would turing, it could result in better outcomes for raise the cost of borrowing to many develop- both the debtor country and creditors. ing countries,66 but by improving the pricing To be effective, however, the standstill of risk, it should reduce the incidence of crisis has to come at the right time. That timing and the volatility of lending and interest rate has to take into account three factors: one is spreads, and increase incentives for reform. that governments may delay declaring a debt Information about and assessment of standstill, fearing a loss of confidence and national financial systems, including the

156 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

quality and effectiveness of supervision and Orange County bankruptcies in the United implementation of domestic regulations or States, and financial crises in a number of global standards, may be valuable. Market Scandinavian countries) demonstrate. Still, participants may make use of it, and regu- there are potential benefits to better infor- lators in creditor countries may require its mation and disclosure and there are two use in risk management by lending banks different sets of issues under discussion: under their supervision. improvements in information and disclo- sure standards, and better use of informa- Information and monitoring of tion to assess national vulnerabilities and vulnerabilities undertake measures to forestall crises. More good information is always better Transparency and accountability. As in than less. At the same time, complete trans- the Mexican crisis in 1994, the East Asia parency does not exist, and better informa- crisis highlighted weaknesses in the cover- tion (recognizing the limits of costs in com- age, frequency, and timeliness of informa- piling such information) will not tion available to assess vulnerabilities. necessarily prevent crises. Even with elabo- Before the onset of the Asian crisis, and for rate disclosure rules, information asymme- several weeks (if not months) after, the try remains, as recent crises in industrial amounts of foreign liabilities to which the countries (for example, Republic Bank and countries were exposed were not precisely

Spreads dip in Korea… …and in Thailand Figure 3-15 Spreads between corporate Figure 3-16 Spreads between corporate and bank borrowers in the Republic of and bank borrowers in Thailand, 1990Ð97 Korea, 1991Ð97

Basis points Basis points 110 130

100 120 Private corporations

110 90 Private corporations 100 80

90 70 80 60 70

50 60

Commercial banks and financial institutions 40 Commercial banks and financial institutions 50

30 40 1991 1992 1993 1994 1995 1996 1997 1990 1991 1992 1993 1994 1995 1996 1997

Source: Euromoney (loanware). Source: Euromoney (loanware).

157 GLOBAL ECONOMIC PROSPECTS

known. Uncertainty about short-term debt tion is toward greater release of informa- and foreign exchange reserves exacerbated tion to the public, except in clearly defined the financial panic and crisis. The East cases where confidentiality requirements Asian crisis has illustrated the role of pri- override the gains from making informa- vate position-taking by nonfinancial firms, tion public. which are difficult to monitor in a liberal- More and better information can also ized environment. Questions have also been be made available from creditor country raised about the disclosure of information sources and from the BIS. This information and transparency of international agencies can be used to improve risk assessment. themselves. Accordingly, improvements are There is also a case for better private efforts at collecting information, despite the failure of rating agencies to adequately Warning indicators are unlikely to assess risks in Asian countries during the predict crises, particularly their timing, runup to the crisis.68 The G-22 working but they can provide timely and better group on transparency and accountability information about impending problems has also recommended that modalities for so that policymakers can take preventive compiling and publishing data on interna- actions. tional exposure of investment banks, hedge funds, and other institutional investors be needed on information and disclosure at all examined. levels (private sector, national authorities, Warning indicators and manifestations and international financial institutions), but of vulnerability. Warning indicators are international standards need to be applied unlikely to predict crises, particularly their carefully and progressively over time, rec- timing, but they can provide timely and ognizing constraints and costs. better information about impending prob- Following the Group of Seven (G-7) lems so that policymakers can take preven- Halifax proposals, international financial tive actions. The literature has used two institutions, in conjunction with national approaches. The first, the signals approach, authorities, are working to improve the aims at determining characteristic and quality and timeliness of information on abnormal behavior of a set of variables— central bank reserves, short-term foreign leading indicators—preceding crises, rela- currency debt (including central bank tive to tranquil periods. The indicators that derivatives transactions), and domestic predict the most actual crises and produce financial sector indicators (such as nonper- the least false alarms are used as leading forming loans and short-term debt).67 For indicators. The second is the regression national authorities, better disclosure and approach, which looks at the statistical sig- accounting standards, especially about for- nificance of various indicators in models of eign exchange liquidity positions, with crisis determination.69 Warning indicators respect to their financial institutions and of vulnerability usually flash positive sig- private corporations, are also important. nals for many variables at the same time. For international institutions, the presump- There is, however, no uniform, well-defined

158 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

set of indicators. Country conditions tend usually reflect rising capital inflows. to be crucial in determining the significance Deficits should also be carefully monitored of specific indicators (IMF 1998b; Gold- if spending is going to consumption rather stein and Reinhart 1998). Almost all stud- than investment—particularly in the trad- ies have found that traditional indicators of ables sector—since there is presumption of vulnerability—notably those relating to lower risks (because of faster growth of indebtedness, fiscal policy, sovereign risk GDP and exports). The East Asian crisis ratings, and interest spreads—have failed to has shown that the allocation and efficiency send useful warning indicators of currency of the increased investment is also relevant. crises. Warning indicators of banking crises. Vulnerability indicators of currency Most indicators of banking crises are crises. The most important signals of a cur- macroeconomic, and closely related to rency crisis are real exchange rate apprecia- those for currency crises (Goldstein and tion; international illiquidity, as measured Turner 1996; Demirgüç-Kunt and Detra- by the ratio of short-term liabilities to giache 1997; Kaminsky and Reinhart reserves, money stock to reserves, or for- 1997). Work is being done on developing eign assets to liabilities (of banks); and structural or microeconomic warning indi- lending booms financed by foreign borrow- cators. Relevant variables include spreads ing. Other significant indicators are slower between deposit and lending rates, access to GDP and export growth, higher foreign interbank loans, changes in the ratio of interest rates, deteriorating terms of trade, capital to risk-weighted assets, the loans-to- a decline in equity prices, and a banking deposits ratio, foreign currency exposure, crisis (Kaminsky and Reinhart 1997; Gold- government ownership, and the proportion stein and Reinhart 1998). of lending at the discretion of banks and The indicator most commonly associ- directed by government (Honohan 1997; ated with currency crises is the size of the Rojas-Suárez 1998). current account deficit.70 Empirical work has generally failed to find current account Notes deficits helpful by themselves in predicting 1. Greenwald, Stiglitz, and Weiss (1984); Green- crisis, however (Frankel and Rose 1996; wald and Stiglitz (1993); Mishkin (1991, 1997); Sachs and others 1996; Milesi-Ferretti and Stiglitz (1998b, 1998c). 2. Many empirical studies consider currency Razin 1996; Radelet and Sachs 1998). An crises to be episodes of large (Edwards exception is Goldstein and Reinhart 1989; Edwards and Montiel 1989; Frankel and Rose (1998), who find that ratios of current 1996). In contrast, Eichengreen, Rose, and Wyploz account deficit to GDP and to investment (1995) and Kaminsky and Reinhart (1997) favor a top the list of leading indicators of currency broader approach, focusing on devaluations as well as crises. In any case, current account deficits episodes of unsuccessful speculative attacks. Otker and Pazarbasioglu (1996) regard crises as including remain important in assessing vulnerability cases of devaluation, increases in the rate of crawl, and if complemented by analysis of the causal shifts to a more flexible exchange rate system. factors. Large or fast-increasing deficits 3. During 1870–1913 a number of then-emerging should always be monitored, since they economies also received large amounts of capital

159 GLOBAL ECONOMIC PROSPECTS

inflows. For 1870–89 and 1890-1913 the largest vol- 12. Private borrowers in Latin America in the umes (mean absolute value of current account as per- 1990s generally displayed a far greater willingness to centage of GDP) were respectively: 18.7 percent and hedge their foreign exchange liabilities, while borrow- 6.2 percent for Argentina, 8.2 percent and 4.1 percent ers in East Asian countries generally avoided them— for Australia, and 7.0 percent and 7.0 percent for partly because historical nominal exchange rate Canada (Obstfeld 1998). volatility (and the volatility of financial prices) was 4. Stiglitz (1998a); Summers (1998). much higher in Latin America than in East Asia. 5. Kumar, Moorthy, and Perraudin (1998) find 13. The changes in the real exchange rate are par- that a decline in portfolio flows has a stronger impact ticularly welcome if they reflect price adjustments in on the probability of crisis than a decline in FDI. response to fundamental factors such as a permanent 6. Other factors include the greater degree of transfer of resources from increased capital inflows, inherent risks present in developing countries, due to shifts and gains in productivity following reforms, their narrower economic bases (smaller economies improved terms of trade, correction of earlier excessive specialized in fewer economic activities). depreciation, or increased levels of consumption to 7. Surges in capital inflows can occur either equilibrium levels consistent with higher permanent exogenously, because of events in the income (and the need to incur current account deficits). outside the control of policymakers of the economy in 14. The impact is minimized even in the absence question, or endogenously, because of changes in of sterilization simply because, as the currency appre- country policies and circumstances (Hernandez and ciates, the extent of the impact on domestic money is Rudolf 1995; Gavin et al. 1995; Montiel and Rein- reduced by that exact amount of appreciation. hart 1997). They also respond to the macroeconomic 15. Assuming that the prevailing conditions do policy mix of the capital importing country, as well as not justify a rise in long-term equilibrium exchange the capital market structure (Montiel and Reinhart rates. Measuring whether prevailing exchange rates 1997). are misaligned with fundamentals is, however, notori- 8. Fernandez-Arias and Montiel (1995) also find ously difficult. In particular, relative purchasing power that in half of a sample of 12 countries experiencing parity movements may not always provide the correct picture of misalignments from equilibrium exchange the largest inflows relative to the size of their rates, and there may be other, better measures (Broner, economies, reserve accumulation accounted for about Loayza, and Lopez 1998). 40 percent of the inflows. 16. Schadler et al. (1993) and Dasgupta and Das- 9. In Corsetti, Pesenti, and Roubini (1998) this gupta (1995) find such evidence. This may be due to inconsistency is between fixed exchange rate, govern- lack of credibility of low-inflation programs (Kaminsky ment bailout guarantees (and their implication for and Leiderman 1998), to a rise in credit demand, or to monetary policy), and foreign debt accumulation and increased riskiness of the financial sector. current account deficits (or capital mobility). 17. The forms of such tighter fiscal policy may 10. There is evidence showing a significant degree have differential effects. An adjustment that curbs of both openness and capital mobility in developing spending or raises taxes on nontradables would countries. It is based on interest parity tests: Edwards reduce domestic inflation and interest rates. Alterna- and Khan (1985); Khor and Rojas-Suárez (1991); tively, one that curbs spending or raises taxes on trad- Haque and Montiel (1991); Reisen and Yèches (1993); ables would improve the current account deficit and Robinson (1991); and Dasgupta and Dasgupta (1995); reduce borrowing from abroad. Cutting spending and correlation between savings and investment: Doo- would have a more direct and immediate effect on ley, Frenkel, and Mathieson (1987); Wong (1988). aggregate demand than raising taxes, because of lags. 11. The cost of sterilization may be significant: 18. Kaminsky and Reinhart (1997) find that from 0.5 to 2 percent of GDP per year in Chile and more than half of the 26 banking crises they studied Colombia in the 1990s (Williamson 1996), and 0.3 to were followed by a balance of payments crisis within 0.75 percent of GDP per year for Malaysia, Thailand, and Indonesia in 1990–96 (ADB and World Bank three years. Conversely, only about 1 in 10 of the 1998). balance of payments crises were followed by banking

160 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

crises within three years. Also, regression of the mea- bank cases. Poor bank management is a sure of banking crises against the balance of pay- factor in 20 cases. ments measure indicates that balance of payments 31. This consensus also constitutes the core of crises do not help predict banking crises. Sachs, Tor- the IMF (1998a) guidelines. nell, and Velasco (1996) find that banking sector 32. Published by the International Accounting Stan- fragility is a major determinant of currency crisis. dards Committee (IASC) and the International Federa- Milesi-Ferreti and Razin (1996) show that the bank- tion of Accountants (IFAC). The International Organiza- ing sector plays an important role in determining cur- tion of Securities Commissions (IOSCO) is also working on establishing universal principles for rent account sustainability. regulations, improving disclosure requirements, and 19. However, Eichengreen and Rose (1998) find developing standards for cross-border offerings. no evidence for a role of domestic financial fragility in 33. Some may also argue the usefulness of com- predicting banking crises. petition in setting standards as against harmonization. 20. Sundarajan and Balino (1991) provide evi- 34. More radical options would entail the aboli- dence of this effect in the case of the crises in the tion of deposit insurance, the adoption of narrow Southern Cone countries during the 1980s: Chile banking, or the adoption of free banking; see Caprio (1981–83), Argentina (1980–82), and Uruguay and Klingebiel (1996b). (1982–85). 35. Goldstein and Turner (1996); Sundarajan 21. In addition to any measures and regulations and Balino (1991); Kaminsky and Reinhart (1997); on foreign currency exposure and access to foreign Caprio and Klingebiel (1996b); Demirgüç-Kunt and borrowing by banks, discussed below. Detragiache (1997); Gavin and Hausmann (1996). 22. There is some disagreement as to the effec- 36. Eichengreen and Rose (1998) find a highly tiveness of higher reserve requirements as an instru- significant correlation between changes in industrial ment for restraining lending booms. 23. Honohan (1997); Caprio, Atiyas, and Han- country interest rates and banking crises in emerging son (1994). Such limits may be set at high levels that markets. Also, Kaminsky and Reinhart (1997) find would not normally be reached, but restrain occa- that foreign-domestic interest rates signaled crises in sional bursts of overexuberant and risky expansion all 20 cases for which data are available. (World Bank 1998b). 37. Banking , which at first sight 24. The contributions of weak corporate gover- should allow pooling and diversification of risks, does nance and transparency to the East Asian crisis are not necessarily do that: larger banks may still take on analyzed in ADB and World Bank (1998). excessive risks without an adequate management 25. World Bank (1998b). Also, work on develop- structure in place. Bank supervision in the past con- ing standards for corporate governance is being under- ventionally focused on balance sheets, but much more taken within the OECD. attention is now devoted to the soundness of banks’ 26. For extensive discussion see World Bank management processes in assessing and managing risks (1997b) and ADB and World Bank (1998). (Mishkin 1996). 27. See also the G-22 working group report on 38. The ability of foreign financial institutions to strengthening financial systems, October 1998. enter domestic markets, which may be part of external 28. An example is the more recently developed financial liberalization but not formally part of capital value-at-risk techniques for risk management. account liberalization, also contributes to financial 29. These factors determine the balance of bene- integration and provides benefits similar to those of fits (in terms of efficiency and stability) according to trade liberalization, in terms of competitive effects and the type of banking system structure, ranging from improved quality of services and reduced prices. “narrow banking” to “universal banking” (Kaufman Claessens, Demirgüç-Kunt, and Huizinga (1997) pro- and Kroszner 1997). vide empirical evidence that broader foreign owner- 30. Caprio and Klingebiel (1996b) find faulty ship of banks renders domestic banking markets more supervision and regulation to be significant in 26 of 29 competitive and reduces domestic bank costs. Also,

161 GLOBAL ECONOMIC PROSPECTS

foreign banks that are internationally and in terms of fraction of lifetime consumption (Obstfeld 1995; van their activities more diversified help strengthen the Wincoop 1994). Typically, models that allow income domestic financial system. The reverse implication is growth to endogenously depend on diversification appear that the franchise value of domestic banks may fall, to arrive at higher estimates of gains than models where inducing more risk taking. While this negative impact income is fixed. The alternative sets of assumptions also is real and the ensuing risk should be managed, on bal- have differing degrees of ability to account for the stylized ance the benefits and risks of foreign entry are the facts about consumption volatility. same as those associated with FDI (in the financial sec- 45. This ignores longer-term consumption- tor) and warrant similar treatment. smoothing effects, which are important. 39. World Bank (1997b) discusses these benefits 46. Sachs and others (1996) find that countries at length. Another benefit sometimes cited is that with large short-term, variable, interest and foreign financial openness submits governments to the hard currency–denominated debt are more prone to crisis. scrutiny of international markets and would restrain Radelet and Sachs (1998) find that the ratio of short- any tendencies for mismanagement. Also, financial term debt to reserves is strongly associated with the deepening through increased capital flows helps onset of crisis, whereas the ratio of long-term debt to develop capital markets and allows more banking sys- reserves is not. Frankel and Rose (1996) find that the tem intermediation, which are shown to affect growth lower the reliance on FDI flows (compared to total positively (Levine and Zervos 1998). Increased inter- debt), or the greater the reliance on more volatile capi- national competition also enhances the quality of the tal flows, the higher the probability of crisis. financial system. 47. Models of self-fulfilling expectations of cur- 40. Of course the associated benefits of some rency crisis imply that the intrinsic instability of the flows such as trade credit which are closely related international financial system is a major contributor to to trade should not be assessed only within this currency crisis and, therefore, complete openness of framework. the capital account implies greater risks for developing 41. In capital inflow surges the increase in total countries. This issue is discussed below. inflows is due mainly to non-FDI flows, and the con- 48. The evidence is discussed in World Bank tribution is larger when the size of capital inflows is (1997b), chapter 2. See also Montiel and Reinhart (1997). large (greater than 9 percent of GDP). As seen in box 49. For instance, McKinnon and Pill (1997) 3.3, reserve accumulation is also larger. model how excessive foreign borrowing can take place 42. Because these flows have to be serviced, the in a recently liberalized domestic financial system with net gains are a fraction of the gross. FDI flows too inadequate supervision, and the presence of moral have to be serviced through profit repatriation, but a hazard, possibly due to government guaranties, in the significant part of such profits are reinvested, consis- context of unrestricted access to external finance. tent with the long-run nature of such inflows. 50. Some argue, however, that such distinctions 43. Low-income developing countries may also are not operational, that is, that volatility of flows benefit from long-term consumption smoothing. They cannot be distinguished among capital account cate- may borrow and increase their consumption now in gories, due to a high degree of substitution among view of increased income in the future. these categories. Claessens and others (1995) 44. The precise welfare improvement associated researched capital inflows to five developing and five with increased consumption smoothing depends on a industrial countries over a 15-year period (or longer) number of factors, such as the time-preference and the and found no evidence of patterns in the volatility shape of the utility function, as well as assumptions about among components of the capital account. Specifically, market structure, country size, and technology. The esti- long-term flows were as likely to be volatile as short- mates of utility benefits from consumption smoothing term flows. Similar research was later conducted by vary widely, from nearly 0 percent of lifetime consump- Chuhan and others (1996), who also found that vari- tion (Backus, Kehoe, and Kydland 1992; Cole and Obst- ous types of capital flows behave similarly. However, feld 1991; Tesar 1995) to a very significant (15 percent) they rejected the notion that flows are essentially the

162 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

same. They focused on interrelationships of the behav- 56. There may be increased bank exposure, how- ior of flows, that is, on the relative responsiveness of ever, to the extent counterparty risks have increased in one flow to changes in another. They determined that recent years with the proliferation of hedge funds and the composition of capital flows does matter, specifi- investing on margins. cally that short-term inflows are more responsive to a 57. An expansionary monetary policy or lender of change in FDI than the reverse and, therefore, suffer last resort activity to contain financial crisis and provide much more from contagion effects. Some have also liquidity is often counterproductive. Such a policy argued that multinational corporations, for instance, would cause expected inflation to rise and the domestic hedge long-term FDI by rolling over opposite short- currency to depreciate. The depreciation of the currency term currency positions, but there is little empirical would aggravate the domestic financial crisis, since it evidence to support that view. leads to a deterioration in the balance sheets of domestic 51. This may be partly due to lags in the mea- banks and firms that have debt denominated in foreign surement of FDI, with disbursement for new invest- currency. It may also lead to a jump in expected infla- ments spread over many years. tion, which would cause interest rates to rise, worsening 52. A more difficult and controversial measure is the balance sheets of firms and households and poten- to set prudential ratios for firms borrowing abroad— tially causing greater losses to banking institutions. The such as a minimum equity to liability ratio, maximum total net result is a worsening of the situation. An inter- foreign to domestic liability ratio, and maximum open national lender of last resort would help overcome these foreign exchange position. problems and contain the domestic crisis. 53. The so-called Rey Report (May 1996), which 58. These include the various rules discussed recommends that financial systems in emerging mar- above: adequate disclosure requirements for banks, kets be strengthened, that collective action clauses be adequate capital standards, penalties and sharing in added to bond contracts to facilitate orderly work- the costs by managers and shareholders, careful moni- outs, and that international financial institutions con- toring of banks’ risk management procedures, prompt sider “lending into arrears” on sovereign debt owed to corrective action, and so on. private creditors. 59. The U.S. Shadow Financial Regulation Com- 54. The three working groups, established by the mittee (1998) has proposed a mandatory loss-sharing Finance Ministers and Central Bank Governors of 22 system imposing “haircuts” on foreign lenders who systemically significant economies included senior offi- withdraw or fail to roll over their claims before IMF cials from these countries and international financial loans are paid back. institutions, focused on: increasing transparency and 60. The facility allows the Central Bank of disclosure; strengthening financial systems; and Argentina the of issuing short-term dollar- improving the management of international financial denominated government bonds and provincial loans crises. Three reports of the working groups were pub- totaling $6.7 billion to international banks subject to a lished in October 1998. Also the G-7 summit (Birm- buyback clause. It pays a fee as long as the facility is ingham, May 1998) has considered ways to strengthen available, and a spread is determined if it is used. The the global financial architecture. mechanism allows the Central Bank to act as a lender 55. Another issue concerns regional arrangements. of last resort without resorting to domestic money cre- Contagion tends to have, at least initially, a strongly ation, which is not possible under the currency board regional character as demonstrated in both the Mexican arrangement. and East Asian crises. These “neighborhood” spillover 61. International debt workouts are usefully effects may be due to underlying linkages or regional complemented by strong domestic bankruptcy laws similarities as perceived by investors. This provides an and systems of debtor-creditor workouts. argument for institutional arrangements of a regional 62. The case of sovereign debt is discussed in the character to improve monitoring and surveillance and Rey Report of the G-10. help initiate and implement policies to prevent financial 63. In order to avoid the adverse selection effects crises. Regionally coordinated (and pooled) intervention of such contract clauses, industrial countries should may also be useful in responding to crisis. include them in their own government bond contracts.

163 GLOBAL ECONOMIC PROSPECTS

64. There are also questions about the effects of Aggarwal, Vinod. 1998. “Exorcising Asian Debt: international hedge funds on the volatility of exchange Lessons from Latin American Rollovers, Workouts, rates and stock markets in small countries. and Writedowns.” Paper prepared for a World Bank 65. This proposal was made recently by Alan conference on The Aftermath of the Asian Crisis. Greenspan, Chairman of the U.S. Federal Reserve Washington, DC: World Bank. (May 13–14). Board, in a speech before the 34th Annual Conference Backus, David K., Patrick J. Kehoe, and Finn E. Kyd- on Bank Structure and Competition of the Federal land. 1992. “International Real Business Cycles.” Reserve Bank of Chicago. Journal of Political Economy 100(4):745–75. 66. The implicit tax is paid by the borrowing (August). country. This is unlike the imposition of taxes on capi- BIS (Bank for International Settlements). The Matu- tal flows by the borrowing country, where the tax rev- rity, Sectoral, and Nationality Distribution of enues accrue to the government of the borrower. International Bank Lending. Basle: Bank for Inter- 67. To this end, the IMF has established the Spe- national Settlements, Monetary and Economic cial Data Dissemination Standards (SDDS) and the Department (various issues). General Data Dissemination System (GDDS). See also Bordo, Michael D., Bruce Mizrach, and Anna J. the recommendations of the G-22 report of the work- Schwartz. 1995. “Real Versus Pseudo-international ing group on transparency and accountability, Octo- Systemic Risk: Some Lessons from History.” NBER ber 1998. Working Paper 5371. Cambridge, MA: National 68. It has been argued that rating agencies suffer Bureau of Economic Research. (December). from a conflict of interest, having to respect local sen- Broner, Fernando, Norman Loayza, and J. Humberto sitivities to gain and maintain a foothold in emerging Lopez. 1998. “Real Exchange Rate Misalignment markets. It is very unlikely that a rating agency will in Latin America.” Washington, DC: World Bank survive if it brazenly misleads its customers. (mimeo). 69. A survey is in IMF (1998b). Burki, Shahid, and Sebastian Edwards. 1995. “Latin 70. A rule of thumb is that a current account America After Mexico: Quickening the Pace.” deficit greater than 5 percent is often an indicator of Washington, DC: World Bank (mimeo). vulnerability—for sustainability, the growth of that Calvo, Guillermo A. 1996. “Capital Flows and debt which should not exceed the average rate of eco- Macroeconomic Management: Tequila Lessons.” nomic growth. International Journal of Finance and . 1:207–23. Calvo, Guillermo A., Morris Goldstein, and Eduard References Hochreiter, eds. 1996. Private Capital Flows to Alexander, William E., Jeffrey M. Davis, Liam P. Emerging Markets After the Mexican Crisis. Wash- Ebrill, and Carl-Johan Lindgren, eds. 1997. Sys- ington, DC: Institute for International Economics temic Bank Restructuring and Macroeconomic and Austrian National Bank. Policy. Washington, DC: International Monetary Calvo, Sara, and Carmen Reinhart. 1996. “Capital Fund. Inflows to Latin America: Is There Evidence of ADB (Asian Development Bank) and World Bank. Contagion Effects?” In Guillermo A. Calvo, Mor- 1998. “Managing Global Financial Integration: ris Goldstein, and Eduard Hochreiter, eds., Private Emerging Lessons and Prospective Challenges.” Capital Flows to Emerging Markets After the Mex- Manila (mimeo). ican Crisis. Washington, DC: Institute for Interna- Alesina, Alberto, Vittorio Grilli, and Gian Maria tional Economics. Milesi-Ferretti. 1994. “The Political Economy of Caprio, Gerard, Jr. 1997. “Safe and Sound Banking in Capital Controls.” In Leonardo Leiderman and Developing Countries—We’re Not in Kansas Any- Assaf Razin, eds., Capital mobility: the Impact on more.” Policy Research Working Paper 1739. Consumption and Growth. Cambridge, UK/New Washington, DC: World Bank, Policy Research York: Cambridge University Press. Department.

164 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Caprio, Gerard, Jr., and Daniela Klingebiel. 1996a. Corsetti, Giancarlo, Paolo Pesenti, and Nouriel “Bank : Cross Country Experience.” Roubini. 1998. “Paper Tigers? A Preliminary Policy Research Working Papers 1620. Washing- Assessment of the Asian Crisis.” Paper presented at ton, DC: World Bank, Policy Research Depart- the NBER International Seminar on Macroeco- ment, Finance and Private Sector Development nomics, Lisbon. (June 14–15). Division. Dasgupta, Dipak, and Bejoy Dasgupta. 1995. “Inter- ____. 1996b. “Bank Insolvency: Bad Luck, Bad Policy est Rates in Open Economies: Real Interest Parity, or Bad Banking?” Annual World Bank Conference Exchange Rates and Country Risk in Industrial on Development 1996. Washington, DC: World and Developing Countries.” Policy Research Bank. Working Paper 1283. Washington, DC: World Caprio, Gerard, Jr., Izak Atiyas, and James A. Hanson, Bank. eds. 1994. Financial Reform: Theory and Experi- Demirgüç-Kunt, Asli, and Enrica Detragiache. 1997. ence. New York, NY: Cambridge University Press. “The Determinants of Banking Crises: Evidence Carrasquilla, Alberto. 1998. “On Regulating Capital from Industrial and Developing Countries.” World Account Transactions.” Inter-American Develop- Bank Policy Research Working Paper 1828. Wash- ment Bank. Paper presented at a World Bank semi- ington, DC: World Bank Development Research nar. Washington, DC: World Bank, (June 23). Group and International Monetary Fund Research Cardenas, Mauricio, and Felipe Barrera. 1997. “On Department, World Bank and International Mone- the Effectiveness of Capital Controls in Colombia tary Fund. (September). during the 1990s.” Journal of Development Eco- ____. 1998. “Financial Liberalization and Financial nomics 54:27–57. Fragility.” Paper presented at the Annual Bank Cashin, Paul, Manmohan S. Kumar, and C. John Conference on Development Economics. Washing- McDermott. 1995. “International Integration of ton, DC: World Bank. (April 20–21). Equity Markets and Contagion Effects.” Working Dollar, David. 1998. “ and Disappoint- Paper WP/95/110. Washington, DC: International ments: Learning from the Success and Failure of Monetary Fund. Foreign Aid.” Policy Research Working Paper. Chuhan, Punam, Gabriel Perz-Quiros, and Helen Pop- Washington, DC: World Bank. per. 1996. “International Capital Flows: Do Short- Dooley, Michael, Jeffrey Frankel, and Donald Math- Term Investment and Direct Investment Differ?” Pol- ieson. 1987. “International Capital Mobility: icy Research Paper 1669. Washington, DC: World What Do Saving-Investment Correlations Tell Us?” Bank, Policy Research Department. (October). IMF Staff Papers 34:503–30. Washington, DC: Claessens, Stijn, Asli Demirgüç-Kunt, and Harry International Monetary Fund. (September). Huizinga. 1997. “How Does Foreign Entry Affect Edwards, Sebastian. 1989. Real Exchange Rates, the Domestic Banking Market?” Washington, DC: Devaluation, and Adjustment. Cambridge, MA: World Bank, (June). MIT Press. Claessens, Stijn, Michael Dooley, and Andrew Warner. Edwards, Sebastian, and Mohsin S. Khan. 1985. 1995. “Portfolio Capital Flows: Hot or Cold?” “Interest Rate Determination in Developing Coun- World Bank Economic Review 9(1):153–74. tries: a Conceptual Framework.” IMF Staff Paper Cole, Harold L., and Maurice Obstfeld. 1991. “Com- 32:377–403. Washington, DC: International Mon- modity Trade and International Risk-Sharing: etary Fund. How Much Do Financial Markets Matter?” Jour- Edwards, Sebastian, and Peter J. Montiel. 1989. nal of Monetary Economics 28:3–24. (August). “Devaluation Crises and the Macroeconomic Con- Corbo, Vittorio, and Leonardo Hernandez. 1996. sequences of Postponed Adjustment in Developing “Macroeconomic Adjustment to Capital Inflows: Countries.” IMF Staff Paper 36:875–903. Wash- Lessons From Recent Latin American and East ington, DC: International Monetary Fund. Asian Experience.” The World Bank Research Eichengreen, Barry, and Richard Portes. 1995. “Cri- Observer 11 (1):61–85. sis, What Crisis? Orderly Workouts for Sovereign

165 GLOBAL ECONOMIC PROSPECTS

Debtors.” London: Centre for 1998. Economic Development Institute (World Research. Bank) and University of Maryland. Eichengreen, Barry, and Andrew K. Rose. 1998. “Con- Gavin, Michael, and Ricardo Hausmann. 1996. “The tagious Currency Crises: Channels of Con- Roots of Banking Crises: The Macroeconomic veyance.” In Takatoshi Ito and Anne O. Krueger, Context.” Working Paper 318. Washington, DC: eds., Changes in Exchange Rates in Rapidly Devel- Inter-American Development Bank. Also appeared oping Countries (forthcoming). in Ricardo Haussman and Liliana Rojas-Suárez, Eichengreen, Barry, and Charles Wyplosz. 1993. “The eds., Banking Crises in Latin America. Washing- Unstable EMS.” Brookings Papers on Economic ton, DC: Inter-American Development Bank Activity 1(1993):51–124. Washington, DC: Brook- Gavin, Michael, Ricardo Hausmann, and Leonardo ings Institution. Leiderman. 1995. “The Macroeconomics of Capi- Eichengreen, Barry, Andrew K. Rose, and Charles tal Flows to Latin America: Experience and Policy Wyplosz. 1995. “Exchange Market Mayhem: the Issues.” Working Paper 310. Washington, DC: Antecedents and Aftermath of Speculative Attacks.” Inter-American Development Bank. (October). Economic Policy: A European Forum 21:249–312. Glick, Reuven, and Andrew Rose. 1998. “Contagion London: Centre for Economic Policy Research. and Trade: Why are Currency Crises Regional?” (October). Prepared for a Centre for Economic Policy ____. 1996. “Contagious Currency Crises.” Scandi- Research/World Bank Conference Financial Crises: navian Economic Review 98 (4):463–84. Also Contagion and Market Volatility, London. (May). reprinted in Torben M. Andersen and Karl O. Goldstein, Morris. 1995. “Coping with Too Much of Moene, eds., Financial Liberalization and Macro- A Good Thing: Policy Responses for Large Capital economic Stability. Malden, MA: Blackwell. Inflows in Developing Countries.” Policy Research Fernandez-Arias, Eduardo, and Peter J. Montiel. Working Paper 1507. Washington, DC: World 1995. “The Surge in Capital Inflows to Developing Bank, Policy Research Department. Countries: Prospects and Policy Response.” Policy ____. 1998. “Limiting the Moral Hazard in Interna- Research Working Paper 1473. Washington, DC: tional Financial Rescues.” Prepared for a Brook- World Bank, International Economics Department, ings symposium. Washington, DC: Brookings Insti- International Finance Division. tution. (June). Flood, Robert, and Peter Garber. 1984. “Collapsing Goldstein, Morris, and Carmen Reinhart. 1998. “Fore- Exchange-rate Regimes: Some Linear Examples.” casting Financial Crises: Early Warning Signals for Journal of International Economics 17:1–13. Emerging Markets.” Washington, DC: Institute for Frankel, Jeffrey, and Andrew K. Rose. 1996. “Cur- International Economics (forthcoming). rency Crashes in Emerging Markets: An Empirical Goldstein, Morris, and Philip Turner. 1996. “Banking Treatment.” Journal of International Economics Crises in Emerging Economies: Origins and Policy 41:351–66. Options.” BIS Economic Papers No. 46. Basle: Frankel, Jeffrey, and Sergio Schmukler. 1996. “Crisis, Bank for International Settlements. (October). Contagion, and Country Funds: Effects on East Greenwald, Bruce, and Joseph E. Stiglitz. 1993. Asia and Latin America.” Working Paper Number “Financial Market Imperfections and Business PB96-04. San Francisco: Federal Reserve Bank of Cycles.” Quarterly Journal of Economics 108 (1): San Francisco. 77–114. Galbis, Vicente. 1993. “High Real Interest Rates Greenwald, Bruce, Joseph E. Stiglitz, and Andrew Under Financial Liberalization: Is There a Prob- Weiss. 1984. “Informational Imperfections in the lem?” IMF Working Paper 93/7. Washington, DC: Capital Markets and Macroeconomic Fluctua- International Monetary Fund. (January). tions.” American Economic Review 74 (22): Gavin, Michael. 1998. “Fiscal Policy in Emerging 194–99. Market Economies.” Paper presented at the Capi- Group of Seven. 1998. G-7 Leaders’ Statement on the tal Flows Seminar at College Park, MD, May World Economy and Declaration of G-7 Finance

166 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Ministers and Central Bank Governors, October of Disinflation: Is it a Lack of Credibility?” Journal 30. of Development Economics 55:191–214. Haque, Nadeem U., and Peter J. Montiel. 1991. “Cap- Kaminsky, Graciela L., and Carmen Reinhart. 1997. ital Mobility in Developing Countries: Some “The Twin Crises: The Causes of Banking and Bal- Empirical Tests.” World Development 19:1391– ance-of-Payments Problems.” Working Papers in 98. (October). International Economics 37. College Park, MD: Hellman, Thomas, Kevin Murdock, and Joseph E. University of Maryland, Center for International Stiglitz. 1996. “Deposit Mobilization through Economics, Department of Economics. (December). Financial Restraint.” In N. Hermes and R. Kaufman, George G., and Randall S. Kroszner. 1997. Lensink, eds., Financial Development and Eco- “How Should Financial Institutions and Markets nomic Growth: Theory and Experiences from be Structured? Analysis and Options for Financial Developing Economies. London: Routledge. System Design.” Inter-American Development ____. 1997. “Liberalization, Moral Hazard in Bank- Bank, Working Paper Series 338. Washington, DC: ing, and Prudential Regulation: Are Capital Inter-American Development Bank. (February). Requirements Enough?” Research Paper No. Khatkhate, Deena. 1998. “Chasing Shadows: East 1466. Asian Financial Crisis and the IMF.” Lecture to Hernandez, Leonardo, and Heinz Rudolf. 1995. “Sus- Bank Negara Malaysia. tainability of Private Capital Flows to Developing Khor, Hoe Ee, and Liliana Rojas-Suárez. 1991. “Inter- Countries: Is a General Reversal Likely?” Policy est Rates in Mexico: the Role of Exchange Rate Research Working Paper 1518. Washington, DC: Expectations and International Creditworthiness.” World Bank, Policy Research Department. IMF Working Paper WP/91/12. Washington, DC: Honohan, Patrick. 1997. “ Banking System Failures in International Monetary Fund, Research and West- Developing and Transition Countries: Diagnosis ern Hemisphere Departments. and Prediction.” BIS Working Papers No. 39. King, Robert G., and Ross Levine. 1993. “Finance and Basle: Bank for International Settlements, Mone- Growth: Schumpeter Might be Right.” Quarterly tary and Economic Department. (January). Journal of Economics 108(3):717–37. International Monetary Fund. 1996. Exchange King, Mervyn, and Sushil Whadwhani. 1990. “Trans- Arrangements and Exchange Restrictions, Annual mission of Volatility between Stock Markets.” Report. Washington, DC. Review of Financial Studies 3 (1):5–33. ____. 1997a. International Capital Markets. Washing- Krugman, Paul. 1996. “Are Currency Crises Self-Ful- ton, DC. (November). filling?” NBER Macroeconomics Annual 1996: ____. 1997b. Exchange Arrangements and Exchange 345–506. Cambridge, MA/London: MIT Press. Restrictions, Annual Report. Washington, DC. ____. 1998. “What happened to Asia?” in Paul Krug- ____. 1998a. Toward a Framework for Financial Sta- man’s web page (http://web.mit.edu/people/krugman/ bility. Prepared by a Staff Team led by David Folk- index.html). erts-Landau and Carl-Johan Lindgren. World Eco- Kumar, Manmohan S., Uma Moorthy, and William Per- nomic and Financial Surveys series. Washington, raudin. 1998. “Determinants of Emerging Market DC: International Monetary Fund. (January). Currency Crises and Contagion Effects.” Paper pre- ____. 1998b. World Economic Outlook. Washington, sented at the Centre for Economic Policy Research. DC: International Monetary Fund. (May). Financial Crises: Contagion and Market Volatility: a ____. 1998c. Exchange Arrangements and Exchange CEPR/World Bank Conference. London. (May). Restrictions, Annual Report (forthcoming). Levine, Ross. 1997. “Financial Development and Eco- Jeanne, Olivier. 1997. “Are Currency Crises Self-Ful- nomic Growth: Views and Agenda.” Journal of filling: A Test.” Journal of International Econom- Economic Literature: 688–726. (June). ics 43(3/4):163–86. Levine, Ross, and Sara Zervos. 1998. “Stock Markets, Kaminsky, Graciela L., and Leonardo Leiderman. Banks, and Economic Growth.” The American 1998. “High Real Interest Rates in the Aftermath Economic Review. (June).

167 GLOBAL ECONOMIC PROSPECTS

Lindgren, Carl-Johan, Gillian Garcia, and Matthew I. Empirical Estimates.” Policy Research Working Saal. 1996. Bank Soundness and Macroeconomic Paper 1103. Washington, DC: World Bank, Policy Policy. Washington, DC: International Monetary Research Department. Fund. Montiel, Peter J., and Carmen M. Reinhart. 1997. Massad, Carlos. 1994. “Financial Reforms, Opening “The Dynamics of Capital Movements to Emerg- Up and Stability: Policy Issues and Sequencing ing Economies during the 1990s.” Paper prepared Considerations.” In Shakil Faruqi, Financial Sector for the UNU/WIDER project on Short-term Capi- Reforms, Economic Growth, and Stability. Wash- tal Movements and Balance of Payments Crises. ington, DC: World Bank. Mundell, Robert A. 1963. “Capital Mobility and Sta- Masson, Paul. 1998. “Contagion: Monsoonal Effects, bilization Policy Under Fixed and Flexible Spillovers, and Jumps Between Multiple Equilib- Exchange Rates.” Canadian Journal of Economics ria.” Prepared for a CEPR/World Bank Conference and Political Science: 475–85. (November). on Financial Crises: Contagion and Market Nicholl, Peter. 1997. “Market-Based Banking Regula- Volatility. London. (May). tion.” In Suman K. Bery and Valeriano F. Garcia, Mathieson, Donald J., and Liliana Rojas-Suárez. eds., Preventing Banking Sector Distress and Crises 1993. “Liberalization of the Capital Account: in Latin America. Proceedings of a Conference Experiences and Issues.” IMF Occasional Paper held in Washington, DC. April 15–16, 1996. 103. Washington, DC: International Monetary Obstfeld, Maurice. 1986. “Rational and Self-fulfilling Fund. (March). Balance of Payments Crises.” American Economic McKinnon, Ronald I., and Huw Pill. 1997. “Credible Review 76: 676–81. Economic Liberalizations and Overborrowing.” ____. 1995. “Risk-taking, Global Diversification and American Economic Association Papers and Pro- Growth.” American Economic Review 84(5): ceedings of 109th Annual Meeting. (January). 1310–29. Milesi-Ferretti, Gian Maria, and Assaf Razin. 1996. ____. 1996. “Models of Currency Crises with Self-ful- “Current-Account Sustainability.” Princeton Stud- filling Features.” European Economic Review: ies in International Finance 81. Princeton, NJ: 1037–47. (April). Princeton University, Department of Economics. ____. 1998. “The Global Capital Market: Benefactor (October). or Menace?” NBER Working Paper 6559. Cam- Mishkin, Frederic S. 1991. “Asymmetric Information bridge, MA: National Bureau of Economic and Financial Crises: A Historical Perspective.” In Research. R.Glenn Hubbard, ed., Financial Markets and Obstfeld, Maurice, and Kenneth Rogoff. 1996. Foun- Financial Crisis. Chicago: University of Chicago dations of International Macroeconomics. Cam- Press. bridge, MA: MIT Press ____. 1996. “Bank Consolidation: A Central Banker’s Obstfeld, Maurice, and Alan Taylor. 1998. “The Great Perspective.” NBER Working Paper 5849. Cam- Depression as a Watershed: International Capital bridge, MA: National Bureau of Economic Research. Mobility over the Long Run.” In Bordo, Michael ____. 1997. “Understanding Financial Crises: A D., Claudia D. Goldin, and Eugene N. White, eds., Developing Country Perspective.” Annual World The Defining Moment: The Great Depression and Bank Conference on Development Economics the American Economy in the Twentieth Century. 1996. Washington, DC: World Bank. Chicago: University of Chicago Press. ____. 1998. “Systemic Risk, Moral Hazard and the OECD. Financial Markets Trends. Paris: Organisation Lender of Last Resort.” Prepared for a World for Economic Co-operation and Development (var- Bank/Reinventing Bretton Woods Committee Con- ious issues). ference on the Aftermath of the Asian Financial Otker, Inci, and Ceyla Pazarbasioglu. 1996. “Specula- Crisis, Washington, DC. (March 13–14) tive Attacks and Currency Crises: The Mexican Montiel, Peter J. 1993. “Capital Mobility in Develop- Experience.” Open Economies Review 7 (Sup. 1): ing Countries: Some Measurement Issues and 535–52.

168 PREVENTING FINANCIAL CRISES IN DEVELOPING COUNTRIES

Quirk, Peter J., Owen Evans, and others. 1995. Capi- nomic Activity 1(1996):147–215. Washington, tal Account Convertibility: Review of Experience DC: Brookings Institution. and Implications for IMF Policies. IMF Occasional Schadler, Susan, Maria Carkovic, Adam Bennett, and Paper 131. Washington, DC: International Mone- Robert Kahn. 1993. “Recent Experiences with tary Fund. (October). Surges in Capital Inflows.” IMF Occasional Paper Radelet, Steven, and Jeffrey D. Sachs. 1998. “The East 103. Washington, DC: International Monetary Fund. Asian Financial Crisis: Diagnosis, Remedies and Sheng, Andrew. 1996. Bank Restructuring: Lessons Prospects.” Cambridge, MA: Harvard Institute for from the 1980s. Washington, DC: World Bank. International Development. (April 20). Stiglitz, Joseph E. 1998a. “Boats, Planes, and Capital Razin, Assaf, and Chi-Wa Yuen. 1994. “Convergence Flows.” Financial Times. March 21. in Growth Rates: a Quantitative Assessment of the ____. 1998b. “Knowledge for Development: Eco- Role of Capital Mobility and International Taxa- nomic Science, Economic Policy, and Economic tion.” In Leonardo Leiderman and Assaf Razin, Advice.” Annual Bank Conference on Develop- eds., Capital Mobility: the Impact on Consump- ment Economics. (April 20–21). tion, Investment and Growth. Cambridge, UK: ____. 1998c. “The Role of the Financial System in Cambridge University Press. Development.” Presentation at the Fourth Annual Reinhart, Carmen. 1998. “Capital Flows and Conta- Conference on Development Economics in Latin gion: Causes, Consequences, and Policy American and the Caribbean. El Salvador. (June). Responses.” Paper presented at the Capital Flows Summers, Lawrence. 1998. “Deputy Secretary Sum- Seminar at College Park, MD, May 1998. Eco- mers’ Remarks before the International Monetary nomic Development Institute (World Bank) and Fund.” (March 19). Available at http://www.treas. University of Maryland. gov/press/releases/pr2286.htm Reisen, Helmut, and Hélène Yèches. 1993. “Time- Sundarajan, Vasudevan, and Tomas J.T. Balino. 1991. Varying Estimates on the Openness of the Capital Banking Crises: Cases and Issues. Washington, Account in Korea and Taiwan,” Journal of Devel- DC: International Monetary Fund. opment Economics 41:285–305. Tesar, Linda L. 1995. “Evaluating the Gains from Robinson, David. 1991. “Thailand: Adjusting to Suc- International Risksharing.” Carnegie-Rochester cess-Current Policy Issues.” IMF Occasional Paper Conference Series on Public Policy 42(1995): 85. Washington, DC: International Monetary 95–143. : North-Holland. Fund. Valdès, Rodrigo O. 1996. “Emerging Market Conta- Rodrik, Dani. 1998. “Who Needs Capital-Account gion: Evidence and Theory” (mimeo). Massachu- Convertibility?” Symposium paper to appear in setts Institute of Technology. Princeton Essays in International Finance. Wacziarg, Romain. 1998. “Measuring the Dynamic Rojas-Suárez, Liliana. 1998. “Early Warning Indica- Gains from Trade.” Policy Research Working tors of Banking Crises: What Works for Emerging Paper 2001. Washington, DC. (November). Markets? With Applications to Latin America.” Williamson, John. 1996. The Crawling Band as an Washington, DC: Inter-American Development Exchange Rate Regime: Lessons from Chile, Bank. Colombia, and Israel. Washington, DC: Institute Rojas-Suárez, Liliana, and Steven R. Weisbrod. 1996. for International Economics. (October). “Banking Crises in Latin America: Experiences and van Wincoop, Eric. 1994. “Welfare Gains from Inter- Issues.” In Ricardo Haussmann and Liliana Rojas- national Risk-Sharing.” Journal of Monetary Eco- Suárez, eds., Banking Crises in Latin America. nomics 34:175–200. (October). Washington, DC: Inter-American Development Wolf, Holger C. 1997. “Regional Contagion Effects In Bank. Emerging Stock Markets.” Working Papers in Sachs, Jeffrey D., Aaron Tornell, and Adrés Velasco. International Economics G-97-03:1-18. Princeton, 1996. “Financial Crises in Emerging Markets: The NJ: Princeton University, Department of Econom- Lessons from 1995.” Brookings Papers on Eco- ics, International Finance Section. (October).

169 GLOBAL ECONOMIC PROSPECTS

Wong, David Y. 1988. “What Do Savings-Invest- ____. 1998a. Global Development Finance. Washing- ment Correlations Tell Us About Capital ton, DC. Mobility?” Working Paper 88-10. Philadelphia, ____. 1998b. World Development Report 1998/99: PA: Federal Reserve Bank of Philadelphia. Knowledge for Development. New York: Oxford (May). University Press. World Bank. 1997a. Global Economic Prospects ____. 1998c. World Development Indicators. Wash- and the Developing Countries. Washington, ington, DC. DC. Wyplosz, Charles. 1998. “International Capital Mar- ____. 1997b. Private Capital Flows to Developing ket Failures: Sources, Costs, and Solutions.” Back- Countries: The Road to Financial Integration. ground Paper for Global Economic Prospects and A World Bank Policy Research Report. New the Developing Countries 1998. Washington, DC: York: . World Bank.

170