Chapter

II Assessing Crisis Vulnerabilities in Latin America

Author(s):
G. Kincaid, and Charles Collyns
Published Date:
April 2003
Share
  • ShareShare
Show Summary Details
Author(s)
Javier Hamann, Kalpana Kochhar, Timothy Lane, Guy Meredith, Jürgen Odenius, David Ordoobadi, Hélène Poirson and David Robinson 

Financial crises generally result from a combination of balance sheet fragilities and an inability—perceived or real—to mount an effective policy response. The precise mix of these factors differs across crises. In some cases, balance sheet fragilities have been primarily in the financial sector, while in others public debt dynamics have been the main focus of concern. Policy paralysis likewise stems from a variety of economic constraints and political factors. This section examines how lessons from previous crises can be brought to bear in examining the potential vulnerabilities of Latin American countries. It focuses on key elements that have been important in previous crisis episodes but also points to possible gaps in conventional wisdom. In doing so, it stresses that crises tend to reflect vulnerabilities that market participants and policymakers had previously overlooked; thus, some of the success stories of today may be the crisis countries of tomorrow.

As Argentina remains in difficulty and other countries in Latin America experience recurrent pressure, a key question is how widely these troubles are likely to spread.1 Recent events suggest the possibility of a financial crisis that is regional in scope—similar to the 1997–98 Asian crisis. But the experience in Asia suggests that, while no country can be immune to adverse effects from economic cataclysms in neighboring countries, the nature and severity of these effects depend on individual countries’ vulnerabilities as well as on global financial conditions, including banks’ and investors’ appetite for risk.

This section examines how lessons from previous crises can be brought to bear in examining the potential vulnerabilities of Latin American countries and provides an overview of the main elements of vulnerability to crises. In this exercise, the object is not to provide a comprehensive assessment of the countries’ vulnerability, but to focus on key elements that have been important in previous crisis episodes, while also pointing to possible gaps in conventional wisdom.

Elements of Vulnerability to Crises

In general, countries that have succumbed to financial crises have suffered from combinations of different vulnerabilities: balance sheet fragilities and a perceived inability to mobilize an effective policy response. These factors and their interaction influenced both the onset of the crises and the way they unfolded.

A key factor underlying most emerging market financial crises in recent years was fragilities in the balance sheets of the government, financial institutions, corporations, or households. In some instances, these fragilities consisted of balance sheets that were susceptible to movements in interest rates, exchange rates, or both. In some cases, inadequate cushions of liquidity were another important factor. In other cases, important preexisting weaknesses in balance sheets were crucial; the full magnitude of these weaknesses often came to light only after the crisis broke.

At the same time, in most crises another central factor was the inability to mobilize an effective policy response. This policy paralysis took a variety of forms. In some instances, monetary policy failed to respond quickly and decisively to domestic and external shocks. Often it was constrained by an untenable exchange rate regime—an existing exchange rate peg that sooner or later became the focus of market skepticism. In many instances, important economic constraints on policy existed—reflecting the awareness that policy action to stem a crisis would have adverse effects on the financial system or debt dynamics, or on the real economy. A third factor—relevant to varying degree in virtually all crises—was the political process, which was made more complex by a fragile political balance or an electoral transition.

Contagion is the third key aspect examined. A crisis spreads across countries typically through a combination of channels and reflects differing vulnerabilities that determine how each country is affected.

The remainder of this section is structured as follows. The next two subsections examine in turn the role of balance sheet fragilities and policy paralysis. This is followed by an analysis of contagion. The last subsection presents concluding remarks, and an appendix offers more thoughts on assessing the scope for contagion across emerging market economies.

Balance Sheet Fragilities

A key element in most previous financial crises was the vulnerability of balance sheets in the public sector, banks, or corporations. In some instances (such as Mexico and Korea), there were inadequate cushions of liquidity in relation to short-term debt or foreign currency deposits. In other cases, the structure of public and/or private debt made these debts susceptible to interest rate and/or exchange rate movements (Brazil, Turkey). A particularly salient factor in many crises was the prevalence of un-hedged foreign currency borrowing by corporations (Indonesia, Korea), banks (Mexico), or government (Brazil, Russia). Maturity mismatches in the public or private sector (Thailand, Korea) made these balance sheets susceptible to the large changes in domestic interest rates that frequently characterize a crisis. Preexisting nonperforming loans (Korea, Thailand) were an important factor in some cases. Often the full magnitude of these problem loans did not become apparent until the crisis broke, and revelations further contributed to the downturn of market sentiment. In some cases (notably Russia), off-balance-sheet operations of the banking sector were also a source of vulnerability.

In many crises, a perceived lack of medium-term sustainability was critical. In some cases—Brazil, Russia, Turkey, Argentina—concerns were primarily over the public finances. In others—notably Thailand—it was private sector behavior that was most important in generating an unsustainable overall balance of payments. Table 2.1 highlights some elements of financial vulnerability in selected Asian countries just before the Asian crisis. Table 2.2 shows how this picture looked in Latin America at the end of 2001. In particular, the ratio of reserves to short-term debt was very low in Argentina, Brazil, and Uruguay, relative to the comparator sample.

Table 2.1.Selected Asian Countries: Key Vulnerability Indicators Based on Country (Credit Risk) Models(In percent unless otherwise noted)
IndicatorMalaysiaKoreaThailandPhilippinesIndonesiaSample median1
Year: 1996
Ratio of reserves to short-term debt2209.270.177.649.6115.1
Ratio of reserves (in months of imports)4.12.76.33.55.14.6
Ratio of reserves to broad money29.015.425.721.515.425.0
Ratio of total external debt to GDP339.322.359.248.556.839.3
Ratio of total external debt to GDP431.426.358.321.427.123.4
Ratio of total external debt to exports350.789.3193.3196.7258.8196.7
Ratio of total external debt to exports440.4105.4190.386.7123.4123.4
Ratio of current account to GDP−4.4−4.4−8.1−4.8−3.4−3.0
Terms of trade movement3.7−5.26.3−5.0−4.2−0.4
Ratio of central government balance to GDP0.70.10.90.31.2−1.4
Sources: IMF, International Financial Statistics (IFS), World Economic Outlook, and IMF staff estimates; World Bank, Global Development Finance (GDF); and Bank for International Settlements (BIS).

A sample of 41 countries was chosen as comparators for the Latin American countries. In light of the diversity within the Latin American group, the comparator group also included most important emerging markets as well as Hong Kong SAR, Hungary, Iceland, Ireland, Israel, India, Indonesia, Jordan, Korea, Lebanon, Malaysia, Mexico, New Zealand, Pakistan, Peru, Philippines, Poland, Portugal, Russia, Singapore, Slovak Republic, South Africa, Sweden, Thailand, Turkey, Uruguay, and Venezuela.

By remaining maturity.

From GDF (only available for developing countries).

From BIS (refers to loans from banks and debt securities issued abroad; since credits from official sources are not included, these data tend to under-state the indebtedness of developing countries especially).

Sources: IMF, International Financial Statistics (IFS), World Economic Outlook, and IMF staff estimates; World Bank, Global Development Finance (GDF); and Bank for International Settlements (BIS).

A sample of 41 countries was chosen as comparators for the Latin American countries. In light of the diversity within the Latin American group, the comparator group also included most important emerging markets as well as Hong Kong SAR, Hungary, Iceland, Ireland, Israel, India, Indonesia, Jordan, Korea, Lebanon, Malaysia, Mexico, New Zealand, Pakistan, Peru, Philippines, Poland, Portugal, Russia, Singapore, Slovak Republic, South Africa, Sweden, Thailand, Turkey, Uruguay, and Venezuela.

By remaining maturity.

From GDF (only available for developing countries).

From BIS (refers to loans from banks and debt securities issued abroad; since credits from official sources are not included, these data tend to under-state the indebtedness of developing countries especially).

Table 2.2.Selected Latin American Countries: Key Vulnerability Indicators Based on Country (Credit Risk) Models(In percent unless otherwise noted)
IndicatorArgentinaBrazilChileColombiaMexicoPeruUruguayVenezuelaSample median1
Year: 2001 (unless otherwise indicated)
Ratio of reserves to short-term debt241.760.9127.674.7125.6152.538.6116.4125.6
Ratio of reserves (in months of imports) (2000)12.06.79.89.32.311.48.69.75.4
Ratio of reserves to broad money19.923.946.144.932.449.7. . .40.428.4
Ratio of total external debt to GDP3 (2000)51.440.149.341.925.954.040.931.549.0
Ratio of total external debt to GDP4 (2000)38.819.333.422.617.211.032.013.832.0
Ratio of total external debt to exports3 (2000)553.5432.0203.6261.390.3406.4357.2120.1203.6
Ratio of total external debt to exports4 (2000)417.9208.1137.7140.760.182.9279.352.7102.5
Ratio of current account to GDP (2000)−3.2−4.1−1.30.4−3.1−3.1−3.010.8−0.9
Terms of trade movement−0.50.5−4.0−9.1−2.35.44.5−17.0−0.3
Ratio of central government balance to GDP (1999)−2.9. . .−1.4−5.9−1.5−2.2−3.7−2.4−2.2
Sources: IMF, International Financial Statistics (IFS), World Economic Outlook, and IMF staff estimates; World Bank, Global Development Finance (GDF); and Bank for International Settlements (BIS).

A sample of 41 countries was chosen as comparators for the Latin American countries. In light of the diversity within the Latin American group, the comparator group also included most important emerging markets as well as Hong Kong SAR, Hungary, Iceland, Ireland, Israel, India, Indonesia, Jordan, Korea, Lebanon, Malaysia, Mexico, New Zealand, Pakistan, Peru, Philippines, Poland, Portugal, Russia, Singapore, Slovak Republic, South Africa, Sweden, Thailand, Turkey, Uruguay, and Venezuela.

By remaining maturity.

From GDF (only available for developing countries).

From BIS (refers to loans from banks and debt securities issued abroad; since credits from official sources are not included, these data tend to under-state the indebtedness of developing countries especially).

Sources: IMF, International Financial Statistics (IFS), World Economic Outlook, and IMF staff estimates; World Bank, Global Development Finance (GDF); and Bank for International Settlements (BIS).

A sample of 41 countries was chosen as comparators for the Latin American countries. In light of the diversity within the Latin American group, the comparator group also included most important emerging markets as well as Hong Kong SAR, Hungary, Iceland, Ireland, Israel, India, Indonesia, Jordan, Korea, Lebanon, Malaysia, Mexico, New Zealand, Pakistan, Peru, Philippines, Poland, Portugal, Russia, Singapore, Slovak Republic, South Africa, Sweden, Thailand, Turkey, Uruguay, and Venezuela.

By remaining maturity.

From GDF (only available for developing countries).

From BIS (refers to loans from banks and debt securities issued abroad; since credits from official sources are not included, these data tend to under-state the indebtedness of developing countries especially).

Given the importance of these balance sheet weaknesses in determining overall vulnerability, a number of systematic approaches have been developed to examine these weaknesses and their relationship to the incidence of crises. In particular, balance sheet weaknesses are a prominent aspect of early warning and country risk models. These models and their implications for Latin America are briefly reviewed below.2

Early Warning Systems

Early warning system (EWS) models, as well as models of country (credit) risk and so-called crisis cost models, provide a formal framework for examining vulnerability to crisis. EWS models focus on ascertaining the probability of a currency crisis, whereas models of country (credit) risk capture the likelihood of an external debt or funding crisis. Crisis cost models predict the likely impact of a crisis on output and growth, if it were to occur.

The different models have consistently highlighted a few key indicators of vulnerability. EWS models emphasize several external indicators: competitiveness (size of the current account deficit and exchange rate overvaluation); debt burden (external debt as a ratio to GDP or to exports, and default history); growth prospects (export growth, GDP growth, and investment); liquidity (the ratio of short-term debt to reserves, as well as import cover and the ratio of reserves to broad money)3 and financing requirements (Box 2.1). Most EWS models also identify domestic credit conditions as a financial indicator of vulnerability. Some EWS models (such as the Credit Suisse First Boston [CSFB] regional Latin America model) and most spreads and ratings models also highlight the budget deficit as a critical fiscal indicator (Box 2.2). More recently, some EWS models have highlighted corporate sector weaknesses as relevant warning signals of crisis. The main vulnerability areas according to these models are financial leverage (debt-to-equity ratios),4 liquidity (short-term debt over working capital), and corporate governance (proxied by a shareholder-rights index).

Box 2.1.Standard Models of External Crises

In first-generation models, the underlying causes of crisis are fundamental macroeconomic problems, although the proximate triggers of the crisis may be contagion effects or imprudently low levels of foreign exchange reserves. Fundamental causes of crisis include traditional macroeconomic imbalances such as unsustainable current account deficits, overvalued exchange rates, unsustainable fiscal positions, or excessive rates of credit expansion.1

Second-generation models, in contrast, analyze crises as essentially avoidable financial panics—self-fulfilling attacks against otherwise viable economies. The models are characterized by multiple equilibria, where sudden shifts in market expectations and confidence can lead to collapse. Countries with low liquidity ratios are more vulnerable to such self-fulfilling panic. Another factor contributing to vulnerability is policymakers’ trade-offs between conflicting objectives, such as employment and exchange rate stability. Investors’ bets that the authorities will compromise on the latter rather than the former become self-fulfilling as an attack is launched on the currency.2

Interestingly, a number of second-generation models imply that a country is vulnerable to a self-fulfilling attack only over a range of values for the economic fundamentals. Hence, even in this approach, only countries with weak fundamentals will be vulnerable to panic.3 Although this blurs somewhat the mapping between variables and models, first-and second-generation models clearly have different empirical implications: in the first-generation approach, the probability of a crisis increases steadily in the run-up to a crisis, whereas second-generation models do not imply any particular trend in the probability of crisis prior to the crisis itself.

The Asian crisis seemed to tilt the consensus toward the second-generation models as more representative of recent crises. However, this was not unanimous, and third-generation models emerged in the post-crisis literature. These models acknowledged that second-generation models captured some aspects of the Asian crisis (notably the existence of multiple equilibria) but noted that other factors were also at play— namely, corporate and financial weaknesses.4 These models highlighted the flawed incentive structures under which the corporate and financial sectors operated in the Asian crises countries. These structures were characterized by regulatory inadequacy and close links between public and private institutions— including a long tradition of public guarantees to private sector projects—which caused an overreliance of firms on debt and foreign financing. Excessive foreign borrowing was also encouraged by exchange rate policies aimed at stabilizing the value of the domestic currency in terms of the U.S. dollar (thus lowering the risk premium on dollar-denominated debt) and international banks’ apparent neglect of the standards for sound risk assessment. Excessive risk taking and over-investment were reflected in persistent current account deficits and a highly vulnerable corporate financing structure.

More recently, a micro-based branch of this literature has examined the microeconomic and institutional factors underlying the corporate sector problems, such as ownership structure (characterized by links and cross-ownership between the corporate and banking sectors and between the government and both sectors) and flawed corporate governance.5

1 See the seminal contribution of Krugman (1979).2 For example, see Chang and Velasco (1998); and Obstfeld (1996).3 For example, see Flood and Marion (1998).4 For example, see Krugman (1999); and Bris and Koskinen (2000).5 For a review of recent micro-based approaches, see Khatri, Leruth, and Piesse (2001).

Crisis Cost Models

Crisis cost models have identified some important factors that could aggravate the negative growth impact of a crisis, if it were to occur.5 These include the potential for reversal of capital flows (proxied by high pre-crisis cumulative capital inflows and a pre-crisis business cycle boom); high openness to capital flows and low openness to trade; low liquidity (proxied by a high ratio of short-term debt to reserves); and financial sector weaknesses that could lead to an accompanying banking crisis. Rising oil prices are identified as another aggravating factor in the cross-country models, although they would have a favorable effect on some Latin American countries that are oil exporters (notably Venezuela and Mexico).

Box 2.2.Models of Ratings, Spreads, and Crisis Costs1

Sovereign ratings are intended to represent long-term forward-looking assessments of an economy’s prospects based on economic fundamentals. They typically serve as a benchmark (ceiling) for private sector ratings. The existing literature identifies a number of quantitative factors that help to explain ratings: per capita income, GDP growth, inflation, fiscal balance, external balance, ratio of external debt to exports or GDP, ratio of short-term debt to reserves, and interest rate spreads (domestic lending rates minus London interbank offered rate, LIBOR). Several studies have also found that after the Asian crisis the importance of interest rate spreads and the ratio of short-term debt to reserves as determinants of sovereign ratings increased.2

Most spread models start from some specifications of the probability of default and tie it to a spread. Long-run equilibrium or “fair value” spreads reflect a country’s capacity to pay and can be estimated across countries and over time using cointegration methods, as a function of solvency and liquidity variables (real GDP growth, total external amortization as a ratio to reserves, ratio of external debt to GDP, fiscal balance, trade openness, real exchange rate misalignment), external liquidity conditions (international interest rates), and the country’s default history. The underlying short-run country-specific model (error-correction model) can then be estimated individually for each country, allowing the speed of adjustment to vary across countries (empirical findings suggest that it varies anywhere from 1 to 64 months). Because spreads tend to overshoot, adjustment speeds provide a guide to the relative speed of reversals of spreads back to equilibrium.

The recent literature on the cost of crises identifies a number of factors that help to explain output changes associated with crises. By focusing on the impact of crises on output in terms of level or growth, rather than just the likelihood of crisis, this literature complements early warning system (EWS) modeling. Gupta, Mishra, and Sahay (2001) find that the reversal of capital flows is an important explanation for a growth slowdown after a currency crisis. A surge in private capital inflows prior to the crisis, openness to capital flows and not to trade, depreciation in competitor countries, low export growth, large short-term debt relative to reserves, a rise in oil prices, and high pre-crisis growth rates are key factors associated with a decline in growth following the crisis. Tight monetary policy following the crisis would also exacerbate the depth of the crisis, at least in the short run.

Hoggarth, Reis, and Saporta (2001) show that the resolution costs of a banking crisis tend to be higher in lower-income countries, in countries with a higher degree of banking intermediation, and in countries experiencing a currency crisis that previously had a fixed exchange rate regime. Unlike resolution costs, output losses associated with banking crises tend to increase with the length of crises, suggesting that forbearance may allow output losses to accumulate not only year by year but also at an increasing rate, as actual output deviates further from its potential. Output losses are also higher when a banking crisis is accompanied by a currency crisis.

1 For examples of ratings models, see Cantor and Packer (1996); Ferri, Liu, and Stiglitz (1999); Monfort and Mulder (2000); and Mora (2001). For spread models, see Goldman Sachs (2000, 2001); and Eichengreen and Mody (1998). For crisis cost models, see IMF (1998); Aziz, Caramazza, and Salgado (2000); Gupta, Mishra, and Sahay (2001); and Hoggarth, Reis, and Saporta (2001).2 Ratings changes have also been found to be pro-cyclical, i.e., downgrades tend to take place during economic downturns or crises.

The results of these models must be interpreted with caution. Most models predict only small probabilities of crises. Reviews of an earlier generation of EWS models indicate that, with a couple of exceptions, most such models did not perform well in predicting the Asian crises. The models were updated to include the Asian crises and performed somewhat better in predicting subsequent crises. But the models often aggregate crises that differ greatly in origin.

One striking feature of the experience of recent crises is the diversity of configurations of relevant factors. Any single factor identified could, by itself, result in a crisis, but in the crises that occurred multiple factors have often interacted in varying ways.

Moreover, one cannot assume that models of crises are now complete: there may be other aspects of vulnerability that are currently overlooked that may generate future crises.

While many of the factors identified as important in previous crises have been present in Latin America, some additional factors, which are not in the models, are also relevant. Public debt management—including the large share of foreign-currency-linked, indexed, or floating-rate debt, and the bunching of repayment obligations—extensive dollarization, and a concentration of nonresident deposits are significant potential sources of vulnerability for several countries in the region.

In some instances—notably Brazil and Argentina—a major consideration is the share of debt issued by a single country in relation to emerging markets as a whole (Table 2.3). Bigness in the relevant market is in itself a source of vulnerability. Under these circumstances, a decline in creditors’ appetite for emerging market debt quickly translates into difficulty in rolling over a country’s liabilities. This vulnerability is painfully illustrated by Argentina’s rank, ahead of its default, as the leading emerging market issuer.

Table 2.3.Sovereign Debt Issuance, 1998–2002
In Billions

of U.S. Dollars
In Percent

of Total
Argentina32.416.3
Brazil23.912.0
Turkey20.510.3
Mexico17.68.9
Philippines11.96.0
Russia11.35.7
Colombia9.64.8
Lebanon8.24.1
Hungary5.22.6
South Africa4.32.2
Top ten144.972.9
Total198.7100.0
Source: IMF, Bonds Equities Loans (BEL) database.
Source: IMF, Bonds Equities Loans (BEL) database.

Other countries with low predicted crisis probabilities based on EWS and country risk models nonetheless appear vulnerable in light of weaknesses in their corporate sectors that are not accounted for in standard EWS and credit risk models. Weaknesses in the corporate sectors (low liquidity and high leverage) appear a source of vulnerability in Chile. In other Latin American countries, vulnerability is attenuated by the relatively low degree of reliance on the financial system, the legacy of years of high inflation.

Policy Paralysis

In each of the previous financial crises, a major reason for the loss of market confidence was the markets’ assessment of the capacity of the government to undertake the policy adjustments needed to avert or resolve a crisis. This was due to a combination of economic and political constraints on policymakers.

Economic Constraints

The role of economic constraints on policies is captured in second-generation models of crises: in these models, policymakers are concerned by the adverse economic implications of applying their macroeconomic policy instruments. In many cases, these constraints are the other side of the vulnerabilities of balance sheets to interest rate and exchange rate movements. In many crisis countries, the early policy response was characterized by “fear of floating,” associated inter alia with extensive dollarization of liabilities.6 The authorities clung to unsustainable exchange rate pegs, owing in part to their awareness of the disastrous balance sheet effects of allowing the exchange rate to float. This meant that when the currencies were finally forced to float, the new regime was introduced under much less favorable conditions—in particular, after massive drainage of reserves and a major loss in policy credibility (e.g., Thailand). In the case of Argentina, the consequences of clinging to a peg were much worse: the convertibility regime was maintained through 2001 at the cost of a massive violation of property rights in late 2001 and early 2002.

In many cases (notably in the Asian crisis countries), pressures on exchange rate pegs were motivated by questions about whether the authorities would be able to defend the pegs, given the damaging effect that high interest rates would likely have on weak financial and highly leveraged financial and corporate sectors. Early inaction by the authorities seemed to prove these market perceptions correct. But it soon became apparent that failing to defend the currency peg could be at least equally damaging, given the foreign currency exposures—a no-win situation.

In other cases, the impact of interest rates on the public sector was of foremost importance. In Brazil, Argentina, and Turkey, market participants wondered whether the authorities would be able to raise interest rates because of the effect that this would have on the already precarious public debt dynamics.7 In this setting, it was natural for markets to doubt the authorities’ ability to stomach the effect of higher interest rates, which would result in a need for additional primary fiscal adjustment in order to maintain a sustainable fiscal position.

Political Factors

Another major source of doubt pertained to the political system’s ability to deliver the decisive action needed to tackle a crisis. Political turbulence is a fact of life in the economic policy environment, but it is particularly relevant in financial crises, where averting a prolonged crisis turns on market participants’ perceptions that policies will stay on track over the medium term.

In previous crises, political factors played a role in the depth and duration of the crisis by affecting the market’s perception about the authorities’ willingness and ability to act. The importance of these factors is heightened in the period preceding a political transition, when the degree of market anxiety about the future course of policies is always greatest. For instance, in Indonesia in late 1997 and early 1998, the resistance of President Suharto to many of the proposed policy measures, uncertainty about the succession, and later the political vacuum following Suharto’s ouster impaired an effective policy response. In Malaysia during the first several months of 1998, the tension between different factions of the ruling party was an important factor. In Russia, the background to the 1998 crisis was the many political uncertainties and machinations that undermined policy credibility in the latter part of the Yeltsin era. In Turkey in 2001, the crisis was exacerbated by political disagreements within the governing coalition. In Russia (as more recently in Argentina), fiscal federalism was a major factor constraining policy action at the center. In Korea, political sniping in the run-up to presidential elections undermined the initial policy response; but then a tripartite agreement between the new government, business, and labor contributed to favorable market perceptions of the authorities’ capacity to implement reforms.

In a number of Latin American countries, the role of political factors is intensified by the breadth of differences among contending political forces—which may in turn partially reflect the width of income inequalities and the prevalence of poverty in most countries of the region. In a number of countries— notably Argentina, Brazil, and Venezuela—the political debate has placed on the table fundamental questions about the approach to economic development, and the validity of basic property rights—in contrast with the tendency in many industrial countries for political contests to converge to the center.

Important differences in the political configurations across Latin American countries are relevant for various countries’ abilities to avoid crises. In Argentina in the run-up to the 2001–02 crisis, while broad and strong popular support existed for the currency board, much less consensus emerged about the policies required to defend it: the perception of corruption, mismanagement, and inequality made the public relatively unwilling to bear the costs of fiscal adjustment or financial restructuring. In Brazil in 2002, the incoming president had for years strongly criticized the market-oriented policies pursued during a decade of inflation stabilization, privatization, and a number of structural reforms and advocated default on external debt. The market jitters elicited by the prospect of his election calmed with his shift toward the center during the election campaign and with the tone of moderation in his post-election statements. The lack of consensus in a number of countries reflects, in part, a political effort to manage divergent views inside democratic institutions, which are still being consolidated in Latin America.

Some Latin American countries have experienced more extreme political divisions in recent years. In Venezuela, wide disagreements on the direction of economic policies have been combined with challenges to political institutions and constitutional processes. In Ecuador, several changes of administration have taken place within the past few years outside constitutional processes. In Paraguay, problems with governance have long been serious and widespread, and dissension within the ruling group came to a head during the 2002 crisis.

In contrast, a few Latin American countries have exhibited a relatively high degree of consensus. The high degree of consensus on economic policies seems to have been a positive factor in Chile, which has largely escaped contagion from the turmoil elsewhere in Latin America. In Uruguay, a continuing commitment to democratic processes has been an important anchor for political stability. Although this has not prevented crisis from spilling over from its close neighbor Argentina, there has been little or no question raised in the market about the political capacity of the Uruguayan authorities to carry out the required reforms. In Colombia, democratic processes and a degree of political stability have prevailed in the face of destructive and tragic internal conflicts involving rebel and paramilitary groups.

Contagion

In the wake of previous crises, considerable attention has been paid to the channels by which crises spread. The initial impression was that geography is a determinant, but this only goes so far by way of explanation. The 1994–95 Mexican “Tequila” crisis had a significant impact on other Latin American countries (notably Argentina, where it triggered a deposit run under the currency board) plus Turkey. The 1997 Thai crisis spread to a number of other Asian countries, notably Indonesia, Korea, Malaysia, and the Philippines. The 1998 Russian crisis led to a wave of market unrest that within a few months threatened a generalized collapse of emerging markets—triggering pressures on Brazil and exacerbating Ukraine’s liquidity problems (which were triggered in the wake of the Asian crisis). Contagion from the 2001–02 Argentine crisis has been more complex, given the slow-motion nature of the crisis: during 2001, investors covered their exposures, and depositors moved money out of Argentina to neighboring countries (notably Uruguay). After the deposit freeze and redenomination in Argentina, however, Uruguay’s financial system came under heavy pressure as nonresident deposits were withdrawn, while Latin Amercian emerging markets overall were affected by heightened awareness of risk, exacerbated by contemporaneous political uncertainties.

Channels of Contagion

There are several basic reasons for contagion. First, given imperfect information, creditors who see problems in one country may perceive similar problems in other countries and reduce exposure in anticipation of these problems emerging. At times, there is an element of “regional profiling” as investors reduce their exposure to an entire region in response to troubles in one country. This type of contagion should diminish with more and better information that would allow investors better to differentiate among countries: in the immediate wake of the collapse in Argentina, market participants apparently saw Argentina as unique.

Second, real linkages involving trade flows play an important role. These can work via import substitution, competition in third markets, or upstream effects on suppliers of inputs. These linkages were important in transmitting the Asian crisis, as discussed below.

A third important set of factors involve financial linkages, through bank lending (common creditors, nonresident deposits, portfolio reallocation, etc.) and, to a lesser extent, bank and corporate ownership. In some cases, these linkages are reinforced by technical factors. For instance, when asset returns on countries in a region are correlated, investors wishing to reduce their exposure to one country but unable to do so (given market imperfections) may reduce their exposure vis-à-vis other countries. Such behavior in turn tends to reinforce the correlations among returns. (This is a channel by which concerns about Brazil could be transmitted to Chile.)

A fourth variant is the “wake-up call” hypothesis.8 Market participants would seek to reduce their exposure to countries that are seen as having common vulnerabilities, even in the absence of direct spillovers.9

Another consideration, discussed further below, is the possibility of policy spillovers, such as competitive devaluations or spillovers from fiscal tightening.

Appendix 2.1 focuses on the aspects of contagion for which there are clear measures: trade linkages and financial sector linkages, including both cross-border lending and systemic factors. The growing body of empirical evidence on contagion suggests that financial market linkages are the most important factor systematically explaining contagion in recent crises.10 These studies have typically found marginally significant effects from trade spillovers. A more powerful channel arises from the fact that countries in a given region tend to borrow from banks based in similar countries, leading to a “common creditor” effect. Risks are particularly high for banking systems in Latin America, given the substantial ownership linkages across countries to institutions in high-risk and crisis-afflicted countries (Table 2.4).11 From these calculations, banking systems in Bolivia and Paraguay could be affected by a credit event else-where in the region.

Table 2.4.Latin America: Foreign Banks’ Share of the Top Ten Banks’ Assets1(In percent unless otherwise noted)
Owners of Foreign BanksBolivia

(Mar.

2002)
Brazil

(Dec.

2002)
Chile

(May

2002)
Colombia

(May

2002)
Ecuador

(Jun.

2002)
Guatemala

(Dec.

2002)
Mexico

(Mar.

2002)
Paraguay

(Dec.

2002)
Peru

(Dec.

2001)
Uruguay

(May

2002)
Venezuela

(Jun.

2002)
Industrial countries23.715.046.510.06.95.681.167.261.420.542.1
Spain16.75.437.510.05.643.87.927.89.238.2
Santander Central Hispano16.7(1)5.4(1)30.7(2)16.1(2)7.9(1)4.4(1)20.1(1)
Banco Bilbao Vizcaya Argentaria6.8(1)10.0(1)27.7(1)7.9(1)18.1(1)4.8(1)18.1(1)
I.F. Group5.6(1)1.8(1)
United States7.04.55.04.730.821.46.511.3
Citibank7.0(1)2.1(1)5.0(1)4.7(1)29.1(1)21.4(1)4.7(1)5.5(1)
Bank Boston (FBF)2.4(1)1.8(1)5.8(1)
JP Morgan Chase1.7(1)
Canada4.05.43.83.9
Bank of Nova Scotia4.0(1)5.4(1)3.8(1)3.9(1)
United Kingdom2.12.210.6
HSBC2.1(1)
Lloyds Bank2.2(1)10.6(1)
Netherlands3.01.216.9
ABN AMRO3.0(1)16.9(1)
ING Bank1.2(1)
Italy/France10.523.3
INTESABCI/Banque Sudameris10.5(1)23.3(1)
Sources: National authorities; and IMF staff estimates.

The number of foreign banks’ subsidiaries in the top ten largest banks in each country is shown in parentheses.

Sources: National authorities; and IMF staff estimates.

The number of foreign banks’ subsidiaries in the top ten largest banks in each country is shown in parentheses.

Policy Spillovers

What about contagion arising from policies taken in response to the crisis—such as exchange rate depreciation, trade protection, and contractionary spending policies? Trade spillovers appear to have been significant in transmitting the impact of the Asian crisis across the region. However, the spillovers were generally a by-product of currency movements and demand contractions over which the authorities had little control, as opposed to active policy decisions.

Nonetheless, there are isolated examples where discretionary exchange rate depreciation had repercussions on trading partners. The most obvious was the decision by Taiwan Province of China to allow its currency to depreciate against the U.S. dollar in mid-October 1997. This triggered strong pressures on neighbors, notably Hong Kong SAR, and arguably was an important factor contributing to the second leg of the Asian crisis in late 1997.12

How important were spillover effects from currency movements on neighbors? During the Asian crisis, for instance, bilateral depreciations of individual crisis currencies translated into large real effective depreciations, notwithstanding movements in neighbors’ exchange rates. Table 2.5 indicates that, on average, changes in real effective exchange rates (REERs) in the crisis countries amounted to some-what over half the size of depreciations against the U.S. dollar. Nevertheless, the export volume response was muted. Volume growth was actually lower after the depreciations than before, while real activity in partner countries grew at about the same pace.

Table 2.5.Selected Asian Countries: Change in Export Volume Growth versus Different Measures of Competitiveness
Change in

Export Volume

Growth1

(percentage points)
Change in

Partner Country

Real GDP Growth1

(percentage points)
Change in

Exchange Rate

Versus U.S. Dollar

(log change x 100)
Change in REER:

CPI-based

INS Measure2

(log change x 100)
Change in REER:

Export Deflators Using

Bilateral Trade Weights2

(log change x 100)
Crisis countries
Korea0.9−0.3−54.7−35.9−15.4
Malaysia1.3−0.2−44.2−24.12.3
Indonesia−4.4−0.2−141.2−78.4−13.0
Thailand5.0−0.2−47.8−24.2−9.0
Philippines−3.50.0−44.2−20.913.8
Average−0.1−0.2−66.4−36.7−4.3
Other countries
Japan2.00.0−18.1−6.8−1.3
China−2.00.50.44.53.6
Hong Kong SAR1.2−1.1−0.117.18.2
Taiwan Province of China5.90.3−19.7−8.9−1.0
Sources: IMF, IFS, and IMF staff estimates.

Average growth in 1999–2000 less that in 1994–96.

Change from 1996 to 1998 in real effective exchange rate (REER), calculated as the IMF’s Information Notice System (INS) value based on the consumer price index (CPI); decline indicates greater competitiveness.

Sources: IMF, IFS, and IMF staff estimates.

Average growth in 1999–2000 less that in 1994–96.

Change from 1996 to 1998 in real effective exchange rate (REER), calculated as the IMF’s Information Notice System (INS) value based on the consumer price index (CPI); decline indicates greater competitiveness.

This suggests that the increase in export competitiveness from conventional REER measures was overstated, or that other factors were at work to limit the export response. A different view of the change in competitiveness is provided by looking at an REER measure constructed using export unit values instead of consumer price indices (CPIs), and bilateral trade weights as opposed to multilateral trade weights, as in the IMF’s Information Notice System (INS). This measure indicates an average depreciation in the Asian crisis countries of only 4 percent. On this basis, depreciations in other countries—and changes in their export prices—offset almost all of the nominal depreciations against the U.S. dollar in Asian countries.

Exchange rate depreciations in Latin American countries such as Chile and Mexico have not had the same spillovers in the current crisis (Table 2.6). Flexible currency arrangements can contribute to reducing contagion through trade channels. Moreover, the low share of intraregional trade in most Latin American countries reduces the potential magnitude of adverse trade spillovers (although a low trade share increases vulnerability to a debt crisis by reducing a country’s ability to export its way out of debt-servicing difficulties). But there are considerable differences across countries, both in the trade share and in the direction and composition of trade (Table 2.7).

Table 2.6.Selected Latin American Countries: Change in Export Volume Growth versus Different Measures of Competitiveness
Change in

Export Volume

Growth1

(percentage

points)
Change in

Partner Country

Real GDP Growth1

(percentage

points)
Change in

REER: CPI-

based INS

Measure2

(log change x 100)
Change in REER:

Export deflators

Using Bilateral

Trade Weights3

(log change x 100)
Crisis countries
Argentina4.40.8−87.2−11.2
Brazil3.20.6−17.6−2.6
Uruguay13.21.3−12.7−29.4
Other countries
Chile0.00.5−12.9−7.5
Mexico−2.8−0.68.90.2
Peru−0.50.210.84.6
Colombia2.4−0.63.2−2.6
Sources: IMF, World Economic Outlook (WEO) database, IFS, and IMF staff estimates.

WEO projected average growth in 2003–04 less that in 1998–2000.

Change from 2000 to 2002 WEO projections in real effective exchange rate (REER), calculated as the IMF’s Information Notice System (INS) value based on the consumer price index (CPI) assumed to remain constant in the second half of 2002 at June level; a decline indicates greater competitiveness.

Change from 2000 to 2002 based on WEO projections.

Sources: IMF, World Economic Outlook (WEO) database, IFS, and IMF staff estimates.

WEO projected average growth in 2003–04 less that in 1998–2000.

Change from 2000 to 2002 WEO projections in real effective exchange rate (REER), calculated as the IMF’s Information Notice System (INS) value based on the consumer price index (CPI) assumed to remain constant in the second half of 2002 at June level; a decline indicates greater competitiveness.

Change from 2000 to 2002 based on WEO projections.

Table 2.7.Selected Asian and Latin American Countries:Trade Shares
Total Trade (percent of GDP)
19952000
Asia
Korea61.886.4
Malaysia191.8230.6
Philippines81.1103.8
Thailand91.0125.5
Indonesia54.382.4
Singapore340.5339.6
Hong Kong SAR298.2289.5
Latin America
Argentina20.622.5
Bolivia42.042.8
Brazil15.523.2
Chile52.359.6
Colombia30.738.5
Ecuador56.972.7
Uruguay36.739.4
Paraguay111.481.3
Peru30.534.6
Venezuela48.944.5
Source: IMF, World Economic Outlook.
Source: IMF, World Economic Outlook.

Another potential aspect of policy contagion is that of contractionary fiscal policies.13 Here, the concern is that fiscal contraction aimed toward medium-term sustainability could dampen economic activity, with an adverse effect on neighboring countries’ exports that would undermine their growth and thus their fiscal sustainability. That conventional Keynesian fiscal multipliers are higher in Latin America than, for instance, in the Asian crisis countries would tend to make this effect larger;14 on the other hand, this effect works mainly through bilateral trade links among neighboring countries, which are not large among most Latin American countries. More important, in most recent crises fiscal consolidation has made a comparatively minor contribution to the economic downturn: the recessions have resulted mainly from the collapse of private domestic demand.15

Concluding Remarks

Any assessment of vulnerability must be taken with strong caveats. Predicting crises is still more an art than a science, and the experience of previous crises suggests that one should expect the unexpected. Crises tend to reflect vulnerabilities that market participants and policymakers had previously overlooked. Time and again, experience has shown that the star performers of today may turn out to be the crisis countries of tomorrow; it is thus essential to scrutinize factors that, in the midst of market enthusiasm for a particular country, may have passed unnoticed. These considerations imply that there are inherent limits to the ability of any methodology to predict crises—since any reliable model of predicting crises is likely to be used by market participants to cover their exposures and eschew the inflows of capital that contribute to balance sheet vulnerabilities. But examining the experience of past crises is helpful in attempting to prevent future crises and to understand and manage those that nonetheless occur.

Appendix 2.1. Assessing the Scope for Emerging Market Contagion

As in the capital account crises of the 1990s, the risks of contagion across countries have been an important source of vulnerability across emerging markets. Three potential channels of contagion are considered: a retrenchment of cross-border lending, trade flows, and the financial sector. While these channels of contagion pose potential risks to the asset class as a whole, there are indications that portfolio investors have adopted an increasingly discriminatory stance toward individual markets, largely based on the country’s policy regime and track record.

In recent years, banks have built increasingly global operations in part through direct equity investments in emerging markets. As business platforms have been broadened, risks have been diversified. The assessment of risks, however, tends to be conducted in broad terms. In times of heightened risk aversion, this puts emerging markets at the fore-front, and banks may choose to retrench from emerging markets as whole, while not always taking into account the particular circumstances of individual markets.

Such decisions tend to be taken on the basis of a bank’s total exposure, without discriminating whether the exposure was built up via their bank’s local branch operations or through operations in the major financial centers. The analysis that follows captures this notion by using Bank for International Settlements (BIS) data, which consolidates all on-balance-sheet claims of BIS registered banks on private and public entities, including those claims held by local branches and subsidiaries. These data, therefore, may also differ significantly from international debt statistics.

Retrenchment of Cross-Border Lending

Bank exposure to Brazil is large—$142 billion at end–2001—and relatively concentrated. Whereas U.S. banks provided $34 billion in short-and long-term financing, Europe’s four largest lenders (aggregated by nationality) provided $68 billion in financing to Brazil; Spain alone provided $26 billion (Table 2.8).

Table 2.8.Selected Emerging Markets: BIS Lending and Common Creditor Index1
Share of Total (percent)
Four largest European lenders
Common

Creditor

Index
BIS Loans,

End–2001

(US$ billions)
United StatesTotalSpainUKNetherlandsGermanyOther
Brazil142.424.047.618.011.110.77.728.4
Colombia0.8316.425.052.128.67.95.210.422.9
Argentina0.8173.929.145.924.48.63.79.125.0
Mexico0.80215.135.947.939.52.72.13.716.2
Taiwan Province of China0.8032.236.633.90.016.712.64.529.6
Korea0.7973.222.422.80.29.53.99.254.8
Philippines0.7922.421.733.60.611.54.916.644.8
Venezuela0.7721.615.064.944.77.13.59.520.1
Chile0.7743.916.964.749.33.93.87.718.4
South Africa0.7622.314.138.30.28.46.223.647.6
Indonesia0.7637.49.041.40.210.09.321.949.5
Thailand0.7542.410.329.60.09.810.59.360.1
Bulgaria0.741.511.451.20.62.317.730.737.4
Turkey0.7439.711.343.50.97.87.027.845.2
Russia0.6941.56.061.90.61.56.853.032.1
Sources: IMF, IFS; and Bank for International Settlements (BIS).

The common creditor index average is 0.49 across a group of 165 developing countries with outliers, such as Sierra Leone (0.20) and Tonga (0.06).

Sources: IMF, IFS; and Bank for International Settlements (BIS).

The common creditor index average is 0.49 across a group of 165 developing countries with outliers, such as Sierra Leone (0.20) and Tonga (0.06).

There is a risk of rollover problems of these loans in Brazil and emerging markets in general, including movement for “safe havens.” A potential retrenchment of U.S. lending could adversely affect Mexico and Taiwan Province of China, both having raised more than one-third of their external bank financing in the United States. In addition, Mexico raised 40 percent of its bank financing from Spain, highlighting the risk of a potential reduction of resources from this source. A broad-based retrenchment by European lenders would heavily affect Latin America as well as emerging markets in Europe. More than half of bank financing was provided by Europe’s four largest lenders to Russia, Bulgaria, Colombia, Chile, and Venezuela, with Spain having provided the bulk of financing to Chile (49 percent) and Venezuela (45 percent).

Private and public borrowers in countries with borrowing patterns similar to those of Brazil could also suffer from a potential retrenchment of bank lending. The common creditor index in Table 2.8 measures the similarity of countries’ borrowing patterns relative to those of Brazil. The index ranges between 0 and 1, with a higher index value indicating patterns more akin to those of Brazil.

Redirection of Trade Flows

Sharp exchange rate movements and a subsequent redirection of international trade flows can also act as a channel of contagion, although these have proved to be of less importance in the financial crises of the 1990s (Figure 2.1). We apply a widely used trade competition index, which—analogous to the common creditor index—measures the similarity of countries’ trading patterns with those of Brazil (Figure 2.2). The index ranges between 0 and 1, with a high index value indicating a high degree of overlap of a country’s export markets with those of Brazil. The index, however, does not take into account differences in the product composition of exports, and a high index value, therefore, is a necessary but not sufficient condition for a high degree of competition in third markets. Assessments of vulnerability through this channel should also take into account the overall current account position (Figure 2.2): countries with substantial current account surpluses—such as those in Asia—will have a cushion against trade-related contagion.

Figure 2.1.Common Creditor Index versus Ratio of Foreign Exchange Reserves to Short-Term Debt, 2001

Sources: International Monetary Fund, World Economic Outlook database; and IMF staff estimates.

Figure 2.2.Trade Competition Index versus Current Account, 2001

Sources: International Monetary Fund, World Economic Outlook database; and IMF staff estimates.

Dollarization and Systemic Risks

Dollarization of financial sector balance sheets poses systemic risks and is a potential further channel of contagion, particularly as depositors observe the experience in neighboring countries. Nonresidents’ withdrawals of foreign exchange deposits triggered bank runs in Uruguay and Paraguay in the aftermath of the Argentine crisis. Because commercial banks’ liquid foreign exchange assets were inadequate to cover the demand for deposits, central banks provided lender-of-last-resort financing, thereby straining their official foreign exchange reserves. Besides the risk of a sudden outflow of deposits, banks are exposed to credit risks when extending loans denominated in foreign exchange to unhedged borrowers. Credit risk thus can cause or aggravate mismatches of assets and liabilities in foreign exchange.

Dollarization is prevalent in Latin America and is a policy issue for other countries too, including Turkey (Figure 2.3). In Latin America, regulation has prevented dollarization in Brazil and Mexico, and most economies exhibiting a high degree of dollarization are relatively small. This suggests that the global systemic risks associated with dollarization are more likely to manifest as a result of a crisis in the region rather than as a potential cause—not unlike the experience of Uruguay and Paraguay.

Figure 2.3.Dollarization in Selected Emerging Markets, End–2001

(Foreign exchange deposits in percent of total deposits)

Source: IMF staff estimates

References

    AzizJahangirCaramazzaFrancesco and SalgadoRanil2000Currency Crises: In Search of Common Elements,” IMF Working Paper 00/67 (Washington: International Monetary Fund, March).

    BergAndrew and CatherinePatillo1999aAre Currency Crises Predictable? A TestIMF Staff Papers Vol. 46 (June) pp. 10738.

    BergAndrew and CatherinePatillo1999bPredicting Currency Crises: The Indicator Approach and an Alternative,Journal of International Money and Finance Vol. 18 No. 4 pp. 56186.

    BrisArturo and YrjöKoskinen2000“Corporate Leverage and Currency Crises” (unpublished;Yale University and Stockholm School of Economics).

    CantorRichard and FrankPacker1996“Determinants and Impact of Sovereign Credit Ratings,”FRBNY Economic Policy Review (New York: Federal Reserve Bank of New York, October).

    CaramazzaFrancescoLucaRicci and RanilSalgado2000Trade and Financial Contagion in Currency Crises,” IMF Working Paper 00/55 (Washington: International Monetary Fund).

    ChangRoberto and AndresVelasco1998Financial Crisis in Emerging Markets: A Canonical Model,” NBER Working Paper6606 (Cambridge, Massachusetts: National Bureau of Economic Research).

    EichengreenBarry and A.Mody1998What Explains Changing Spreads on Emerging Market Debt: Fundamentals or Market Sentiment?” NBER Working Paper 6408 (Cambridge, Massachusetts: National Bureau of Economic Research).

    FerriGiovanniLi-GangLiu and JosephStiglitz1999The Procyclical Role of Rating Agencies: Evidence from East Asian Crisis,Economic Notes Vol. 28(3) pp. 33555.

    FloodRobert and Nancy P.Marion1998Perspectives on the Recent Currency Crisis Literature,” IMF Working Paper98/130 (Washington: International Monetary Fund).

    GhoshAtish2002IMF-Supported Programs in Capital Account Crises, Occasional Paper210 (Washington: International Monetary Fund).

    GoldmanSachs2000“Introducing GS-ESS: A New Framework for Assessing Fair Value in Emerging Markets’ Hard-Currency Debt,”Global Economics Paper No. 45 (New York,June).

    GoldmanSachs2001“Introducing GS-ESS Second Edition,”Emerging Markets Strategy (New York,July26).

    GoldsteinMorris1998The East Asian Financial Crisis: Causes, Cures, and Systemic Implications (Washington: Institute for International Economics).

    GoldsteinMorris1993International Capital Markets, Part I: Exchange Rate Management and International Capital Flows (Washington: International Monetary Fund,April).

    GuptaPoonamDeepakMishra and RatnaSahay2001“Output Response to Currency Crises,”paper presented at the IMF’s Annual Research ConferenceNovember;http://www.imf.org/external/pubs/ft/staffp/2001/00-00/pdf/pgdmrs.pdf

    HoggarthGlennRicardoReis and VictoriaSaporta2001“Costs of Banking System Instability: Some Empirical Evidence” (London: Bank of England).

    International Monetary Fund1997World Economic Outlook—Interim Assessment (Washington: International Monetary Fund,December) Chapter 3.

    International Monetary Fund1998“Financial Crises: Characteristics and Indicators of Vulnerability,”World Economic Outlook (Washington,October) Chapter 2.

    International Monetary Fund2002Global Financial Stability Report—Market Developments and Issues (Washington,September) Chapter 3.

    KhatriYougeshLuc E.Leruth and JeniferPiesse2001“Corporate Performance and Governance in Malaysia” (unpublished; Washington: International Monetary Fund).

    KrugmanPaul1979A Model of Balance of Payments Crisis,Journal of Money Credit and Banking Vol. 11 pp. 31125.

    KrugmanPaul1999Balance Sheets, the Transfer Problem, and Financial Crises,International Tax and Public Finance Vol. 6. No. 4.

    MonfortBrieuc and ChristianMulder2000Using Credit Ratings for Capital Requirements on Lending to Emerging Market Economies: Possible Impact of a New Basel Accord,” IMF Working Paper00/69 (Washington: International Monetary Fund).

    MoraNada2001Sovereign Credit Ratings: Guilty Beyond Reasonable Doubt?” Working Paper (Cambridge, Massachussets: MIT).

    ObstfeldMaurice1996Models of Crises with Self-Fulfilling Features,European Economic Review Vol. 40 (April) pp. 103747.

    RoseAndrew and ReuvenGlick1999Contagion and Trade: Why Are Currency Crises Regional?Journal of International Money and Finance Vol. 8 No. 4 pp. 60317.

The initial version of this section was prepared in the summer of 2002.

In addition, IMF staff examine the vulnerabilities of the financial system through the Financial Sector Assessment Program (FSAP). Two of the stated goals of the FSAP are to identify strengths, risks, and vulnerabilities of financial systems; and to determine how key sources of risk are being managed. The FSAPs include stress tests of the financial sector, examining the impact on balance sheets of various hypothetical shocks.

The short-term debt variable has been identified in some studies as the single most important indicator of the likelihood of an external crisis occurring within the next 12 to 24 months. See Berg and Patillo (1999a and 1999b). In contrast, the money-and import-cover-based indicators have turned out to have little explanatory power.

Especially when the leveraged financing of the corporate sector is done mostly through the banking sector.

The results of crisis cost models must, however, be treated even more cautiously than those of EWS models, since the literature is in its infancy and there has been little opportunity to test the results out of sample.

Fear of floating and dollarization played out differently in Brazil, which in late 1998 provided foreign exchange hedges to its corporate and banking sectors. This helped to attenuate the macroeconomic consequences of the January 1999 devaluation when it occurred. But by fiscalizing the cost of the devaluation, it contributed to the buildup of debt that was the focus of market pressures in 2002.

Similar issues of debt dynamics lay behind Italy’s experience of the crisis in the European Monetary System’s exchange rate mechanism (ERM) in 1992 (see, for instance, Goldstein and others, 1993).

Some evidence supporting this hypothesis has been found by Rose and Glick (1999).

The impact of contractionary fiscal policy is discussed further in Section VI.

The high Keynesian multipliers in Latin America reflect relatively low saving rates, import shares, and tax pressure. Of course, the overall effect of fiscal consolidation on domestic demand also reflects the confidence effects on interest rates as well as directly on consumption and investment, which are more likely to be positive given these countries’ relatively high public debts.

    Other Resources Citing This Publication