What Is International Financial Contagion

International Finance 6:2, 2003: pp. 157–178

What Is International Financial Contagion? Thomas Moser International Monetary Fund.

Abstract Despite the growing popularity of blaming ‘contagion’ for international financial crises, contagion remains an elusive concern. Without a clear understanding of financial contagion and the mechanisms through which it works, we can neither assess the problem nor design appropriate policy measures to control it. This paper organizes and evaluates recent research on international financial contagion. The main finding is that the different mechanisms by which a crisis can spread greatly differ from each other both in their causes and implications. Policy measures that do not take these differences into account may do more harm than good.

I. Introduction Blaming financial crises on contagion has proved to be highly contagious among economists and politicians alike. For each of the international financial crises of the 1990s, financial contagion has increasingly received more credit, and some believe this to be an indispensable consequence of

I thank Barbara Do¨beli, Morris Goldstein, Werner Hermann, Maria Rueda Maurer, Jakob

Schaad, Umberto Schwarz, Charles Wyplosz, Fritz Zurbru¨gg and an anonymous referee for valuable comments and suggestions. The views in this paper are my own and do not necessarily reflect those of the IMF or IMF policy.

r Blackwell Publishing Ltd. 2003, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA

158

Thomas Moser

the ‘new global economy’ (IMF 1999, p. 66; Summers 2000, p. 6). This popularity has led to a flood of research on the topic.1 If it were true that financial crises can now easily spread across countries like contagious diseases, the consequences would be serious. It has been claimed that in the new global economy sound economic policies may not be enough to prevent financial crises, as ‘the threat of contagion makes even the virtuous vulnerable to currency runs’ (Feldstein 1999, p. 93). A financial crisis in any country may be a threat to the stability of the global financial system. The large international financial rescue packages of the 1990s and the subsequent call for a new international financial architecture have been responses to these concerns. To cure or at least mitigate financial contagion is one of the main aims of the ongoing efforts to reform the international financial system (Fischer 1999a, p. 557). Despite the advanced stage of the policy debate, however, and the large body of research it has generated, financial contagion remains an elusive concern. Economists do not agree and are often not clear about what they mean when they refer to it. Without a clear understanding of financial contagion and the mechanisms through which it works, though, we can neither seriously assess its importance nor design appropriate policy measures. This paper organizes and evaluates recent research on the subject.2 The main finding is that the mechanisms by which a crisis can spread greatly differ both in their causes and implications. This is of practical importance: Appropriate policy measures against contagion depend on the specific contagion mechanism. Otherwise, they may do more harm than good.

II. What Financial Contagion Does and Does Not Mean Not only ‘contagion’ but also ‘financial crisis’, the supposedly transmitted disease itself, is a remarkably imprecise term. The classical view in the tradition of Thornton ([1802] 1978) and Bagehot ([1873] 1962) restricts the term to problems in the banking sector (Wood 1999). In this classical view, crashes in asset markets that do not threaten the stability of the banking system are not financial crises; see, for example, Bordo et al. (1998). Most contemporary authors, however, use the term more broadly, even generically for events grouped into currency, banking and debt crises. Currency crises are habitually defined as periods of ‘large’ changes in the exchange rate or some index of market pressure, additionally allowing for 1

See, for example, the conference volume by Claessens and Forbes (2001).

2

For other attempts, see Claessens et al. (2001) and Pritsker (2001).

r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

159

changes in foreign reserves or interest rates to capture both successful as well as unsuccessful currency attacks (Eichengreen et al. 1996; Kaminsky and Reinhart 1999). Definitions of banking crises typically employ ratios of non-performing assets to total assets in the banking system or incidents of public sector crisis management such as rescue operations, takeovers, deposit freezes, generalized deposit guarantees or bank holidays (Demirgu¨c¸Kunt and Detragiache 1998; Kaminsky and Reinhart 1999). Debt crises are identified either by a certain level of arrears or by incidents of default or debt restructuring agreements (Detragiache and Spilimbergo 2001). Economists have used the concept of contagion even more widely. Contagion is a long standing idea in explanations for the spread of bank runs, but it has also been used to describe the spread of strikes across firms or industries (Price 1890, p. 425; Mavor 1891, p. 205), the spread of wage increases secured by labour unions to non-union firms or sectors (Ulman 1955, p. 397), the spread of business fluctuations across economies (Mack and Zarnowitz 1958, p. 18), the diffusion of technology and growth convergence across countries (Findlay 1978, p. 3; Baumol 1994), and the spread of speculative trading across individuals (Murchison 1933, p. 75; White 1940). In the recent literature, the concept is supposed to describe incidents in which a (suitably defined) financial crisis in one country brings about a crisis in another. In its broadest sense, therefore, financial contagion has to do with the propagation of adverse shocks that have the potential to trigger financial crises. The crux of the matter is to identify potential propagation mechanisms and define those that represent contagion. As a first step it is helpful to understand what contagion does not mean.

A. Independent Shocks and Common Shocks Simultaneous crises are not a sufficient condition for contagion. Contagion requires causal connection. The (near) simultaneous occurrence of financial crises may instead result from coincidence or common cause rather than causal links. In the case of coincidence, independent shocks hit countries at about the same time with no connection between the different crises, and diagnosing contagion would be a post hoc fallacy. With common cause, several countries are hit by a common global or regional external shock (other than a financial crisis). Candidates for such adverse common shocks with the potential of inflicting balance-of-payment difficulties, particularly in emerging market economies, are changes in global (US) interest rates, exchange rates between major currencies, commodity prices, or recessions in major industrial countries. r Blackwell Publishing Ltd. 2003

160

Thomas Moser

A rise in US interest rates adversely affects the funding of emerging market economies not only by increasing their debt servicing costs and lowering their creditworthiness (Calvo et al. 1996), but also by exacerbating asymmetric information problems (Mishkin 1997). Similarly, unanticipated drops in export prices may impair the corporate sectors’ capacity to service debts in countries that heavily depend on export earnings and deteriorate financial sector balance sheets. Movements in exchange rates between major currencies may also inflict deterioration in balance sheets through adverse effects on price competitiveness and export growth. Overall, such common external shocks adversely affect economic fundamentals in a number of countries simultaneously, potentially triggering crises in some of them. In fact, for each of the major crises over the last two decades one can easily identify a common shock in the run up to the crisis. The Latin American debt crisis of 1982 and the Mexican crisis of 1994 were both preceded by a sharp increase in US interest rates. A significant appreciation of the D-Mark starting in the mid-1980s paved the way for the ERM crisis of 1992–93. The Asian crisis of 1997 followed a sharp appreciation of the US dollar against the yen, persistent economic stagnation in Japan, and a steep decline in semiconductor prices. Masson (1999b) proposes the term ‘monsoonal effects’ rather than contagion for common shocks that result from such economic shifts. Others tend to generalize the term contagion to any kind of external shock that includes common shocks (Summers 2000, p. 6). However, this is misleading. Contagion should be confined to describe situations in which a crisis in one country causes crises in other countries, or at least makes them more likely. Though common shocks and contagion are both a source of systemic risk, they are different phenomena.

B. Interdependence of Fundamentals Even confined to crises that are causally connected, it is controversial whether any transmission of country specific shocks should be considered contagion. In particular, interdependence among countries arises because their economic fundamentals are linked through the balance of payments. A crisis in one country can affect another’s fundamentals. A well-known channel of such shock propagation is international trade. A crisis in one country, usually associated with economic contraction and exchange rate devaluation, negatively affects exports of trading partners through a fall in demand and a loss in price competitiveness. Such trade related shock propagation does not only work through bilateral trade links, but also through indirect trade links resulting from common third markets. r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

161

In the recent Argentine crisis, the devaluation of the peso in early 2002 followed by a severe economic contraction had clear repercussions for Mercosur, the trade bloc including Argentina, Brazil, Paraguay and Uruguay. It is no coincidence that the other three members approached a financial crisis in 2002, particularly Uruguay with Argentina its second largest trading partner. In turn, Brazil’s devaluation in 1999 was an important reason for the protracted period of contraction that Argentina and Uruguay suffered in the run up to the crisis. Kindleberger (1985, p. 227) has argued that such trade propagation mechanisms are ‘too strung out with lags to explain the near simultaneity of y crises’, but agents in financial markets respond in anticipation of real impacts causing immediate corrections in asset prices. However, asset prices and capital flows do not react only to anticipated trade shocks. Independent financial linkages can be either bilateral direct cross-border investments, or indirect through common investors. As an example of a bilateral linkage, the Argentine crisis of 2002 led many Argentines to withdraw dollar deposits held in Uruguay, causing banks in Uruguay to withdraw required reserves and accelerate the decline in central bank reserves, finally leading to the floating of the peso in June 2002.3 Regarding indirect links, international investors including banks and institutional investors such as mutual funds, hedge funds and pension funds, diversify their portfolios over financial assets of numerous countries. Even if they hold only a small fraction of their portfolio in emerging markets, these fractions can represent a significant proportion of the relatively small capital markets in those countries, and portfolio rebalancing may cause significant capital flows (Kaminsky et al. 2001). Several authors have pointed out that international investors could undertake such portfolio rebalancing in response to a crisis in one country, and thereby put other emerging economies in trouble by selling their assets or calling loans. Schinasi and Smith (2001) show how cross-country transmission of shocks can be explained by basic portfolio theory as an optimal portfolio rebalancing response to a financial crisis in one country. They show that an adverse shock to a single asset’s return distribution can lead to a reduction in other risky asset positions, depending on the portfolio management rule and the parameterization of the joint distribution of asset returns. In the presence of leverage, the optimal response to a realized loss on a specific position is to deleverage and reduce risky asset positions in all markets. 3

However, the withdrawal of deposits by Argentines also contained a significant element of ‘domino effects’. Many Argentines were forced to raise liquidity due to the fact that their government imposed a freeze on withdrawals from domestic banks, the corralito.

r Blackwell Publishing Ltd. 2003

162

Thomas Moser

Overall, their analysis suggests that, under plausible conditions, a crisis in one market can generate substantial reductions in the optimal positions in non-crisis markets in the move to a new equilibrium. In the literature on financial contagion, there is yet little convergence of views about whether cross-country propagation of shocks through fundamentals should be considered contagion. A number of authors call for discrimination between ‘pure contagion’ and shock propagation through fundamentals, which they suggest labelling ‘transmission’ (Bordo et al. 1998), ‘spillovers’ (Masson 1999b), ‘interdependence’ (Forbes and Rigobon 2002) or ‘fundamentals-based contagion’ (Kaminsky and Reinhart 2000). The main reason for discrimination is that shock propagation through fundamentals is the result of an optimal response to external shocks. As a crisis in one country upsets the equilibrium in other countries, real and financial variables adjust to a new equilibrium. Financial market responses only reflect (anticipated) changes in fundamentals and speed up adjustment to the new equilibrium, but they do not cause the change in equilibrium. In other words, rather than causing the crisis, financial market responses bring the crisis forward.

III. Mechanisms of Pure Contagion To explain cross-country propagation of shocks not related to or explained by economic fundamentals, one has to resort to market imperfections. I will divide the different explanations into information effects and domino effects. Information effects involve a crisis in one country that triggers a crisis in others as agents take the crisis as a signal to update information. In a domino effect, a crisis in one country leads to crises in others because of financial connectivity.

A. Information Effects Information imperfections and costs of acquiring and processing information make a correct assessment of fundamentals difficult and a certain degree of ignorance rational. As a result, market participants are uncertain about the true state of a country’s fundamentals. A crisis elsewhere might lead them to reassess the fundamentals of other countries and cause them to sell assets, to call in loans, or to stop lending to these countries, even if their fundamentals remain objectively unchanged. The literature offers a number of reasons why a crisis elsewhere could lead to a reassessment of objectively unchanged fundamentals. r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

163

Signal extraction failures One group of explanations follows the framework adopted by King and Wadhwani (1990). The basic idea is that due to incomplete information, market participants are uncertain about the relevance of a financial crisis in one country for the fundamentals of another country. There are two possible stories of how this could result in contagion. In what could be called ‘fictitious interdependence’, market participants falsely assume interdependence of fundamentals or at least overestimate the extent of interdependence. As a result of the former, a country specific shock that would have no effect on the other country in the case of full information would nevertheless be transmitted. As a result of the latter, the propagation would be excessive relative to a situation with full information (Pritsker 2001). Second, in what could be called the ‘lump together hypothesis’, market participants assume that countries considered ‘similar’ to the crisis country face potentially similar problems. For instance, investors could misinterpret a country specific shock as a common shock for a number of countries. Alternatively, lumping several countries together could result from some form of statistical discrimination, insofar as the problems that cause a crisis in one country are thought to reveal a characteristic of a wider group of countries. In both cases of signal extraction failures, fictitious interdependence or lumping together, a crisis in one country leads to an inefficient revision of fundamentals and a less accurate assessment. This inaccuracy would give better-informed investors scope for profits. However, one could argue that the resulting short-run pressures suffice to lead the countries concerned into trouble. Wake-up call A different story as to why a crisis in one country could lead to a reassessment of objectively unchanged fundamentals in other countries is what Goldstein (1998, p. 18) calls the ‘wake-up call’ hypothesis. A crisis in one country may serve as a ‘wake-up call’ for market participants if it causes them to take a closer look at fundamentals similar to those in the crisis country. Contagion occurs if this leads them to detect problems or risks they failed to see before.4 The difference to the former explanation is that, in the signal extraction story, the initial crisis makes market participants assume problems that do not exist, whereas, with a wake-up call, it makes them aware of existing problems. This sort of contagion is the result of an efficient correction and leads towards a more accurate assessment of fundamentals. 4

See Goldstein (1998, p. 18): ‘I refer to it as a wake-up call because to judge from most market indicators of risk, private creditors and rating agencies were asleep prior to the outbreak of the Thai crisis.’

r Blackwell Publishing Ltd. 2003

164

Thomas Moser

Expectations interaction This line of argument takes a crowd psychology approach and employs a form of ‘mental contagion’ in the tradition of the classical financial panic view, depicting financial crises as self-fulfilling expectations phenomena or coordination failures among agents in the presence of multiple equilibria. Contagion arises if a crisis elsewhere serves as the ‘sunspot variable’ that leads agents to co-ordinate their expectations on crisis equilibria. A crisis in one country may cause the loss of public confidence in financial markets or international investments in general, setting off runs in other countries.5 Alternatively, such a story would be consistent with Bayesian updating in which market participants raise the probability they assign to the occurrence of a crisis in similar circumstances on the basis of drawing another favourable observation (Drazen 2000). There are two frameworks for modelling self-fulfilling financial crises, one along the lines of the Diamond–Dybvig (1983) model of bank runs, the other along the lines of the Obstfeld (1986) model of speculative attacks. Chang and Velasco (2001) and Marshall (1998) apply the Diamond–Dybvig approach to countries, and the cause for financial crises are simultaneous liquidity withdrawals, not driven by economic fundamentals, which result in costly liquidations and asset price collapses. Two equilibria arise when a country’s financial system has potential short-term obligations in foreign currency that exceed the amount of foreign currency it can access on short notice. The crisis equilibrium occurs if international creditors stop lending and withdraw funds for fear that all other will do so, forcing the financial system to liquidate potentially profitable investments at a loss. Alternatively, in the Obstfeld (1986) model, the multiplicity of equilibria arises because expectations feed back in an adverse way into the fundamentals, eventually inducing the government to change its policy. For instance, expectations of devaluation affect aggregate demand, debt servicing costs and bank balance sheets through higher interest rates. Consequently, a country will abandon the exchange rate peg if coordinated behaviour of speculators raises the costs of maintaining the peg sufficiently high. Common to both approaches is the implication that such crises are apparently unnecessary as they are not determined by underlying fundamentals, but every agent will find it individually advantageous to participate in an expected run or attack. Both frameworks, however, require initial fundamentals to be in a range of ‘discomfort’. Consequently, while self-fulfilling crises seem possible, not just any country can be subject to such a crisis. The Obstfeld model requires a sufficiently unfavourable initial state of the 5

Note that contagion across different markets or institutions requires more than herding behaviour with respect to one market or institution, explaining the notion ‘in general’.

r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

165

economy to make adverse market sentiment possible and the country vulnerable to the cost it may cause. In the Diamond–Dybvig approach, the multiple equilibrium phenomenon is essentially related to liquidity risk, a risk that is evident for banks due to the unique characteristics of their balance sheet and deposit contracts, but not imperative for countries. To construct an international financial contagion story, however, one could still apply the original Diamond–Dybvig approach confined to the banking industry. For instance, when in late April 2002 residents in Uruguay started to join Argentines in withdrawing dollar deposits, some saw this as a matter of pure contagion across banking institutions. However, considering the unresolved problems in some banks in Uruguay, the signs of a deepening recession, the decline in international reserves and the rising doubts about the government maintaining the exchange rate peg, there was good reason for demanding immediate withdrawal. More generally, there is not much evidence that depositor runs have driven solvent banks into insolvency, not even during the exemplary banking panic of the Great Depression (Calomiris and Mason 1997). Moral hazard plays Dooley (2000) offers an explanation for contagion where a crisis in one country leads investors to revise their bail-out expectations. This line of argument links contagion to moral hazard, as expected insurance against losses is an important factor in generating capital flows in the first place. Investors overlook weak fundamentals as long as reserves plus the amount of potential assistance by the international community is thought to be large enough to insure against losses. Contagion may occur if a crisis elsewhere leads investors to revise potential international assistance below some critical level where the expected liquidation value of the investments net of insurance is less than the value of claims. Such a downward revision could occur whether or not the crisis country receives official lending. If it does, this will reduce potential lending available for bail-outs for any other country, as the overall size of funds available for bail-outs is likely to be limited. Conversely, if the country is not rescued (e.g., Russia in 1998), investors may interpret this as a change in the crisis management policy of the international community and revise downward the expected value of the overall size of funds available for bail-outs. In either case, contagion would mean that investors bring their positions closer into line with economic fundamentals.

Membership contagion Drazen (2000) provides an explanation for contagion in currency crises that focuses on the actions of policy makers. Countries with fixed exchange rates r Blackwell Publishing Ltd. 2003

166

Thomas Moser

can be thought of as members of a ‘no-devaluation club’ with the value of membership depending positively on who else is a member. Such a club can be defined explicitly (e.g., the Exchange Rate Mechanism (ERM) of the EU) or implicitly if politicians attach weight to being ‘lumped together’ with successful neighbour countries that have fixed exchange rates. Membership contagion may then result if the devaluation by one ‘club member’ weakens the membership criterion and makes other members more likely to devalue because governments assign a lower value to the commitment. One could also broaden the story and argue that governments, sacrificing domestic economic goals to maintain a fixed exchange rate, can assume market participants take a more favourable view of a devaluation if it follows in the wake of other devaluations, particularly if contagion can be blamed. Contagion would then result from the lower costs of breaking a commitment if others do so.

B. Domino Effects In this group of explanations, a crisis in one country spreads to others rather mechanically in domino fashion as a result of direct or indirect financial connections. This story comes in three variations.

Counterparty defaults The most straightforward version is the one of ‘cascading defaults’ and is a matter of ordinary credit risk (Marshall 1998). High exposure to troubled private or sovereign debtors may lead to cross-border transmission of failures if losses for creditors are large enough. This is particularly likely if banks are among the major creditors, as they operate with relatively low capital-to-asset ratios. Accordingly, a country could experience a banking crisis if major banks suffer large losses from defaults on foreign loans or cross-border settlements. Concerns over loan defaults were one of the main motivations for the international rescue operations during the Latin American debt crisis of 1982, with claims on troubled debtors heavily concentrated in the hands of few leading banks.6 A famous example for cross-border settlement failure is the closure of the Herstatt Bank in Germany in 1974, which occurred during the business day in New York and led to counterparty losses for banks that had already fulfilled their side of foreign exchange transactions with Herstatt. 6

See, for example, Fischer (1987, p. 166). In 1982, the nine large US money centre banks had more than 250% of their capital in loans to less developed countries.

r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

167

Portfolio rebalancing due to liquidity constraints A number of authors have pointed out that liquidity or capital constraints could impose greater than optimal asset reduction on international investors affected by a crisis in one market, forcing them to unwind positions in other markets to raise liquidity. This unwinding could occur if investors need to meet margin calls or other collateral requirements, or fulfil investor redemptions in the case of mutual funds. As originally suggested by Valde´s (1997), liquidity needs could also arise if market liquidity sharply declines or virtually dries up because a market maker or important market participant suffers large losses and withdraws funds, or if several participants unwind similar positions at the same time. If market participants rely on such a market for funding or hedging, they might be forced to cover by selling assets in unrelated liquid markets. Such concerns seem to have been a factor in the Federal Reserve’s intervention to prevent forced liquidation of the hedge fund Long Term Capital Management (LTCM) in Autumn 1998 (Greenspan 1998). Anecdotal evidence also suggests that, in the financial market turbulence following the Russian crisis during the Autumn of 1998, market liquidity dried up temporarily even in such deep and mature markets as the US repo market and the yen/dollar currency market (IMF 1998, p. 51). It has also been reported that, in the wake of the 1998 Russian crisis, international investors put short positions in the relatively deep markets for Hungarian and Brazilian debt in place to hedge against long positions in Russian securities (CGFS 1999). As pointed out by Masson (1999a), explaining the persistence of such effects requires explaining why other investors, seeing the fundamentals unchanged, do not take advantage of the buying opportunity created by a few institutions in trouble. Calvo (1998) offers that a ‘lemons problem’ could arise because of information asymmetries. If buyers cannot distinguish whether sellers sell because of liquidity constraints or because they know something about the fundamentals, investors in need for liquidity can sell assets only at firesale prices.

Portfolio rebalancing due to capital constraints Capital constraints are another factor that could impose greater than optimal asset reduction on investors affected by a crisis. Particularly, capital requirements that induce banks to adjust their capital ratios might cause them to cut back foreign loans or, in the case of ratios to risk-weighted assets, shift into low-risk assets such as government securities. A bank could also respond by improving the numerator in the capital–asset ratio, perhaps by issuing equity or by retaining earnings, although this might be difficult in times of crisis. Evidence that this channel can be economically significant r Blackwell Publishing Ltd. 2003

168

Thomas Moser

even between advanced countries has been provided by Peek and Rosengren (1997) who show that risk-based capital requirements associated with the Japanese stock market decline resulted in a significant decrease in lending by Japanese banks in the USA in the early 1990s. Peek and Rosengren (2000) further show that this loan supply shock had real effects on economic activity in the USA in that it was not compensated by lending from other sources.

IV. Does Contagion Matter? While the above discussion suggests that pure contagion is possible, to be a matter of political concern it has to be empirically relevant. The empirical research on contagion seems to arrive at the following conclusion. First, while common shocks and interdependence of fundamentals, particularly trade links, explain a large part of the simultaneous occurrence of financial crises, the empirical role for pure contagion seems to be relatively limited. Second, regarding this limited role, empirical evidence points to the existence of portfolio rebalancing effects.

A. Testing for Contagion Measuring contagion encounters empirical difficulties (Rigobon 2002). One problem is that financial contagion is associated with high frequency events while, for most economic fundamentals, high frequency data is not available, making it difficult to control for fundamentals. A widely used approach to test for contagion is to study cross-market correlation of financial variables and to define contagion as a significant increase in correlation coefficients during the crisis period.7 Several of the initial studies that applied the correlation approach to the crises of the 1990s found that cross-market correlation did increase significantly during these crises. However, Forbes and Rigobon (2002) pointed out that correlation coefficients are biased in the presence of heteroscedasticity, overstating the magnitude of correlation when market 7

The logic of this approach can be shown in a simple framework following Forbes and Rigobon (2002). The financial market variable of interest (exchange rate, interest rate, stock prices) of country i, yi,t, can be described by yi,t 5 aixt1bizt1ei,t, where x is a vector of the respective variable in countries other than i, z is a variable that captures common shocks, and e represents country-specific shocks. Vector a measures the impact of country-specific shocks elsewhere on fundamentals of country i (interdependence) and vector b the effect of common shocks on country i. Contagion is captured by a change in either ai or bi after a shock to an other country, as reflected by a change in the correlation between yi,t, and xt.

r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

169

volatility increases during a crisis.8 Based on an extensive set of tests, they find that, if correlation coefficients are adjusted for heteroscedasticity, there is virtually no evidence for contagion during recent crises. Since they find considerable cross-market linkages in all periods, their analysis suggests that most shocks are common or transmitted across markets through interdependence of fundamentals. Some recent studies have found somewhat more favourable evidence for pure contagion even after correcting for the bias described by Forbes and Rigobon (2002), but they do not establish a strong case for contagion.9 Baig and Goldfajn (1999) study cross-market correlation during the Asian crisis with respect to foreign exchange, equity, interest rates and sovereign debt markets, and find a significant increase in correlation only in foreign debt markets, whereas the evidence in stock markets is mixed at best. They further try to distinguish co-movements driven by fundamentals and nonfundamentals by creating a set of daily variables constructed from news reported by the press that proxy for movements in fundamentals. Testing for impacts of cross-border news on the markets after controlling for owncountry news and other fundamentals, they find evidence for such impact in the currency and equity markets. Gelos and Sahay (2001) employ a VAR framework to study daily stock and exchange market data during the 1997 Czech crisis, the Asian crisis and the 1998 Russian crisis, and find that Russian stock returns ‘Granger caused’ returns in European transition economies during the Russian crisis, while this was usually not the case during tranquil times. They find no evidence for an increase in heteroscedasticity-adjusted correlations in stock markets, but do so in exchange markets. Bordo and Murshid (2001) examine cross-country correlations in bond prices and interests rates during seven international financial crises from the pre World War I era to the Asian crisis. After adjusting for heteroscedasticity, they find only a very small number of isolated cases of pure contagion. The conclusions of Forbes and Rigobon (2002) coincide with Moreno and Trehan (2000) who analyse the extent to which common external shocks alone can explain simultaneous currency crises. Analysing a sample of 121 countries over the 1974–97 period, Moreno and Trehan (2000) find that common external shocks can explain close to 75% of the variation in the total number of crisis over time. In fact, according to their findings, changes 8

The intuition behind this is that, due to an increase in the variance of xt, a greater proportion of the fluctuations in yi,t is explained by xt and, therefore, estimated correlation between yi,t and xt increases, even if the value of a remains unchanged.

9

For a critical discussion of the Forbes–Rigobon approach, see also Corsetti et al. (2001) who find the Forbes–Rigobon test to be biased towards the null hypothesis of interdependence.

r Blackwell Publishing Ltd. 2003

170

Thomas Moser

in US interest rates and US inflation alone explain about half of the variation in the aggregate number of crises over time. Overall, their results suggest that common shocks are responsible for a large fraction of the simultaneous occurrence of currency crises.

B. Testing for Specific Propagation Mechanisms A number of studies attempt to disentangle the different mechanisms through which shocks are propagated. A general finding is the importance of trade links for crisis transmission, coinciding with the result that the driving force is interdependence of fundamentals rather than pure contagion. Eichengreen et al. (1996) estimate a probit model with a panel of data for 20 industrial countries for the period 1959–93 and define contagion as a crisis in one country increasing the probability of a crisis occurring in other countries. After finding evidence of such contagion, they discriminate between different channels of transmission by weighting crises elsewhere by the importance of trade and by the similarity of macroeconomic policies and outcomes. They find the trade weighted measure to be significant and conclude that trade links rather than revisions of expectations have been the dominant channel of crisis propagation. This result is confirmed by Glick and Rose (1999) who estimate a probit equation across countries to test for the importance of direct and indirect trade links in five episodes of currency crises from the breakdown of the Bretton Woods system in 1971 to the Asian crisis in 1997–98. They find that currency crises mainly spread along trade links. Gelos and Sahay (2001) find that composite indices of exchange market pressure are moderately correlated across European transition economies during the 1990s, a correlation that can partly be explained by direct trade linkages but not measures of other fundamentals. De Gregorio and Valde´s (2001) examine the performance of alternative crisis indicators as a function of initial macroeconomic conditions and a weighted average of crisis indicators. Alternative weighting schemes are used to capture different channels of propagation. They find that, during the 1982 debt crisis and the Mexican and Asian crises of the 1990s, trade links were very important, although they cannot account for the entire regional effect. Forbes (2002) obtains a similar result by decomposing the trade channel into separate competitiveness and income effects and utilizing industry level trade data rather than aggregate trade flows. Examining a sample of 58 countries during 16 crises from 1994 through 1999, she finds that countries competing with exports from a crisis country (competitiveness effect) and exporting to the crisis country (income effect) had significantly lower stock market returns. However, these trade effects explain only a portion of stock market r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

171

movements during crises, perhaps pointing to other channels such as portfolio rebalancing. These conclusions also hold at the firm level. Forbes (2000) looks at firmlevel data including 46 countries with more than 10,000 companies to test how the Asian and the Russian crises affected individual company’s stock market returns. Estimating the effect of different firm characteristics on stock market returns during the crisis, Forbes confirms the importance of both direct and indirect trade links as transmission channels. Additionally, she finds that, during the later part of the Asian crisis, more liquid stocks had significantly lower returns than the rest of the sample and suggests that this result points to transmission through portfolio effects. Some recent studies find evidence for portfolio rebalancing effects, particularly with regard to bank lending.10 These studies, however, do not discriminate between shock propagation resulting from an optimal portfolio response and propagation resulting from liquidity or capital constraints. Additionally, trade and bank lending linkages tend to be highly correlated, making it difficult to distinguish between the two. Kaminsky and Reinhart (2000) estimate conditional probabilities that a crisis will occur in a given country for 80 currency crises and test for a common lender channel of shock propagation. First, they group countries in accordance with their exposure to US banks and Japanese banks respectively to test for propagation through foreign bank lending. Second, they form two clusters of countries that exhibit a high degree of historic co-movement in their asset returns to capture cross-hedging strategies and test for propagation through globally diversified portfolios. They find that both types of clusters add information in that a crisis in the respective cohort significantly increases the conditional probability of a crisis, suggesting evidence for crisis propagation through portfolio rebalancing. Limited to bank lending (but also including European banks), Van Rijckeghem and Weder (2001) test the importance of a ‘common bank lender’ channel of contagion relative to trade links during the Mexican, Asian and Russian crises of the 1990s. For each crisis episode, they run regressions with a number of alternative crisis indicators and trade linkages, competition for bank funds, and a set of macro fundamentals as independent variables, and find that contagion through common bank lenders was important. Caramazza et al. (2003) employ panel probit estimation and construct a trade-link variable and common creditor variables to examine the occurrence of currency pressures during the major crisis episodes of the 10

In fact, Froot et al. (2001), exploring data on daily international portfolio flows from 1994 through 1998, find that institutional investors did not abandon emerging equity markets during the Mexican or the Asian crisis.

r Blackwell Publishing Ltd. 2003

172

Thomas Moser

1990s. They find that financial linkages with the major creditor of the crisis country appears to substantially raise the probability of a crisis. Rather than looking at prices, Van Rijckeghem and Weder (2000) test the common lender effect by looking at volumes, studying panel data on bank flows to a subset of 30 emerging markets disaggregated by 11 creditor countries. Their regressions point to the existence of a common lender effect in the Mexican and Asian crises of the 1990s. For the Russian crisis, they find a more generalized withdrawal of funds, interpreted as an increase in perceived risk or a general increase in risk aversion.

V. Policy Implications While much goes by the name of financial contagion, empirical research suggests common shocks rather than contagion explain an important part of the simultaneous occurrence of financial crises. As to the other part, there are good reasons for discriminating between shock propagation through fundamentals and shock propagation unrelated to fundamentals (pure contagion). Although the role of pure contagion seems to be rather limited, this paper identifies eight potential mechanisms that differ sharply in causes and implications. For instance, contagion may equally stem from an inefficient reassessment of fundamentals or from efficient correction toward a more accurate assessment. Policy makers had better take these differences into account. A case in point is the call for an international lender of last resort as a defence against financial contagion (Fischer 1999b). Some authors have pointed out that policy implications differ with regard to common shocks and interdependence of fundamentals versus pure contagion (Chang and Majnoni 2001; Forbes and Rigobon 2002; Baig and Goldfajn 1999). They conclude that liquidity assistance is an appropriate response only to pure contagion, while common shocks and interdependence, matters of economic fundamentals, have to be dealt with through policy measures that seek to improve the fundamentals. Otherwise, liquidity support might only delay necessary adjustment. This paper suggests that, even in the case of pure contagion, the advisability of short-term liquidity assistance depends on the specific contagion mechanism. If contagion is channelled through information effects, liquidity assistance is an appropriate response only in the case of signal extraction failures and expectations interactions. With signal extraction failures, if market participants either misjudge interdependence or falsely lump apparently similar countries together, it is reasonable to assume such contagion will only give rise to temporary market pressure either because r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

173

participants learn about the fundamentals or better-informed participants take advantage of profit opportunities. In the second case, if expectations interaction triggers a self-fulfilling crisis, liquidity assistance stands a good chance of success. If the country is vulnerable to self-fulfilling runs because of international illiquidity, liquidity assistance is just what the doctor ordered and the prospect of assistance might even stabilize expectations sufficiently to eliminate self-fulfilling runs in the first place. If the country is vulnerable to adverse market sentiment because of its initial state of fundamentals, the availability of additional foreign reserves at relatively low cost at least increases the critical size of the attack that is required to be successful (Disyatat 2001). As to the remaining information effects, liquidity assistance is less promising. In the case of membership contagion, availability of additional reserves at low cost does influence the government decision, but hardly enough to substantially weaken the temptation to take advantage of the ‘sharing-the-blame’ effect so as to solve a conflict between the fixed exchange rate and other domestic goals. A case in point is the ERM crisis of 1992–93, where mutual liquidity assistance was available. Wake-up calls, where participants detect previously unnoticed problems, and moral hazard plays, where participants overlook problems as long as insurance is thought to be large enough, are matters of fundamentals; problems that temporary liquidity assistance are unable to solve. Liquidity assistance would even further spur the moral hazard play. In the case of domino effects, results are just as mixed. With counterparty defaults, the problem is solvency not liquidity. Accordingly, the problem has to be dealt with through measures that re-establish solvency. In the case of portfolio rebalancing effects, the reason for rebalancing matters. If the crosscountry transmission of the shock is due to liquidity constraints, the capital outflow pressure may be a short-term matter as investors return after liquidity constraints are lifted and short-term financing could help weather the storm. If portfolio rebalancing is due to capital constraints, the nature of the shock is more likely to be longer term as it takes more time to raise capital than to raise liquidity. Given that the simultaneous occurrence of financial crises seems to be mainly a matter of common shocks and interdependence of fundamentals, and that what is left for pure contagion seems to hint at portfolio rebalancing, international liquidity assistance alone should not be expected to cure what is perceived, rightly or wrongly, as international financial contagion. Thomas Moser Advisor to Executive Director IMF, OEDSZ r Blackwell Publishing Ltd. 2003

174

Thomas Moser

700 19th Street, NW, Washington, D.C. 20431 USA [email protected].

References Bagehot, Walter ([1873] 1962), Lombard Street: A Description of the Money Market. Homewood, IL: Irwin. Baig, Taimur, and Ilan Goldfajn (1999), ‘Financial Market Contagion in the Asian Crisis’, IMF Staff Papers, 46, 167–95. Baumol, William J. (1994), ‘Multivariate Growth Patterns: Contagion and Common Forces as Possible Sources of Convergence’, in W. Baumol, W. Wolff and R. Nelson (eds), Convergence of Productivity: Cross-National Studies and Historical Evidence, Oxford: Oxford University Press, 62–85. Bordo, Michael, and Antu Murshid (2001), ‘Are Financial Crises Becoming Increasingly More Contagious? What is the Historical Evidence on Contagion?’, in S. Claessens and K. Forbes (eds), International Financial Contagion, Boston, MA: Kluwer Academic Publishers, 367–405. Bordo, Michael, Bruce Mizrach and Anna J. Schwartz (1998), ‘Real versus Pseudointernational Systemic Risk: Some Lessons from History’, Review of Pacific Basin Financial Markets and Policies, 1, 31–58. Calomiris, Charles, and Joseph Mason (1997), ‘Contagion and Bank Failures During the Great Depression: The June 1932 Chicago Banking Panic’, American Economic Review, 87, 863–83. Calvo, Guillermo A. (1998), ‘Understanding the Russian Virus, with Special Reference to Latin America’, Mimeo, University of Maryland. Calvo, Guillermo A., Leonardo Leiderman and Carmen M. Reinhart (1996), ‘Inflows of Capital to Developing Countries in the 1990s’, Journal of Economic Perspectives, 10, 123–39. Caramazza, Francesco, Ricci Luca and Ranil Salgado (2003), ‘International Financial Contagion in Currency Crises’, Journal of International Money and Finance, forthcoming. Chang, Roberto, and Giovanni Majnoni (2001), ‘International Contagion: Implications for Policy’, in S. Claessens and K. Forbes (eds), International Financial Contagion, Boston, MA: Kluwer Academic Publishers. Chang, Roberto, and Andre´s Velasco (2001), ‘A Model of Financial Crises in Emerging Markets’, Quarterly Journal of Economics, 116, 489–517. Claessens Stijn, and Kristin Forbes eds, (2001), International Financial Contagion. Boston, MA: Kluwer Academic Publishers. r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

175

Claessens, Stijn, Rudiger Dornbusch and Yung Chul Park (2001), ‘Contagion: Why Crises Spread and How This Can Be Stopped’, in S. Claessens and K. Forbes (eds), International Financial Contagion, Boston, MA: Kluwer Academic Publishers. Committee on the Global Financial System (1999), A Review of Financial Markets Events in Autumn 1998. Basel: Bank for International Settlements. Corsetti, G., M. Pericoli and M. Sbracia (2001), ‘Correlation Analysis of Financial Contagion: What One Should Know before Running a Test’, Discussion Paper 822, Economic Growth Center, Yale University. De Gregorio, Jose´, and Rodrigo O. Valde´s (2001), ‘Crisis Transmission: Evidence from the Debt, Tequila, and Asian Flu Crises’, World Bank Economic Review, 15, 289–314. Demirgu¨c¸-Kunt, A., and E. Detragiache (1998), ‘The Determinants of Banking Crises in Developing and Developed Countries’, IMF Staff Papers, 45, 81–109. Detragiache, E., and A. Spilimbergo (2001), ‘Crises and Liquidity: Evidence and Interpretation’, IMF Working Paper No. 01/2. Diamond, Douglas W., and Philip H. Dybvig (1983), ‘Bank Runs, Deposit Insurance, and Liquidity’, Journal of Political Economy, 91, 401–19. Disyatat, Piti (2001), ‘Currency Crises and Foreign Reserves: A Simple Model’, IMF Working Paper No. 01/18. Dooley, Michael P. (2000), ‘A Model of Crises in Emerging Markets’, Economic Journal, 110, 256–72. Drazen, Allen (2000), ‘Political Contagion in Currency Crises’, in P. Krugman (ed.), Currency Crises, Chicago and London: University of Chicago Press, 47–67. Eichengreen, Barry, Andrew Rose and Charles Wyplosz (1996), ‘Contagious Currency Crises’, Scandinavian Economic Review, 98, 463–84. Feldstein, Martin (1999), ‘A Self-Help Guide for Emerging Markets’, Foreign Affairs, 78, March/April, 93–109. Findlay, Ronald (1978), ‘Relative Backwardness, Direct Foreign Investment, and the Transfer of Technology: A Simple Dynamic Model’, Quarterly Journal of Economics, 92, 1–16. Fischer, Stanley (1987), ‘Sharing the Burden of the International Debt Crisis’, American Economic Review, Papers and Proceedings, 77, 165–70. Fischer, Stanley (1999a), ‘Reforming the International Financial System’, Economic Journal, 109, F557–76. Fischer, Stanley (1999b), ‘On the Need for an International Lender of Last Resort’, Journal of Economic Perspectives, 13, 85–104. Forbes, Kristin (2000), ‘The Asian Flu and Russian Virus: Firm-level Evidence on How Crises are Transmitted Internationally’, NBER Working Paper 7807. r Blackwell Publishing Ltd. 2003

176

Thomas Moser

Forbes, Kristin (2002), ‘Are Trade Linkages Important Determinants of Country Vulnerability to Crises?’, in S. Edwards and J. A. Frankel (eds), Preventing Currency Crises in Emerging Markets, Chicago: The University of Chicago Press, 77–124. Forbes, Kristin, and Roberto Rigobon (2002), ‘No Contagion, Only Interdependence: Measuring Stock Market Comovements’, Journal of Finance, 57, 2223–62. Froot, Kenneth, Paul O’Connell and Mark Seasholes (2001), ‘The Portfolio Flows of International Investors’, Journal of Financial Economics, 59, 151–93. Gelos, Gaston, and Ratna Sahay (2001), ‘Financial Market Spillovers in Transition Economies’, Economics of Transition, 1, 53–86. Glick, Reuven, and Andrew K. Rose (1999), ‘Contagion and Trade: Why are Currency Crises Regional?’, Journal of International Money and Finance, 18, 603– 17. Goldstein, Morris (1998), The Asian Crisis: Causes, Cures, and Systemic Implications. Washington: Institute for International Economics. Greenspan, Alan (1998), ‘Testimony before the Committee on Banking and Financial Services’, US House of Representatives, 1 October. Available at www.federalreserve.gov. International Monetary Fund (1998), World Economic Outlook and International Capital Markets: Interim Assessment. Washington: International Monetary Fund. International Monetary Fund (1999), World Economic Outlook: May 1999. Washington: International Monetary Fund. Kaminsky, Graciela, and Carmen Reinhart (1999), ‘The Twin Crises: The Causes of Banking and Balance-of-Payment Problems’, American Economic Review, 89, 473– 500. Kaminsky, Graciela, and Carmen Reinhart (2000), ‘On Crises, Contagion, and Confusion’, Journal of International Economics, 51, 145–68. Kaminsky, Graciela, Richard Lyons and Sergio Schmukler (2001), ‘Mutual Fund Investment in Emerging Markets: An Overview’, World Bank Economic Review, 15, 315–40. Kindleberger, Charles (1985), ‘International Propagation of Financial Crises: The Experience of 1888–93’, in Charles Kindleberger Keynesianism vs. Monetarism and Other Essays in Financial History, London: Allen and Unwin, 226–39. King, Mervyn, and Sushil Wadhwani (1990), ‘Transmission of Volatility between Stock Markets’, Review of Financial Studies, 3, 5–33. Mack, R. P., and V. Zarnowitz (1958), ‘Causes and Consequences of Changes in Retailers’ Buying’, American Economic Review, 48, 18–49. Marshall, David (1998), ‘Understanding the Asian Crisis: Systemic Risk as Coordination Failure’, Federal Reserve Bank of Chicago, Economic Perspectives, 22, 13–27. r Blackwell Publishing Ltd. 2003

What Is International Financial Contagion?

177

Masson, Paul (1999a), ‘Multiple Equilibria, Contagion, and the Emerging Market Crises’, IMF Working Paper No. 99/164. Masson, Paul (1999b), ‘Contagion: Monsoonal Effects, Spillovers, and Jumps Between Multiple Equilibria’, in Pierre-Richard Agenor, Marcus Miller, David Vines and Axel Weber (eds), The Asian Financial Crisis: Causes, Contagion and Consequence, Cambridge, UK: Cambridge University Press, 265–80. Mavor, James (1891), ‘The Scottish Railway Strike’, Economic Journal, 1, 204–19. Mishkin, Frederic (1997), ‘Understanding Financial Crises: A Developing Country Perspective’, in M. Bruno and B. Pleskovic (eds), Annual World Bank Conference on Development Economics, 1996, Washington: World Bank, 29–62. Moreno, Ramon, and Bharat Trehan (2000), ‘Common Shocks and Currency Crises’, Federal Reserve Bank of San Francisco Working Paper No. 05. Murchison, C. T. (1933), ‘Requisites of Stabilization in the Cotton Textile Industry’, American Economic Review, Papers and Proceedings, 23, 71–80. Obstfeld, Maurice (1986), ‘Rational and Self-Fulfilling Balance-of-Payments Crises’, American Economic Review, 76, 72–81. Peek, Joe, and Eric Rosengren (1997), ‘The International Transmission of Financial Shocks: The Case of Japan’, American Economic Review, 87, 495–505. Peek, Joe, and Eric Rosengren (2000), ‘Collateral Damage: Effects of the Japanese Bank Crisis on Real Activity in the United States’, American Economic Review, 9, 30–44. Price, L. L. (1890), ‘The Position and Prospect of Industrial Conciliation’, Journal of the Royal Statistical Society, 53, 420–59. Pritsker, Matt (2001), ‘The Channels for Financial Contagion’, in S. Claessens and K. Forbes (eds), International Financial Contagion, Boston, MA: Kluwer Academic Publishers, 67–97. Rigobon, Roberto (2002), ‘Contagion: How to Measure it?’, in S. Edwards and J. A. Frankel (eds), Preventing Currency Crises in Emerging Markets, Chicago: The University of Chicago Press, 269–334. Schinasi, Garry J., and R.Todd Smith (2001), ‘Portfolio Diversification, Leverage, and Financial Contagion’, in S. Claessens and K. Forbes (eds), International Financial Contagion, Boston, MA: Kluwer Academic Publishers, 187–223. Summers, Lawrence (2000), ‘International Financial Crises: Causes, Prevention, and Cures’, American Economic Review, Papers and Proceedings, 90, 1–16. Thornton, Henry ([1802] 1978), An Enquiry into the Effects of the Paper Credit of Great Britain. Fairfield: Augustus Kelley. Ulman, Lloyd (1955), ‘Marshall and Friedman on Union Strength’, Review of Economics and Statistics, 37, 384–401. Valde´s, Rodrigo (1997), ‘Emerging Markets Contagion: Evidence and Theory’, Documento de Trabajo del Banco Central de Chile No.7. r Blackwell Publishing Ltd. 2003

178

Thomas Moser

Van Rijckeghem, Caroline, and Beatrice Weder (2000), ‘Spillovers Through Banking Centers: A Panel Data Analysis’, IMF Working Paper No. 00/88. Van Rijckeghem, Caroline, and Beatrice Weder (2001), ‘Sources of Contagion: Is It Finance or Trade?’, Journal of International Economics, 54, 293–308. White, Horace (1940), ‘Foreign Trading in American Stock-Exchange Securities’, Journal of Political Economy, 48, 655–702. Wood, Geoffrey (1999), ‘Great Crashes in History: Have they Lessons for Today?’, Oxford Review of Economic Policy, 15, 98–109.

r Blackwell Publishing Ltd. 2003