Financial contagion

From Wikipedia, the free encyclopedia

This article (or section) may need to be wikified to meet Wikipedia's quality standards.
Please help improve this article, especially its section layout, and relevant internal links. (help)
This article has been tagged since January 2007.

There are very few or no other articles that link to this one.
Please help introduce links in articles on related topics. After links have been created, remove this message.
This article has been tagged since August 2006.

Financial contagion refers to the phenomena when one country's economy is jolted because of changes in the asset prices of another country's financial market.
During the last decade international financial markets have been affected by a series of major currency and financial crises: the speculative attacks on the parities of the Exchange Rate Mechanism in 1992, the Mexican “Tequila” peso collapse of 1994, the East-Asian flu of 1997, the Russian default in 1998, and the Brazilian devaluation of 1999. International capital markets in general appear volatile, both on the down-side and the up-side, with emerging market economies suffering most. But last decades showed the most interesting characteristic of these crises - the transmission of shocks from one country to another markets of different sizes and structures around the world. These spillovers often can not be explained on the basis of traditional linkages between economies. Existence of these excessive transmissions of crises has led to conclusion that there is another driving force that makes economies co-move in extreme directions. And the name for this is contagion, a force that threat the benefits of international diversification – only the “free lunch” in finance. On one day of October 1987, the U.S. stock market crashed by some 20 percent, or about 20 times its historical daily standard deviation. The British, Japanese and German markets dropped between 8 and 15 times their normal standard deviations. Australian stock market reacted to the US crisis by 44 percent decline. Other bond and currency market indicators also witnessed large declines. Another example is the case of Russian default. In the month after the 1998 devaluation of the Russian ruble, the Brazilian stock market fell by over 50%. These two economies are located in separate geographic regions, have very different structures, and have virtually no direct linkages through channels like a trade. But many researchers agree that this was the case of contagion (Forbes and Rigobon, 2001).

What is contagion?
Despite of substantial progress in research, there is still no consensus on definition of contagion. Generally it may be defined as an excessive unanticipated increase in cross-market linkages after a shock to an individual country (or group of countries). Dornbusch, Park and Claessens (2000) offer such a definition for contagion: "Contagion, in general, is used to refer to the spread of market disturbances - mostly on the downside - from one country to the other, a process observed through co-movements in exchange rates, stock prices, sovereign spreads and capital flows." Another definition is “association of excess returns in one country with excess returns in another country (or group of countries) after controlling for the effects of fundamentals.” (Eichengreen, Rose and Wyplosz, 1996). This definition is closely related to “true” contagion, as defined in Kaminsky and Reinhart (2000), arising in the absence of, or after controlling for, common shocks and all positive interconnection channels. Contagion should be distinguished from interdependence . Interdependence that represented by fundamental cross-country linkages are usually controlled and related changes are anticipated – for instance, through trade and financial flows or other common links. Different researchers use different fundamentals to control excess volatility. Exchange rates, interest rates, CPI inflation, the current account and budget balances, stock market indices, indicators of domestic credit and labour market, political conditions. Unfortunately, very few fundamentals are found to be statistically significant control variables (Dornbusch, R, Park,Y. and Claessens, S., 2000). Researchers argue that interdependence can be measured and anticipated. However, an existence of extreme, asymmetric patterns of excess volatility that can not be explained by fundamental links between economies has led experts to suggest that contagion - rational or irrational - can be the only remaining explanation.

Channels of contagion
The driving forces behind international propagation of crises can conceptually be divided into two categories (Forbes and Rigobon, 2001 and Pritsker,1999):

 Fundamental causes. This group includes contagion resulting from the normal interdependence among market economies. The interdependence will mean that shocks, whether of a global or local nature, will be transmitted across countries because of their real and financial linkages. Calvo and Reinhart (1996) term this type of crisis propagation “fundamentals-based contagion”.

i)Common Shocks
ii)Trade Links
iii)Competitive Devaluations
iiii)Financial Links

 Behavioural causes This group involves financial crises that are mostly active during crisis and cannot be linked to observed changes in macroeconomic or other fundamentals. Under this definition, contagion arises when there is co-movement that cannot be explained on the basis of fundamentals. In this case, international financial contagion is considered as the market imperfection – irrational force that represents expectations and actions of all the investors and other agents involved in crisis. These irrational phenomena can be financial panic, herding behaviour, loss of confidence, and increases in risk aversion. A crisis in one country may, for example, lead investors to withdraw their investments from many markets without distinguishing differences in economic fundamentals. But, these phenomena can be individually rational, and still lead to a crisis. Information asymmetry and speculative attacks caused, as history shows, huge outflows from crisis countries and often lead to destabilized economies, in many cases magnified by activity of international policy makers such as International Monetary Fund. (Stiglitz, 2002).
This group can be divided to three main causes:
First, there are actions which are initially individually rational, but which do lead to what could be called excessive co-movements, in the sense of not being explained by real fundamentals. This can broadly be called the investors’ practice channel: contagion is transmitted through the actions of investors outside the country, each of which acts individually rational. Conceptually, one can further distinguish this investor behaviour into:
i. liquidity problems
ii. incentive problems
iii. informational asymmetries and market coordination problems.

Second, there can be cases of multiple equilibrium, similar to those in models of commercial bank runs, which can imply contagious behaviour.
Third, there can be changes in the regulations of international financial system, which make investors behave differently following an initial crisis.

While it is difficult to distinguish whether co-movements have been irrational or fundamental, empirical work has been able to document patterns in the vulnerability of countries to volatility and identify possible channels of transmission of contagion. Trade links, regional patterns, and macro-similarities have been found to make countries vulnerable to volatility. Common creditor and other links through international financial centres are found to be mechanisms through which volatility is transmitted from a particular country to other countries at a particular point. These regularities have helped to identify those countries which are at risk of contagion. Less is known on the importance of micro-economic conditions and institutional factors in propagating shocks, including specific financial agents’ actions through which contagion appears to happen such as big speculators.

International stock markets nowadays are the cross-related system where different agents affect each other. The degree of impact depends on many factors: geographic location, regional linkages, size of economy and integration in global market. Contagion often has greater impact on markets located in the same geographic region. For example, South East Asian markets are strongly influenced by leading economy in the region – Japan. Spillovers can spread even to different geographical regions and cause chain reaction. Macroeconomic changes in, for example United States can influence market returns in Japan via close informational interdependency (U.S. and British indices are believed to be most influencing world markets). In other words, there are causal relationships between market indices. It is so called meteor-shower effect – a transmission of volatility across different markets world-wide. The main difference between heat-wave and meteor-shower effects is dependence of market’s volatility on, correspondingly, internal and external factors. In other words, heat wave-effect is, econometrically speaking, autoregression whereas meteor-shower shows interdependency of the different markets’ returns.
Consistent with prior research, strongest impact originates from most influencing countries. For example, Turkish collapse and Argentinean crisis did not cause significant correlation breakdowns on the global market while two crises in US, especially terrorist attacks of 9/11, induced strong turbulences worldwide. Another case is Russian default which triggered a sequence of macroeconomic shocks around the globe and became the main reason of the crash of Long-Term Capital Management Hedging Fund that suffered huge losses of $4.6 billion during four months of 1998. It is believed that on of the main factors that undermined the LTCM’s global portfolio of bonds and derivatives was contagion – unanticipated short-term increased correlations of depreciating assets broke long-term hedging strategy of the Fund. The profits from LTCM's trading strategies were generally not correlated with each other and thus normally LTCM's highly leveraged portfolio benefited from diversification. However, the general flight to liquidity in the late summer of 1998 led to a re-pricing of all risks and these positions then did all move in the same direction. As the correlation of LTCM's positions increased, the diversified aspect of LTCM's portfolio vanished causing large losses to its equity value. Thus the primary lesson of 1998 and the collapse of LTCM for Value-at-Risk users is not a liquidity one, but more fundamentally that the underlying covariance matrix used in Value-at-Risk analysis is not static but changes over time. The incident is consistent with the insight often attributed to the economist John Maynard Keynes, who has warned investors that although markets do tend toward rational positions in the long run, "the market can stay irrational longer than you can stay solvent."
Another interesting feature of contagious crises is investors’ “over-reaction” to bad news. It was observed that nearly any contagion was strongest directly after a shock and then decreased rapidly within a few months, sometimes a few weeks. This evidence is consistent with behavioural explanations of contagion. For example, 9/11 after-shock was in many cases very short by its duration, caused mostly by investors’ expectations and overall uncertainty. This could happen if investors become more uncertain about underlying long-term economic and financial growth rates and trends and therefore attach large significance to relatively small pieces of news. It appears that in long-term prospect much of short-term impact of contagion lose its strength. Another explanation is after-shock market corrections of a crisis that had occurred in the past. Investors’ over-reaction during and after economic shock tend to calm down over time when market begins to stabilize.
Another point should be considered. The common feature of the studies which quantify the reduction in gains from diversification is the assumption that cross-market linkages increase during periods of high turbulence. However, there are cases where cross-market relationships actually decrease, thereby potentially increasing the gains from diversifying across markets. Assuming that markets exhibit greater co-movements at times of distress is a limiting assumption. Researchers should really focus on changes in cross-market linkages and not exclusively on increases. Markets’ simultaneous reaction can be different from the same event. Herd may run simultaneously but in absolutely different directions.

Implications
The phenomenon of global stock market contagion is now too familiar now and serious to ignore and has become an integral part of the stock market activity. International investors are the most interested researchers of the mechanisms and channels of shocks transmissions. An understanding of the logics and patterns of the transmission of idiosyncratic shocks has important implications for portfolio management. Their main concern is that increasing interdependence between markets may mitigate benefits of international diversification – the technique that allows to diversify away idiosyncratic risks of domestic portfolio by including uncorrelated international assets. The key variable – low correlation between world markets has been increasing in last decades due to growing global integration. As markets become more interdependent, a particular market will be influenced by external factors to a greater extent than when markets are segmented. When all markets follow the same direction even well diversified international portfolio is likely to step off an efficient frontier. This problem requires update international diversification rationale.
Intensifying effect is that markets move together more closely during periods of big declines than they do at other times. In other words, interdependence mostly affects portfolios during periods of turbulence, i.e. when diversification benefit is mostly needed. This phenomenon is sometimes referred to as “correlation breakdown”.
Numerous studies argued that interdependence can be measured and anticipated and effectively incorporated in construction of diversified portfolios. But do increased correlation coefficients between integrated markets mean that global co-movements increase over time and especially during periods of turbulence? Should long-term investors worry about correlation behaviour? Researchers argue that one should be very cautious using traditional tests of breakdowns in correlation coefficients which usually find excessive transmission of shocks and discontinuities in the data-generating process. Researchers warn about econometric specifications of raw correlation and propose several techniques to measure contagion without bias. Generally approaches to measure contagion can be divided to two main types:

 Correlations that simply illustrate degree of co-movements between variables. Adjusted correlation is an extended analysis of raw correlations corrected for econometric problems and biases. However, one should be aware that high correlation can not be interpreted as an evidence of strong interdependence between markets. In many cases it appears to be statistical anomaly. Comparison of adjusted correlation to raw correlations would be a great interest to estimate the “true” contagion. Also, using different timeframe, obtained results can be compared to prior literature related to adjusted correlation.
 Measures of causation (and contagion is causation) between co-movements such as conditional probabilities and volatility spillover estimation (GARCH). These approaches let extend analysis of contagion to reveal causal relationships between variables. Additionally to estimation of existence and degree of contagion effects, conditional probabilities can measure probability of crisis in one country given the strength of interdependence with crisis country. Conditional probabilities method can be used to define countries which are highly vulnerable to contagion. The main advantage of GARCH approach is that it can be used to measure direction, as well as impact of volatility spillover.

Estimation of contagion

There are several different approaches to measure contagion such as:

a. Correlation analysis
b. Adjusted correlation approach
c. Conditional probabilities
d. Volatility spillover estimation

a. Correlation analysis.

As financial contagion implies excess returns, then the problem of defining these is immediately raised. Early analyses of the existence of contagion effects focus on changes in correlation coefficients. The central idea is to assess whether the correlation coefficient between two stock markets changes across periods of low and high volatility. The asset price tests consist of estimates of correlation coefficients of changes in interest rates, stock prices, and sovereign spreads of different economies. Under this approach, a marked increase in correlations among different countries’ markets is considered as evidence of contagion. Most studies estimating correlations among markets find evidence of large co-movements in a variety of asset returns.

b. Adjusted correlation approach.

Some authors argue that increase in correlations among different countries’ markets may not be sufficient proof of contagion. Numerous studies have recognized that focusing on raw correlations can be misleading. This measure suffers from econometric and adjustment problems. The conclusion of a correlation breakdown is derived by estimating the correlation in periods of high volatility of returns, so called “conditional” correlation. However, many researchers have shown that this is a biased sampling estimate of the true correlation . Also, correlation, being a function of variance, is biased upwards during tail events. Empirical distributions of financial markets returns are typically non-normal and tend to have fat tails (leptokurtic distribution). Another problem is volatility clustering (time-varying heteroscedasticity / GARCH effects). In other words, the occurrence of large positive or negative returns is more frequent then expected under normal distribution. Moreover, raw correlations approach does not properly model the interdependence across international financial markets. If markets are historically cross-correlated, then a sharp change in one market will naturally lead to changes in the other markets and markets could exhibit an appreciable increase in correlations during crisis periods. Some researchers adjusted for the effects of heteroskedatisticity, endogeneity and omitted variables by making some assumptions. Generally, studies making use of adjusted correlation coefficients find almost no evidence of contagion.

c. Conditional probabilities analysis

Another way to control for the role of fundamentals is to study conditional correlation or probabilities rather than raw correlations and thus use a narrower definition of contagion. The most commonly used methodology, introduced by Eichengreen, Rose, and Wyplosz (1996), is to examine whether the likelihood of crisis is higher in a given country when there are crises in one (“ground-zero”) country or several countries. The approach taken is generally estimating the probability of a crisis conditional on information of the occurrence of crisis elsewhere, taken into account fundamentals or similarities. One advantage of this definition is that it readily allows statistical tests of the existence of contagion. These tests can also try to investigate the channels through which contagion may occur, distinguishing, among others, trade and financial links.

d. Volatility spillover estimation

Another way is to test estimates of spillover in volatility, i.e., cross-market movements in the second moments of asset prices. So far, these approaches do not control for fundamentals and do thus not distinguish between pure and fundamental based contagion.