Financial Contagion and International Portfolio Flows

Abstract

Financial contagion is an issue of interest to policymakers and portfolio managers. In the presence of contagion, the benefits of geographical diversification may be overstated. Contagion is defined independently of changes in macroeconomic fundamentals. It corresponds to a situation in which shocks in one country spread to another country in a significantly more pronounced manner than during quiet times. In the study reported here, I attempted to measure contagion from daily cross-border portfolio flows. Using data from State Street Bank and Trust Company, I constructed three indexes for 1996 through 2000. I found little statistical evidence of contagion, but the weighted index that measured the “contagiousness” of flows reached its four-year low in August 1998, suggesting that the Russian crisis was characterized by both contagion and large aggregate outflows. I also found that contagion appears to be regional and that the developed countries seem to be sheltered from contagion.

Without doubt, a phenomenon such as the Black Death can be identified as contagious, but we cannot make the same claim about crises in financial markets. Financial contagion is an issue of interest not only to policymakers but also to portfolio managers, for in the presence of contagion, the benefits of geographical diversification may be overstated. The waves of crises in the 1990s have prompted economists to examine the factors that trigger such crises and to propose models that issue warning signals. But identifying and measuring contagion depend on how contagion is defined.

In the traditional view, contagion is a result of unexpected changes in macroeconomic fundamentals. The change may arise from aggregate shocks that affect the fundamentals of several countries simultaneously or from a shock in one country that affects the fundamentals in other countries. This definition can be challenged. As to the first cause, aggregate shocks do not constitute contagion because they are not transmitted from one country to another; they hit all countries simultaneously. As to the second cause, a shock in one country may affect others because the countries are linked by trade ties or a common economic policy and are hence part of the same real economy. Separating the amount of contagion that may be caused by common macroeconomic factors from the contagion provoked by market news or shifting investor expectations and risk preferences is a challenge.

My approach is based on research that focused on the propagation of shocks between countries, wherein contagion was reflected in stock prices. I investigated whether investment flows, however, detect contagion. Capital flows may provide a way of detecting contagion better than stock prices do because flows reflect both prices and quantities. They are also linked to the preferences and beliefs of investors and are thus likely to provide valuable information.

I extend previous research on the propagation of shocks between countries in several ways. First, I tested the model on a unique database of cross-border equity flows generated by international investors. Second, I constructed daily aggregate contagion indexes for the 1996–2000 period. Third, to test whether contagion is more global or more regional in nature, I built these indexes so that they could be decomposed into their regional components—the “safe” havens of the developed markets, the emerging markets of Latin America, Asia, and Europe, and the commodity-oriented countries.

I found little statistical evidence of contagion. The first of the three indexes, which quantified the average likelihood of contagion across 23 countries, indicated that after reaching a high point in August 1997 (the Asian currency crisis), contagion decreased. It then increased again and culminated in the middle of April 1998, so when the Russian crisis occurred four months later, contagion was already at a historically high level.

The two other indexes of contagion attempt to measure the “contagiousness” of equity flows. They take into account the magnitude and direction of net equity flows, as well as the likelihood of contagion, thus distinguishing positive from negative contagion. The Weighted Index indicates that the risk of contagion was extremely high in April 1998. The index reached its four-year low at the end of August 1998, suggesting that the Russian crisis was characterized by both contagion and large aggregate outflows. This pattern differs from the behavior of the index during the Asian crisis. The increase in the index between September 1997 and March 1998 suggests that a flow of capital out of the turbulent regions was accompanied by an inflow to safer countries.

As to the regional aspect of crises, the indexes indicate that the region that suffered the least from contagion was the safe countries group, followed by Latin America. Asia experienced the highest proportion of contagious days.

I found little evidence of propagation from one region to another, although some evidence indicates that turbulence in Asia may propagate to emerging European and commodity-oriented countries. The group of safe countries appears to be sheltered from contagion.

Assuming that the indexes capture financial contagion to some extent, the next question is: Is contagion predictable? Analysis of the time-series properties of the indexes indicates that they possess some persistence. The results of a regime-switching model tend to support the existence of various turbulent regimes, with the turbulent regime being highly transitory.