Testing International Financial Contagion with Unknown Structural Breaks

While financial crises are often regarded as contagious outside academia, the empirical evidence for the existence of international contagion of financial crisis is mixed in the literature.  Forbes and Rigobon (2002, Journal of Finance) show that although inter-market correlation tends to increase during a crisis period, there is little evidence of contagion once heteroskedasticity is accounted for.  However, Corsetti et al. (2005 Journal of International Money and Finance) and Dungey et al. (2005 Quantitative Finance) present evidence that the adjusted correlation approach of Forbes and Rigobon is biased toward the rejection of contagion.

This study tests for the existence of international contagion of financial crises, using the same definition of contagion as in Forbes and Rigobon (2002), i.e. significant increase in cross-market linkages, but with a different methodology.  Instead of testing for increased correlation, we follow Dungey et al. (2005) to directly test for structural breaks in the cross-market linkages.  However, while Dungey et al. propose a Chow-type test assuming the break points are known, we treat the break points as unknown and use the methods of Bai and Perron (1998 Econometrica and 2003 Journal of Applied Econometrics) to estimate them.  This approach allows one market’s crisis to spread to another market with a substantial latent period instead of an instantaneous spillover as assumed in previous studies.  Our method is also consistent with the observation that the starting and ending of a crisis are varied across affected countries.

Using the new method, we test for structural breaks in the linkage between the daily stock market return of each of the eight Asian countries (China, Indonesia, Malaysia, Korea, the Philippines, Taiwan, India and Hong Kong) and the return of the Thailand market where the 1997 Asian crisis started.  The stock market returns are calculated from January 1997 through December 1998, using the IFC Investable Index and the IFC Global Index of the Emerging Market Data Base published by Standard and Poor’s.

We find a significant upward shift in the linkage between the stock returns of Thailand and seven out of eight Asian countries, providing strong evidence for the existence of international financial contagion.  The estimated break points are different across markets approximately ranging from one month to five months since the floating of the Thai Baht (7/2/1997) although all such upward shifts are in the second half of 1997.  As for the linkage between the Hong Kong market and other Asian markets, we obtain similar but weaker evidence for contagion.  The evidence for contagion is also maintained when global and regional factors are controlled for by the U.S. and Japanese stock returns.