Volatility analysis during the Asia crisis: a multivariate GARCH-M model for stock returns in the U. S., Germany and Japan (Q2722295)

The analysis of international financial systems and the interconnection of markets have become major topics in financial econometrics in recent years. The availability of daily data and the connectedness of financial markets have inspired the analysis of the transmission mechanism in terms of the mean and variances between time series. In the present paper the stock returns of Germany, Japan and the U.S. for a period of 2 years (June 21, 1996 to June 22, 1998) are examined. A multivariate VAR-GARCH-M model is estimated to explore the relationships in the returns and the conditional variances or, also called, volatilities. It is found that a multivariate ARCH-M model is better than traditional VAR and VAR-GARCH models. The model is estimated by MCMC methods and then compared them by marginal likelihood criteria. The MCMC approach allows the introduction of new concepts and to find exact (small sample) results for characteristics of the dynamic process, like the impulse response function or the predictive distributions. The VAR-GARCH-M model is tested for a break point at October 24, 1997, the day when the Asia crisis hit the Dow Jones index. It is found a clear evidence of a changed dynamic interaction pattern for the period before and after October 24, 1997.