On volatility transfers during the Asia crisis in 1997-1998

To test the heat wave or meteor shower hypothesis, like Booth et al. (1997), we build a vector autoregressive (VAR) model for the conditional variances of the 5 variables, later for 11 variables. With 5 or 11 variables, standardmethods available for 3 dimension-cases do not work (multivariate vector AR-GARCH models do not work). Therefore, we use an ad hoc method. The period for estimation is from 1994 to 1998. We estimate univariate MAGARCH, AR-GARCH or ARMA-GARCH models for the volatilities and we try to find out which volatilities might be the causes for others.