Market conditions and time varying conditional correlations

This paper shows how the dependency of time-varying conditional crosscorrelation on prevailing market conditions can be modeled. With this modelling approach it is possible to empirically investigate how correlations between different markets are dependent on market volatilities and other external factors. The paper shows that the modeling of conditional correlations can be simplified to modeling a univariate GARCH-process. The advantage of this approach is that it allows for utilization of existing GARCH methodology and software to estimate the dynamics of correlation. Our empirical results reveal that time-varying correlations between stock markets are largely unpredictable and are at best dependent on world market volatilities. Weaker evidence is found that correlations are driven by market downturns.