Testing for Asset Market Linkages: A new Approach based on Time-Varying Copulas

This paper proposes a new approach based on time-varying copulas to test for the presence of increases in stock market interdependence after financial crises, also known as shift-contagion process. We show that the previous approaches that take into account changes in volatility regimes are biased when the DGP is either copula based or when there is a break in variance significantly different from the one in correlation. A sequential algorithm is then elaborated to remove this bias. Applied to the recent 1997 Asian crisis, it confirms that breaks in variances always precede those in conditional correlation. It also turns out that this financial turmoil has been characterized by shift-contagion.