Coexceedances in Financial Markets - a Quantile Regression Analysis of Contagion

This paper analyzes simultaneous exceedances (coexceedances) of several stock index returns for different thresholds with a focus on the Asian crisis in 1997. We introduce a new concept of computing and estimating time-varying coexceedances and usethe quantile regression model to analyze contagion. We find contagion (i.e. more and larger coexceedances) during the Asian crisis among countries and also across regions. The results are based on a comparison of coexceedances of the full period to a benchmark period conditional on certain regimes of extreme market movements. An analysis of the evolution of coexceedances also reveals the time-varying character of contagion. Conditional density estimates additionally point to the existence of multiple equilibria in crisis periods