Contagion in the Presence of Stochastic Interdependence

Abstract

Contagion represents a significant change in cross-market linkages precipitated by a crisis and is properly measured only after taking into account the interdependence or extant linkages prevailing between markets. Since it is well known that stock return volatilities and correlations are stochastic in the absence of a crisis, interdependence between markets should reflect the time varying nature of these covariances. We measure contagion in the presence of stochastic interdependence using data on stock indices from South East Asian countries around the July 1997 crisis. Since stock return covariances are observed with error, this suggests casting our model in a state space framework which is estimated using a multivariate Kalman filter. In the presence of stochastic interdependence, we find reliable evidence of contagion between Thailand and Indonesia, Malaysia, and the Philippines but not between Thailand and Hong Kong or Singapore.