Time-Varying Integration, Interdependence and Contagion

Abstract

Bekaert, Harvey, and Ng (2005a) define contagion as "correlation over and above what one would expect from economic fundamentals". Based on a two-factor asset pricing specification to model fundamentally-driven linkages between markets, they define contagion as correlation among the model residuals, and develop a corresponding test procedure. In this paper, we investigate to what extent conclusions from this contagion test depend upon the specification of the time-varying factor exposures. We develop a two-factor model with global and regional market shocks as factors. We make the global and regional market exposures conditional upon both a latent regime variable and two structural instruments, and find that, for a set of 14 European countries, this model outperforms more restricted versions. The structurally-driven increase in global (regional) market exposures and correlations suggest that market integration has increased substantially over the last three decades. Using our optimal model, we do not find evidence that further integration has come at the cost of contagion. We do find evidence for contagion, however, when more restricted versions of the factor specifications are used. We conclude that the specification of the global and regional market exposures is an important issue in any test for contagion.

Keywords: Contagion, Financial integration, Volatility spillover models, Time-varying correlations, Regime-switching models

JEL Classification: G15, G12, F30, F32, F35