Detecting financial contagion in a multivariate system

This paper proposes an original three-part sequential testing procedure (STP), with which to test for contagion using a multivariate model. First, it identifies structural breaks in the volatility of a given set of countries. Then a structural break test is applied to the correlation matrix to identify and date the potential contagion mechanism. As a third element, the STP tests for the distinctiveness of the break dates previously found. Compared to traditional contagion tests in a bivariate set-up, the STP has high testing power and is able to locate the dates of contagion more precisely. Monte Carlo simulations underline the importance of separating variance and correlation break testing, the endogenous dating of the breakpoints and the usage of multi-dimensional data. The procedure is applied for the 1997 Asian Financial Crisis, revealing the chronological order of the crisis events.