Measuring contagion and interdependence with a Bayesian time-varying coefficient model: An application to the Chilean FX market during the argentine crisis

We use a Bayesian time-varying coefficient model to measure contagion and interdependence, and we apply it to the Chilean FX market during the 2001 Argentine crisis. The proposed framework works in the joint presence of heteroskedasticity and omitted variables, without knowledge of the crisis timing prior to the empirical analysis. It can distinguish between contagion and interdependence, as well as between unusually strong or weak market comovements. In a "natural" experiment based on our application, we find that the proposed framework works well in practice. In the application, we find evidence of some contagion from Argentina and some interdependence with Brazil.