An analysis of time-varying exchange rate dependence using Markov switching GAS copulas

We extend the time-varying GAS copula framework described by Creal et al (2013) by introducing a hidden Markov Chain that captures unobserved regime switching and we analyse the time-varying dependence structure of exchange rate pairs consisting of emerging and developed market countries. Our results suggest that this approach is capable of offering relevant insights about the tail dependence dynamics of exchange rate pairs, whereas it is not an improvement relative to a GAS model with structural breaks when determining the dependence structure with elliptical copulas. Our analysis shows that exchange rates pairs of economies with similar backgrounds have a higher dependence structure than pairs from economies with different backgrounds. We find evidence that events of financial distress cause the exchange rate dependence to significantly decrease. Furthermore, we find that these events can impact the tail dependence of exchange rate pairs in the subsequent period by causing asymmetries which are economically interpretable. These differences in dependence structure are relevant for risk managers for whom changes in exchange rate dependence, such as joint currency crashes, are of major importance when hedging exchange rate risk.