Title: Robust and flexible mixture models for the identification of structural shocks of financial time series
Authors: Markus Haas - University of Kiel (Germany) [presenting]
Sebastian Mueller - Christian-Albrechts-University Kiel (Germany)
Abstract: Dynamic mixture or regime-switching models have long been popular in empirical finance due to their good fit and economic interpretability of the extracted regimes. More recently, these models have been employed to identify the contemporaneous structural effects in vector autoregressive models. Gaussian mixtures are typically used in this context, with all structural shock processes changing regimes simultaneously. However, the assumption of (simultaneous) Gaussian regimes is often inappropriate for the structural shocks driving multivariate financial time series, and this may seriously distort the identification and economic interpretation of these shocks. Thus, currently existing approaches to identification via regime-switching effects are extended in order to accommodate thick-tailed innovations and independently switching components. The usefulness of the models is illustrated by applying them to a set of relevant problems in financial economics, such as transmission of shocks in the euro area and price formation in foreign exchange markets.