Title: Copula-MGARCH with dynamic conditional covariance decompositions
Authors: Fabian Raters - University of Goettingen (Germany) [presenting]
Helmut Herwartz - Georg-August-University Goettingen (Germany)
Abstract: The Copula-MGARCH (C-MGARCH) model incorporates standardized copula distributed innovations in MGARCH models. Recent literature suggests that a static continuous decomposition of the standardized innovations' covariance matrix enhances the model's flexibility. A further generalization of the C-MGARCH approach is identified by means of dynamic decompositions of the conditional covariances. These decompositions are assumed to depend on a latent univariate autoregressive process and, considering bivariate BEKK, CCC, and DCC models, dynamically rotate the standardized innovations. Contributing to the recent literature, we suggest a stepwise estimation procedure of the models' parameters by means of Maximum Likelihood and particle filtering. The performance of our extension applied to the aforementioned models is evaluated in a comprehensive study. Empirically, we conduct an application to the log-differences of daily exchange rates by means of in-sample information criteria and ex-ante portfolio Value-at-Risk coverage tests. Our approach contributes to the modeling of nonlinear dependencies between multivariate time series of log-differences in volatility models.