Aas et al. (2009) discovered vine pair-copula constructions (PCC's) for modeling multivariate dependence and thus put an important milestone in the usage of multivariate copulas. At present time PCC's are recognized to be as a most flexible class of multivariate copulas. In this paper we combine vine PCC's and semiparametric copulabased dynamic (SCOMDY) models of Chen and Fan (2006a) with ARMA-GARCH margins. As building blocks of the PCC's we use bivariate t-copulas. Further we consider exchange rates as application and model their dependence structure with regular and canonical vines. Alternatively we also consider a non Gaussian directed acyclic graph (DAG) model, which can be imbedded as a special PCC. A nonnested model comparison of the above SCOMDY models is performed using the adapted Voung's test by Chen and Fan (2006a).      «

Aas et al. (2009) discovered vine pair-copula constructions (PCC's) for modeling multivariate dependence and thus put an important milestone in the usage of multivariate copulas. At present time PCC's are recognized to be as a most flexible class of multivariate copulas. In this paper we combine vine PCC's and semiparametric copulabased dynamic (SCOMDY) models of Chen and Fan (2006a) with ARMA-GARCH margins. As building blocks of the PCC's we use bivariate t-copulas. Further we consider exchange...     »