Estimating and Evaluating the Predictive Abilities of Semiparametric Multivariate Models with Application to Risk Management

In this paper we propose a new semiparametric procedure for estimating multivariate models with conditioning variables. The semiparametric model is based on the parametric conditional copula of Patton (2005a) and nonparametric conditional marginals. To avoid the curse of dimensionality in the estimation of the latter, we propose a dimension reduction technique. The marginals are estimated using conditional kernel smoothers based on local linear estimator. The semiparametric copula model is compared with the parametric DCC model using predictive likelihood as a criterion. The comparison is based on the recent conditional test for predictive abilities of Giacomini & White (2005). We use various simulations and financial series ∗Work in progress: preliminary and incomplete. I express my gratitude to Dr. Cees Diks and Professor Cars Hommes for helpful comments and enthusiastic supervision. I also thank to seminar participants at Tinbergen Institute Amsterdam, University of New South Wales, University of Adelaide and the Quantitative Methods in Finance conference in Sydney. The usual disclaimers apply.