Copula choice with factor credit portfolio models

Abstract

Over the last couple of years we could observe a strong growth of copula based credit portfolio models. So far the major interest has revolved the ability of certain copula families to map specific phenomena such as default clustering or the evolution of prices (e.g., credit derivatives prices). Still few questions have been posed regarding copula selection. This is surprising as the problem of estimating the dependence structure is even unresolved with simple traditional models. For statistical tests of credit portfolio models in general the literature found densitybased tests like that of Berkowitz (2001) the most reasonable option. In this text, we examine its power characteristics concerning factor portfolio models in more detail. Our results suggest that both the copula family as well as the level of dependence is generally very difficult to identify. © 2010 Springer-Verlag Berlin Heidelberg.