Bayesian Model Selection of Regular Vine Copulas

Abstract

Regular vine copulas can describe a wider array of dependency patterns than the multivariate Gaussian copula or the multivariate Student’s t copula. We present reversible jump Markov chain Monte Carlo algorithms to estimate the joint posterior distribution of the density factorization, pair copula families, and parameters of a regular vine copula. A simulation study shows that our algorithms outperform model selection methods suggested in the current literature and succeed in selecting the true model when other methods fail. Furthermore, we present an application study that shows how a vine copula-based approach can improve the pricing of exotic financial derivatives.