C. Ausin, M. Kalli
Copulas are especially useful in capturing the dependence in the tails of the distributions. This can be particularly higher in many situations such as financial crises, where tail dependencies are usually more relevant than overall correlations. However, parametric copulas can be too restrictive in practice and they may have problems capturing both lower and upper tail dependencies without losing coverage of the center area of the joint distribution.
From the Bayesian point of view, only a few nonparametric copula models have been proposed and none of them are able to capture tail dependencies. In this work, we propose a new Bayesian nonparametric copula which can be viewed as a multivariate histogram smoothing. Thus, we impose a prior on the breakpoints and on the volume of the bins. A Gibbs sampling approach can be implemented to sample from the posterior and predictive. We illustrate the procedure using simulated and real data based on multivariate financial time series.
Keywords: Bayesian nonparematrics, copulas, tail dependence, skewness
Scheduled
Bayesian methods II
November 10, 2023 12:00 PM
HC2: Canónigos Room 2