Prediction through a Gumbel Barnett copula

In this paper we investigate the performance of a copula type Gumbel-Barnett, for prediction of the dependence between two variables, X and Y. We fit a copula, using a sample of observations collected at time t and we define predictive areas, at different levels, to detect specific changes in a sample at time t + 1, both samples are governed by the same law (copula). The study addresses the more general problem of detecting dependence changes during different periods of time. This work is developed under a Bayesian perspective, a procedure that allows working with small sample sizes and under different assumptions introduced in the model through the prior distribution.