Dependence in the Dutch market explained using a skewed t-copula in pair-copula models.

The three-dimensional pair-copulas model is used during this research. Different from other research is the bivariate skewed t-copula, which is attempted to use in the pair-copula construction. First the residuals are computed with the use of an AR-GJR-GARCH model followed by Kolmogorov-Smirnov goodness of t tests. After that the copula construction is explained, the bootstrap version of the Anderson-Darling test is used as goodness of t test for the copula models. The limitations of the skew t-copula in a pair-copula model are described. All constructed models did not describe the data correctly, according to the bootstrap version of the Anderson-Darling test. Besides, the value at risk outcomes were not always in favour of the comparison model (the multivariate Student's t-copula).