The t copula and its properties are described with a focus on issues related to the
dependence of extreme values. The Gaussian mixture representation of a multivariate
t distribution is used as a starting point to construct two new copulas, the skewed t
copula and the grouped t copula, which allow more heterogeneity in the modelling of
dependent observations. Extreme value considerations are used to derive two further new
copulas: the t extreme value copula is the limiting copula of componentwise maxima of
t distributed random vectors; the t lower tail copula is the limiting copula of bivariate
observations from a t distribution that are conditioned to lie below some joint threshold
that is progressively lowered. Both these copulas may be approximated for practical
purposes by simpler, better-known copulas, these being the Gumbel and Clayton copulas
respectively.
Gaussian copula's fit in terms of the AIC is worse than that of other copulas. Further, the Gaussian copula seems to underestimate the probability of joint strong risk factor changes for the data sample at hand