On the family of multivariate chi-square copulas

This paper explores the theoretical properties and the practical usefulness of the general family of chi-square copulas that recently appeared in the literature. This class of dependence structures is very attractive, as it generalizes the Gaussian copula and allows for flexible modeling for high-dimensional random vectors. On one hand, expressions for the copula and the density in the bivariate and the multivariate case are derived and many theoretical properties are investigated, including expressions for popular measures of dependence, levels of asymmetry and constraints on the Kendall’s tau matrix. On the other hand, two applications of the chi-square copulas are developed, namely parameter estimation and spatial interpolation.