TIME-VARYING HIERARCHICAL ARCHIMEDEAN COPULAS: A NON-PARAMETRIC APPROACH

A non-parametric approach based on U-statistics in the construction of the time-varying hierarchical Archimedean copula model has been proposed. The procedure aims to analyze changes in the Kendall correlation matrix computed from the observations. This makes it possible to identify local segments of homogeneity and to construct a hierarchical Archimedean copula model on each segment. The simulations performed made it possible to assess the power of the proposed procedure.