An Application of the Pair-Copula Construction to a Non-life Dataset

The modern pair-copula construction (PCC) approach, which defines complex multivariate structures through the use of bivariate copulas, it proves to be an extremely effective tool for respond to problems in various fields of application including the actuarial one. The aim of this paper is to analyze the PCC methodology through an application to a non-life insurance portfolio in presence of categorical and continuous data. The aim is to define a multivariate distribution, highlighting the technical and operational limits in applications in the insurance field. This methodology allows to overcome both the limits of the “traditional” dependence structures and of the more “modern” copula functions. However, since each varied n-distribution has a considerable number of decompositions, the multivariate distribution was determined using Dißmann’s algorithm