Title: Modeling the dependence in compound model using copula representation when the size of frequency is informative
Authors: Woojoo Lee - Inha University (Korea, South)
Jae Youn Ahn - Ewha Womans University (Korea, South) [presenting]
Rosy Oh - Ewha Womans University (Korea, South)
Abstract: While the copula method is a popular choice in modelling the dependence, the choice of the proper copula family is much harder in general compared to the choice of the proper marginal distribution families. Especially, in the modelling of dependence between frequency and average severity, we show, by example, that classical copula approach may mislead the dependence between frequency and severity. We provide the copula model which can safely model the dependence between frequency and individual severity and the dependence between severities. Since the proposed model can distinguish the dependence between frequency and individual severity and the dependence between severities, the model can be applied to other field where the cluster size, the frequency in actuarial context, is informative.