Applying Generalized Multivariate G-h Distributions And Regime-switching Copula To IBNR Reserves

In this paper we deliver our efforts to the analysis of the dependence between two variables in IBNR reserves. The two investigated variables are Claim sizes and Delay from the moment when claim occurred to the moment when the payment has been re-ported. We approximate the distribution of Claim size and the distribution of Delay with generalized multivariate g-h distributions, respectively. IBNR claim reserves have been calculated using Regime-switching Copula model. Bayesian inference is developed where model parameters are estimated simultaneously. This provides an essential difference from previous approaches in literatures where different Copula parameters are estimated separately in separate steps. The proposed Copula model makes it possible to sculpture the dependence structure among the variables which strongly influence the procedure of claims. The proposed approach is illustrated with insurance data from SK holdings.