Font Size: 

We model the dependency structure in the premium and reserve risk sub-module determining the Solvency Capital Requirement (SCR) and the diversification effect (DE). We use the Deep Neural Network (DNN) to estimate marginal distributions modeling the premium and reserve risk of non-life insurance segments, and a copula defining the multidimensional relationship between segments. We use the energy distance to evaluate the error of fitting the copula to the real data. The determined DE when modeling dependencies using the copula method estimated by the use of DNN is compared with DE when modeling dependencies using the method proposed in the Solvency II Directive and using C-vine copulas. The obtained test results indicate that the use of DNN allows for more accurate modeling of the dependency structure, and the determined DE is at the appropriate level.