Integrated modelling for non-life insurance using individual claim reserving

In this paper, we introduce an individual claim reserving model that uses policy characteristics and information on the claim process to estimate the reserves of the Incurred But Not Reported and Reported But Not Settled claims of a non-life insurer. First, we analyse the optimal model specification for the claim process. We use mixture distributions to model the level of the payments and introduce zero adjusted distributions to model claim payments of zero. Second, we study the effect of creating dependence between parts of our model on the respective fit. In addition, we examine whether including policy characteristics increases our model fit. We find that the inclusion of both policy characteristics and claim process characteristics improve the fit of our individual claim reserving model. Based on simulations, we examine the effect of incorporating covariates in the model on the prediction of IBNR and RBNS reserves. We conclude that the inclusion of covariates increases the accuracy of our individual claim reserving model. Our model can be used by insurance companies to coherently estimate their reserves.