CMStatistics 2017: Start Registration
View Submission - CFE
Title: Censored copula modeling for micro-level reserving in non life insurance Authors:  Olivier Lopez - Sorbonne Universite Paris (France) [presenting]
Abstract: In non life insurance, two main strategies exist to evaluate the amount of reserves required to cover the claims of the customers. Chain-Ladder approaches use aggregated information to forecast the amount to be paid, but they miss some important individual information one has on the claims. On the other hand, micro-level reserving aims to use each available information on the claim to produce individualized predictions of its final amount. Nevertheless, this second class of methods usually fails to achieve a reasonable performance for guarantees with potential large duration between the occurrence of the claim and the final payment (lifetime of the claim). We present a general method to address this issue, by modeling separately the hazard rate of the lifetime of the claim, and its amount, using a censored copula methodology to capture the dependence structure of these two variables.