Abstract
The classical approaches to claims reserving problem are based on the limiting assumption that the claims in different years are independent variables. If this assumption is violated then the classical techniques provide incorrect pre- diction of the claims reserves.
Hence, methods that enable modeling the dependencies among the observations are needed. In this article we show the application of generalized estimating equations (GEE) for estimation of claims reserves, and we discuss the potential benefits of this approach.
It is shown that this approach leads to reasonable estimates of the outstanding loss liabilities. The method is illustrated on a Tweedie's compound Poisson model for the incremental claims since the Tweedie's family of distributions provides a flexible approach for modeling incremental claims within the framework of generalized linear models (GLM).
Details on computation and implementation in the R program are provided.