Loss Reserving Chain Ladder Methods Applied to a Small Midwestern Insurance Company

Abstract

Estimating future losses is integral to setting aside appropriate reserves in the insurance industry. This study analyzes different Chain Ladder reserving methods based on weighted-least square regression that consider different function of weights. These methods are tested on 78 NAIC fully developed loss triangles. While the CRE Chain Ladder method is selected based on its performance, this method does not work well for a small number of NAIC companies that may have erratic changes in their loss trends. For these outliers, two other methods were explored for the early development years; the nearest neighbor technique and mixture of linear regressions. A recommendation is then made to a small Midwestern insurance company on the best methodology to use for estimating the loss reserves based on the actual data provided. These results can be useful to any other insurance company currently using Chain Ladder methods in loss reserving practices.