Variability of Loss Reserves

This paper addresses the issue of biases in the loss reserving process, some of which may be intentional. Using an empirical analysis of data from 169 companies over a seventeen year period, it is observed that the level of loss reserves exhibits cyclical behaviour, is different for companies of different sizes and is different for reinsurers than for direct insurers. Furthermore, after these factors are accounted for, the differences in levels between individual companies accounts for about three-quarters of the explainable variation. The paper suggests that greater independence on the part of the loss reserve specialist could lead to more objective estimation and could reduce historical variability by about 40%.