Empirical Bayesian Credibility for Workers' Compensation Classification Ratemaking

This paper demonstrates how a company can derive accurate classification relativities. The method uses an empirical Bayesian credibility formula as taken from the paper "Credibility for Loss Ratios" by Buhlmann and Straub and modified by the IS0 Credibility Subcommittee. The data required for this method can be purchased from the National Council. A classification review is performed on three years of live data, Relativities predicted by both this method and the present ratemaking formula are compared with the actual relativities from a fourth year of data.