A Nonparametric Approach to Evaluating Reinsurers' Relative Financial Strength

There have been a number of past attempts aimed at taking financial data on individual companies, and using this data to produce a predictive model of insurance company solvency. These models have come in two forms: parametric and non-parametric. For example, the NAIC, with its eleven Early Warning Tests, is taking a non-parametric approach to this problem, while the AIA has used a parametric approach in producing its formula for assessing an insurer’s financial strength. In trying to evaluate a reinsurer’s financial strength, however, these two systems have several shortcomings. The first is that all past models were developed by analyzing a primary company data base, and it is not clear that a model created for primary companies will be effective when applied to reinsurance companies. the second is that even when applied to primary companies, the predictive power of these models are questionable. Our model for assessing a reinsurer’s financial strength is non-parametric. By using five years of data (1980-84) and two sets of ten companies (financially strong vs. financially weak), we identified those financial ratios which have historically discriminated between the strong and weak companies. These historically consistent discriminators have been used to produce a current ranking of reinsures based on relative financial strength. Based on this model, most of the companies which currently have poor Best’s ratings are identified as financially weak using 1981 data, which is a far better performance than any of the existing solvency-tracking models. Going forward, we feel that given similar economic/insurance industry conditions as existed during the early 1980’s, this model will do a good job of predicting future problem reinsures. Reinsurance Research - Risk Loads/Profitability