IDENTIFYING FINANCIAL DISTRESS IN THE INSURANCE INDUSTRY - A SYNTHESIS OF METHODOLOGICAL AND EMPIRICAL ISSUES

The problem of insolvency in the insurance industry is well recognized. Numerous approaches for identifying financial distress have been considered. Some of these procedures have statistical limitations that have not been addressed in applications for the insurance industry. This article reviews the methodological issues and analyzes a nonparametric qualitative response model as well as a generalized qualitative response model that circumvents some of the statistical limitations. The generalized models include the probit and logit models as special cases. The results of applying these robust procedures to a large data base of insolvent and solvent insurers are compared to those obtained from applying multiple discriminant analysis and probit and logit models. Comparisons of models are made on an estimation sample and a holdout sample. The more robust qualitative technique provides important advantages, although several procedures provide similar results.