Identifying life insurer financial distress: Parametric and nonparametric classification models, insolvency theories and empirical evidence

This dissertation provides empirical evidence on the strength of three types of bankruptcy detection models: multiple discriminant analysis, logistic regression, and recursive partitioning (RP), as applied to the life insurance industry.; A further focus of the dissertation is the application of bankruptcy theory to aid in the selection of important variables with respect to the problem of financial distress for life insurers. The study utilizes NAIC financial data for a sample of 1,605 solvent insurers and 80 insolvent insurers that either did or did not fail between 1986 through 1992.; Empirical models suggest that surplus measures and leverage measures are strong indicators of insurer financial strength. While no evidence is found for a strong relationship between state capital requirements and insolvency, weak evidence for stock insurers being riskier than mutual insurers is present.