| Financial and statistical issues associated with solvency surveillance in the property/casualty insurance industry are addressed. The lack of proper utilization of insurance data bases and shortcomings in sampling methodology employed in most previous insolvency-prediction studies are demonstrated.; An overall financial description of the annual statement pertinent to the insurance data base at the University of Texas at Austin is initially provided. Then, the procedures employed to create the master data sets, the basis of all the financial-empirical analyses in this study, are carefully described.; The effectiveness of the various statutory accounting ratios in distinguishing between solvent and insolvent insurers is examined extensively utilizing data from the A. M. Best's Standard Tapes (1980-1989), which almost cover the entire population of companies in terms of premiums. These ratios include the Insurance Regulatory Information System (IRIS) ratios and others developed in the study. The raw data base is manipulated in a unique and sophisticated way to demonstrate and correct potential problems introduced by the traditional analyses performed in previous insolvency studies.; There have been a number of insolvency-prediction studies for the property/casualty insurance industry. However, most previous studies suffered from several methodological shortcomings that tend to make their statistical predictions unreliable. Among the problems, the most serious one is related to the sampling methods employed in these studies. Extensive computer simulations based on empirical data, that have not been attempted in previous bankruptcy studies for any industry, are performed to examine potential problems related to sampling methods. The results from these simulations show that the high prediction accuracy produced by any insolvency study that employs a single sample with a large insolvency-frequency rate (the ratio of the number of insolvent to solvent companies in the sample) may not be reliable since the reported accuracy rate could be highly sample-specific. The methodology developed here is completely new and has great practical significance for future insolvency studies in any industry. |
