Denenberg [7] hypothesized that Best's financial ratings provided an accurate estimate of those insurers that would not become insolvent. Several researchers, most notably Pinches and Trieschmann, sought to quantify those predictions through the use of multivariate discriminant analysis. The authors of this paper incorporate Best's ratings into the discriminant analysis through a system of dummy variates. Best's ratings are then compared to the results obtained by the use of financial variables. Finally, a two-stage discriminant technique is introduced and its results are shown to be better for predicting insolvency for property-liability firms. Denenberg created an heuristic analysis of solvency prediction for the insurance industry [7]. Seeking to formalize the common practices of agents and analysts, he examined the Best's ratings for size and financial strength for the six years preceding insolvency. His methodology confirmed the hypohthesis that Best's financial ratings were useful in predicting future solvency status of both life and property-liability firms. Pinches and Trieschmann [13] made several improvements on solvency prediction. They considered a wide range of financial ratios in the context of multivariate discriminant analysis. The improvements indicated which financial characteristics were most highly correlated with insolvency and provided a framework in which the statistical validity of the conclusions could be objectively determined. Harmelink [9] used the same multivariate technique to predict the Best's general policyholder rating rather than insolvency. His Jan Mills Ambrose is a Ph.D. candidate at The Wharton School of the University of Pennsylvania. J. Allen Seward is a Ph.D. candidate at the Wharton School and Assistant Professor of Insurance and Finance at the Hankamer School of Business, Baylor University.. The authors gratefully acknowledge the support of the S.S. Heubner Foundation as well as the helpful comments provided by J. David Cummins, two anonymous referees, and the editor of this journal. This content downloaded from 157.55.39.127 on Wed, 29 Jun 2016 04:22:39 UTC All use subject to http://about.jstor.org/terms 230 The Journal of Risk and Insurance analysis did not attempt to determine whether or not the ratings were indeed a significant indicator of insolvency. A different technique to study the same problem was employed by Harrington and Nelson [10]. An insurer's premium-to-surplus ratio was regressed against a range of insurer characteristics. Analysis of the residuals from the regression equations was then used to forecast insolvency. The results of this approach were mixed when contrasted to the performance of the Insurance Regulatory Information System (IRIS) ratios. The purpose of this paper is to incorporate Best's general policyholder rating and financial size rating' with variables created from a firm's readily available financial information. The rating variables are then used to alter the prior probabilities of classification under multivariate linear discriminant analysis. The organization of this study is as follows: The methodology and sample selection criteria for the subsequent statistical analysis are presented. The results of the tests of three hypotheses for predicting insolvencies are given, and conclusions and recommendations are made. Research Methods and Sample Description
[1]
H. S. Denenberg.
Is "A-Plus" Really a Passing Grade? Insurer Risk Capacity and Financial Ratings: The Use of Financial Ratings of Insurance Reporting Services to Avoid Potentially Insolvent or Otherwise Delinquent Insurers
,
1967
.
[2]
D. F. Morrison,et al.
Multivariate Statistical Methods
,
1968
.
[3]
E. Deakin.
Discriminant Analysis Of Predictors Of Business Failure
,
1972
.
[4]
G. E. Pinches,et al.
A Multivariate Model for Predicting Financially Distressed P-L Insurers
,
1973
.
[5]
G. E. Pinches,et al.
The Efficiency of Alternative Models for Solvency Surveillance in the Insurance Industry
,
1974
.
[6]
H. S. Diamond,et al.
Pattern recognition and the detection of corporate failure
,
1976
.
[7]
Leo Breiman,et al.
Classification and Regression Trees
,
1984
.
[8]
C. Zavgren,et al.
The prediction of corporate failure: The state of the art
,
1983
.
[9]
E. Altman.
Corporate financial distress : a complete guide to predicting, avoiding, and dealing with bankruptcy
,
1983
.
[10]
H. Frydman,et al.
Introducing Recursive Partitioning for Financial Classification: The Case of Financial Distress
,
1985
.
[11]
Jack M. Nelson,et al.
A Regression-Based Methodology for Solvency Surveillance in the Property-Liability Insurance Industry
,
1986
.