Type of Document Dissertation Author Gatzlaff, Kevin Michael Author's Email Address firstname.lastname@example.org URN etd-07122009-140707 Title Dimensions of Property-Liability Insurer Performance Degree Doctor of Philosophy Department Risk and Insurance, Department of Advisory Committee
Advisor Name Title James Carson Committee Co-Chair Kathleen McCullough Committee Co-Chair Cassandra Cole Committee Member Jeffrey Clark Committee Member Randy Dumm Committee Member Richard Corbett Committee Member William Christiansen Outside Committee Member Keywords
- Insurer Insolvency
- Financial Strength
- Quantile Analysis
- Insurer Performance
Date of Defense 2009-06-19 Availability unrestricted AbstractThis dissertation examines several aspects of insurer performance. Much attention in the literature has been paid to insolvency; however, most firms are neither insolvent nor near insolvency. In the past, states have restricted insurer capacity by regulating maximum premium-to-surplus ratios. While high levels of premium-to-surplus have been shown to be associated with insolvency, it is unclear if high levels of premium-to-surplus ratios damage top performers, which may mean that for some firms, regulation of this ratio is unnecessary. Similarly, other factors known to influence insolvency may have little or no impact on top performers. In Essay 1, I analyze whether factors known to influence property-liability insurer insolvency are associated with top performance through the use of quantile analysis. I further investigate the presence of non-linear relationships between performance and these factors, as revealed by changes in the direction of the relationship of the factors and performance as performance quantiles advance. There are two main findings from this essay. First, I find statistically significant differences between the coefficients of the independent variables across performance quantiles. Second, I find evidence that for some variables, such as leverage, concentration, and reinsurance ceded, the direction of the correlation changes as performance quantiles advance. Taken together, these findings indicate that factors associated with insolvency are not necessarily associated with top performance. Researchers and others examining the performance of a cross-section of insurers should take note that the impact of factors on performance changes as performance improves/declines.
Previous literature has assumed that hedging motivations explain the use of derivatives by property-liability insurers. In the second essay, I examine whether property-liability insurers use derivatives as a substitute for reinsurance. I also analyze if insurers use derivatives to generate income in addition to that earned through underwriting and traditional investments. I examine the relationship between derivatives and reinsurance in order to evaluate what I term the substitution hypothesis, which posits that insurers use derivatives as a substitute for reinsurance. I examine the relationship between derivatives usage and return on assets to evaluate support for what I term the income-generation hypothesis, which posits that insurers use derivatives primarily to generate income in addition to what they could normally expect to earn from underwriting and traditional investment sources.
Since decisions about the use of derivatives and reinsurance are made simultaneously, I account for endogeneity when evaluating the relationship between derivatives and reinsurance by instrumenting the demand for reinsurance based on factors known to influence it. Because the characteristics of firms using derivatives can vary considerably from those firms that do not, the possibility of selection bias exists. I use two approaches to control for potential selection bias. Since most property-liability insurers use no derivatives, a series of Tobit equations is first employed, using four different measures of derivatives usage to determine the relationship between derivatives and reinsurance. In all four specifications of the Tobit model, I find a negative and significant relationship between derivatives and reinsurance, indicating support for the substitution hypothesis. I find no relationship between return on assets and derivative usage, indicating a lack of support for the income-generation hypothesis.
For robustness, a two-stage Heckman technique is used to evaluate the substitution hypothesis and the income generation hypothesis. Again I find a negative and significant relationship between the instrumented reinsurance variable and derivatives usage, indicating support for the substitution hypothesis. I also again find no relationship between return on assets and derivatives usage, which indicates a lack of evidence for the income-generation hypothesis.
Filename Size Approximate Download Time (Hours:Minutes:Seconds)
28.8 Modem 56K Modem ISDN (64 Kb) ISDN (128 Kb) Higher-speed Access GatzlaffKDissertation2009.pdf 924.36 Kb 00:04:16 00:02:12 00:01:55 00:00:57 00:00:04