Equity valuation and the determinants of equity risk in the property-casualty insurance industry

This dissertation consists of two empirical studies conducted in the setting of the property-casualty (“P&C”) industry. The first essay documents the relative explanatory power of alternative equity valuation models for the share prices of publicly traded insurers. I develop a dual valuation model that incorporates the underlying economic and accounting characteristics of P&C insurers, and investigate the influence of underwriting profitability on the role of underwriting valuation variables. I hypothesize and find that the dual model provides greater explanatory power for insurers' share prices than asset-only and earnings-only valuation models, suggesting that the stock and flow components in the dual model capture different aspects of equity value. Also as expected, underwriting profitability affects the economic and statistical significance of underwriting-related valuation variables. Consistent with prior studies, my results support the hypothesis that fair values are incrementally value-relevant to historical cost measures, however I do not find evidence that fair values offer relatively greater explanatory power than historical cost measures for P&C share prices.; The second empirical essay examines the relations between unpredictable accrual errors, equity risk, executive incentives, and various dimensions of risk management activity in the P&C industry. I empirically capture the endogeneity inherent in firms' risk management activities by modeling the determinants of insurers' equity risk and risk management choice variables in a simultaneous equation system. I extend the accruals literature by exploiting specialized accrual error disclosures in the insurance industry to develop proxies for the uncertainty inherent in firms' underwriting results. As hypothesized, my proxy for the uncertainty in the loss reserve estimation and reporting process is positively associated with insurer equity: risk, and with three corporate risk management choice variables: capital levels, reinsurance activity, and asset portfolio risk. Executives' personal incentives are not found to be incrementally significant after controlling for other economic determinants of equity risk. The results of my study have implications beyond the insurance industry; my findings suggest that simple OLS models that fail to incorporate the simultaneities inherent in firms' risk taking activities may be subject to severely biased coefficient estimates.