Catastrophic Shocks in the Property-Liability Insurance Industry: Evidence on Regulatory and Contagion Effects

Article excerpt

INTRODUCTION

In the past few years, natural disasters in the United States have resulted in huge losses, much of which were recovered from insurers.(1) Given the magnitude of these losses and the burdensome regulations in the industry, it is essential to analyze the financial and regulatory impact of such catastrophes on the insurance industry. We take a step in this direction by examining the effects of Hurricane Andrew and its regulatory aftermath on the stock prices of property-liability insurance firms. As Schwert (1981) suggests, tests with stock price data are more powerful than other measures because stock prices are more accurate and they incorporate all relevant information as soon as they become available; also, the existence of well-specified models of expected return allow us to isolate firm- and industry-specific effects from marketwide movements.

Insurance experts have suggested two opposing, but not mutually exclusive, effects of major natural disasters on insurance firms. The first and obvious effect is that they cause losses to insurers due to the large reimbursements paid to policyholders for their damages. The expectation of these losses should cause insurance stocks to decline at the time of the disaster (the negative effect). But there is another more subtle effect that such disasters are hypothesized to have on insurance firms: they lead to larger profits because of increases in future premiums. Different theories have been proposed to explain such premium increases. Cagle and Harrington (1995), Cummins and Danzon (1996), Gron (1989, 1990), and Winter (1992) analyze models where the supply of insurance is an increasing function of the insurer's capital. Natural disasters reduce the capital of insurance firms, thus reducing their short-run supply of insurance and causing premiums to increase. Another explanation for higher premiums is that there is an easing of regulatory pressures on insurance firms following these disasters. This allows firms to raise premiums without incurring significant opposition from the regulators. For example, Forbes magazine (January 8, 1990, p. 184) wrote in the wake of Hurricane Hugo and the California earthquake that "Insurers are expected to use the two disasters as an excuse to boost rates modestly in 1990 - and few people will fight the increases." Finally, Gron (1990) and Shelor, Anderson, and Cross (1992) argue that natural disasters increase the demand for insurance from previously uninsured consumers, leading to premium increases. The expectation of higher premiums and profits will cause insurance stock prices to increase at the time of the disaster (the positive effect) or at the very least to partially offset the negative effect. Therefore, the net effect of a natural disaster on insurance stocks would be positive, negative, or zero depending on the relative strengths of these two effects. A corollary to the hypothesis about higher premiums is that, if there is an unexpected tightening of the regulatory environment following the disaster, insurance stocks should fall since the premium increases expected at the time of the disaster are now less likely to occur.

This article tests for the presence and relative strengths of these two effects by examining the impact of Hurricane Andrew on the stock prices of 48 publicly-traded property-liability insurance firms. We chose Hurricane Andrew because it is the costliest natural disaster ever to occur in the United States, making it relatively easy to detect the two hypothesized effects (if they are present). We use two techniques to measure Andrew's impact on insurance stocks: a generalized least squares (GLS) estimation technique based on Zellner's (1962) seemingly unrelated regression model and the standard event study cumulative abnormal returns (CAR) tests adjusted for cross-sectional dependence. The results from both techniques indicate that Andrew had a large and significantly negative impact on insurance stocks with a three-day CAR = -1. …