Natural Disasters, Insurance Stocks and the Numeraire Portfolio

This study employs a generalized numeraire portfolio to benchmark insurance stocks to detect abnormal returns in the presence of natural disasters and insurable losses. We show that under the benchmark approach the efficient markets hypothesis holds in the presence of extreme insurable loss whereas other common methods such as the market model and Fama- French three factor model often fail due to the accumulation of estimation errors. We construct a portfolio of U.S. insurance firms and observe the market reaction to a set of major insured natural disasters. Numeraire denominated or benchmarked returns are shown to be are natural measures of abnormal returns. Using the benchmark approach we observe no significant trend in the cumulative abnormal returns of insurance securities following a natural disaster. Using both the traditional market model and the Fama-French three factor model however, we observe significantly positive cumulative abnormal returns following an insured event. The errors inherent in the market model and three-factor model for event studies are shown to be eliminated using the benchmark approach.