Causal Relationships Between Premiums and Losses, and Causes of the Underwriting Cycles

This research explored two major insurance-market issues. First, it investigated the dynamic interactions between premiums and losses using vector autoregressive (VAR) models. Second, it showed how premiums respond to shocks to losses, surplus, interest rates, the variance in losses, and the variance in interest rates. New empirical results based on a decomposition of variance methodology suggest that changes in losses, surplus, interest rates, and uncertainty do explain substantial percentages of variations in premiums. Our results are substantially consistent with an extrapolation hypothesis and a hypothesis of rational expectations with an institutional lag, a capacity-constraint hypothesis, an interest-rate- fluctuation hypothesis, and a change-in-expectations hypothesis. Interestingly, current premiums were found to explain a substantial percentage of the variation in future losses in a two-variable system. This result is consistent with the predictions offered by rational expectations with institutional lags and some financial pricing models for insurance coverages. New evidence on the response of premiums to shocks in various financial variables is also provided by this study. In summary, the pattern generated by impulse response functions for premiums seem to support Cummins and Outreville's hypothesis of rational expectations with an institutional lag, a result which differs somewhat from the observations of Niehaus and Terry (1993).