Files
Abstract
An important issue in the agricultural actuarial literature is the extent to which sample period selection affects the accuracy of insurance rating. A conditional Weibull distribution approach is developed which explicitly models the interaction of weather, technology, and other variables on
probabilistic yield outcomes to address this issue. Results from an application with an extensive producer-level yield dataset representing commercial-scale Illinois firms suggest that the impact of weather heterogeneity on risk estimation across reasonable samples is likely not as great as is often claimed. The results also suggest that yield risk is decreasing significantly through time, and indicate the presence of trend acceleration. A rating analysis indicates that violations in the
risk evolution assumptions of the rating approaches used in the Federal Crop Insurance
Program—which implicitly assume increasing yield risk through time when yields trend—result in severely biased rates, with typical overstatements of 200% to 400% for Midwest corn.