Impacts of Weather and Time Horizon Selection on Crop Insurance Ratemaking: A Conditional Distribution Approach

An important issue in the agricultural risk and actuarial literatures is the extent to which sample period selection affects the accuracy of insurance rating. This is typically more problematic in agriculture than in other types of insurance because of the variation in weather through time. A conditional Weibull distribution approach is developed that 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.