A Composed-Error Model for Estimating Pest-Damage Functions and the Impact of the Western Corn Rootworm Soybean Variant in Illinois

We describe a composed-error model for estimating pest-damage functions. The composed-error model, originally developed to account for statistical noise when estimating technical efficiency, removes the effect of experimental errors when estimating the variance of yield loss from pest damage. As a result, the estimated variance of yield loss is often less than for a conventional model, which has economic implications when the analysis incorporates risk aversion. We find that, depending on the level of risk aversion, the western corn rootworm soybean variant reduces farmer certainty-equivalent returns 16–26% more with a conventional model than with the composed-error model.