The purpose of this paper is to be an academic inquiry into rating issues confronted by the US Federal Crop Insurance program stemming from changes in participation rates as well as the weighting of data to reflect longer‐run weather patterns.
The authors investigate two specific approaches that differ from those adopted by the Risk Management Agency, building upon standard maximum likelihood and Bayesian estimation techniques that consider parametric densities for the loss‐cost ratio.
Both approaches indicate that incorporating weights into the priors for Bayesian estimation can inform the distribution.
In most cases, the authors' results indicate that including weighting into priors for Bayesian estimation implied lower premium rates than found using standard methods.
