Potential Enhancements to Ratemaking Techniques

The previous section considered the use of non-insurance information in testing or modeling insurance experience. However, non-insurance information is generally considered to be supplementary to rather than as a replacement for insurance experience. The following discussion considers several approaches to improving the accuracy and increasing the stability of the rates based solely on insurance experience.

One recent proposal for improving the accuracy of the rates is to create fixed rating territories consisting of adjacent counties with similar agronomic characteristics. The rating territories could vary depending upon the crop being rated. The rationale for this proposal is that it would eliminate a perceived problem with the current concentric circle technique, due to the potential inclusion of experience from neighboring counties having dissimilar agronomic characteristics.

A more technically demanding approach is known as spatial smoothing. For example, the current concentric circle method is a simple form of spatial smoothing. A more sophisticated approach, known as locally weighted regression smoothing, has previously been introduced into Crop Hail insurance ratemaking by Dr. Michael Lewis. The advantages of this technique are its ability to produce smoother results than the concentric circle method and its ability to take spatial correlation into account. This technique may also eliminate the need to spread excess losses for each county across the state. A detailed explanation of the spatial smoothing process is included in the appendix.