Using Non-Insurance Information in Ratemaking - Cont.

Suppose that the distribution of the yields over time is known or can be estimated. For example, the actual yields for the past 20 years could be considered. The distribution of yields for the coming year can be estimated by applying trend factors to yields from past years. Using the known relationship between yields and pure premiums, estimates of the potential pure premium outcomes for the coming year can be determined. The average of these outcomes is an estimate of the expected pure premium.

While it is unlikely that this technique would be used as part of the standard ratemaking process, several aspects may prove useful. One simple use is to identify data processing errors by identifying years in which the losses and yields are not consistent. A second use is to estimate pure premiums when insufficient loss information is available. For example, when a new crop is introduced into a county, it may be possible to estimate the variability of the yields based on the variability of the yields for other crops or other counties. A third use of this technique would be in smoothing past experience. Since a large portion of the losses are produced in a few abnormally poor years, the number of abnormal years in the experience period has a strong influence on the average pure premium. Because the number of abnormal years is always an integer, the average pure premiums can increase or decrease sharply when an abnormal year enters or leaves the experience period. These fluctuations can be reduced by taking the distribution of yields or pure premiums into consideration. For example, the expected number of abnormal years can be used in place of the actual number, with the corresponding severity based on a larger body of experience.