Abstract
Revising a classification structure requires both underwriting judgment and an analysis of statistical data. The data can be used to estimate the loss experience of the various proposed classifications and to compare it with the loss experience of the existing classifications. When the merging of two classification is contemplated, the data on dand can be sued to compile experience on both separate and combined bases and to test the effect of the merger. However, when the splitting of the established classifications is under consideration, the proposed classifications are often not identified in the data base, so a special effort is required to obtain the necessary data. This may involve obtaining additional information from the applications, imputing classification data from other data on hand or conducting on-site inspections of a number of risks. This paper illustrates a method for revising a classification system in which a mail survey was used to obtain statistical data.
Volume
LXI
Page
103
Year
1974
Publications
Proceedings of the Casualty Actuarial Society