Implications of Sampling Theory for Package Policy Ratemaking, [Discussion]

It is rare for a paper presenting essentially new material to have such broad scope as this one. The author begins with a lucid discussion of his statistical philosophy of ratemaking and insurance. Next, he presents two technical sections: a discussion of the value of stratification in sampling design, and an introduction to ratio estimation, with special consideration of its power for inference and the control of its bias. Then he applies both techniques to package policy ratemaking in general and to an example, the Special Automobile Package Policy. Leaving the personal lines, he reviews the Bailey, Hobbs, Hunt, and Salzmann paper, "Commercial Package Policies - Rating and Statistics, ""'1 mediates the indivisible premium-component rating debate, and shows that the key to the whole problem is a good multiple-line statistical plan.