Adjusting Size for Loss Distributions for Trend

The aim of this paper is to discuss the methods that can be used to forecast the shape of a size of loss distribution in some future time period based on its shape in the recent past. In other words we will examine the ways one can adjust historical loss distributions for "trend". Section (1) of this paper provides a working definition and an example of the term "trend" as it is used in the ratemaking sense. Section (2) discusses trend in relation to size of loss distributions and indicates the basic model often used to adjust them for trend. Also included in this section are empirical methods that can be used to test the appropriateness of this model. Section (3) develops an alternate trend model and the paper concludes with Section (4) which exhibits the impact of this alternate model on increased limits pricing.