Structured Tools to Help Organize One’s Thinking When Performing or Reviewing a Reserve Analysis

Actuaries regularly update the results of prior analyses that leverage more current information. Actuaries will often apply similar methodologies and thought processes from the prior analysis to the current one. In doing so, actuaries are employing techniques that help them to evaluate the reasonability of prior assumptions as compared to the most recent data emergence and using judgment to update assumptions ranging from selection of loss development factors to initial expected loss rates to point estimates or ranges of ultimate losses. However, actuarial literature to date provides little guidance on the questions that one can ask during each of these steps and the calculations that can be done to help bring a repeatable rigor to the analysis being done.

This paper will identify three distinct series of exercises that can be performed to help bring just such a repeatable rigor to the analysis. Along the way, the exercises will help the actuary frame answers to the following questions:

1. How did losses emerge between the prior review and the current review in relation to what was expected to emerge?

2. Are the selected loss development factors (LDFs) generally in line with the patterns in the underlying data triangles?

3. What is driving the change in ultimate loss estimates from the prior to the current analysis? Is it data (i.e., loss emergence), change of assumptions (i.e., loss development factors or initial expected loss rates), or change in judgment (i.e., the manner in which a point ultimate is chosen relative to a paid or incurred ultimate loss projection for a given accident period)?

By giving actuaries a structured and repeatable methodology to apply in search of answers to these questions, we are providing actuaries with a framework that will bring them a structure to their analyses and help them to identify areas in their analyses that might benefit from further investigation and study.

Keywords: reserving, suitability testing, data diagnostics