It is common for actuaries to think about claim related variables for the target of prediction. Understanding losses is of primary importance for actuarial work. There are, however, many other times when predictive modeling can help the practicing actuary understand the dynamics of other items of interest.

These situations could be where the actuary truly wants to predict future outcomes, or more simply wants to understand key dynamics in the multivariate world of insurance. Join Chris Gross as he discusses some examples of applying predictive modeling techniques to non-claims questions. Some of the examples that will be discussed include:

• Building models of collected premium as a more complete alternative to traditional rate monitoring, and to shed light on use of discretionary pricing
• Building a model of policy level underwriting expense
• “Predicting” claim department case reserves to highlight and test changes in adequacy, as well as shed light on different treatment for different types of claims
• Building a model on the back end of a simulation model to highlight the drivers of the simulation results