|
Abstract:
|
We will look at two examples of statistics impacting business decisions. The first discussion will be centered around the business impact generated by the implementation of a fraud predictive model. The model was developed in an insurance company to aid the Special Investigation Unit (SIU) in selecting claims for investigation from a pool of exiting open claims. The history of past assignments processed by the investigating adjusters were reviewed and the model mimics their selection thought process. As a result, the work flow and in some cases, job responsibilities have changed in the SIU. In the second discussion, we estimate the effects of premium changes on retention/renewals on existing policies based on discrete choice models. Modeled elasticities vary by segments and depend on the amount of premium change. We give one useful business application -- estimating expected renewal rates and aggregate premium changes due to rate or system changes. We also discuss issues in applying these models such as the timing of data and the presence of 6 and 12 month policies.
|
