An Experience Rating Approach to Insurer Projected Loss Ratios

The traditional approach to Property/Casualty rate indications starts with a methodology that uses internal data to forecast the Ultimate Loss Ratio, with losses making up about half of the expenses. For parties that are external to the insurer, this approach to forecasting a key component of future profitability is impractical as they generally do not have access to the necessary data. Using publicly available information, that is, the National Association of Insurance Commissioners Schedule P of the statutory financial statements from 1992 to 2010, we develop by line of business forecasts of the relativity to the industry Loss Ratio. To develop these forecasts, we use a weighted regression methodology that incorporates key ideas from fixed-effects regression, instrumental variables regression, credibility theory, as well as a flexible covariance structure for the residuals. Results indicate that the proposed approach of using lagged relativities from insurer own and other lines of business can provide adequate fits for many lines of business and for the combined results of the insurer as a whole.

Keywords: Experience Rating, Panel Data, Fixed-Effects Regression, Instrumental Variable Regression, Credibility Theory