Forecasting automobile insurance paid claim costs using econometric and ARIMA models

• Published in: International journal of forecasting Vol. 1; no. 3; pp. 203 - 215
• Main Authors: Cummins, J.David, Griepentrog, Gary L.
• Format: Journal Article
• Language: English
• Published: Amsterdam Elsevier B.V 1985; North-Holland; Elsevier; Elsevier Sequoia S.A
• Series: International Journal of Forecasting
• Subjects: Accuracy; Actuarial models; ARIMA models; Automobile insurance; Costs; Econometrics; Estimating techniques; Forecasting; Forecasting techniques; General insurance; Insurance; Insurance claims; Insurance rates; Kfz-Versicherung; Liability; Loss ratio; Mathematical models; Property damage; Rate setting; USA; Actuarial models; Automobile insurance; General insurance; ARIMA models; Forecasting; Insurance
• ISSN: 0169-2070; 1872-8200
• DOI: 10.1016/0169-2070(85)90003-2

Automobile insurance companies in the United States currently utilize simple exponential trend models to forecast paid claim costs, an important variable in ratemaking. This paper tests the performance of econometric and ARIMA models, as well as the current insurance industry method, in forecasting two paid claim cost series. The experiments encompass eight forecast periods ranging from 1974 through early 1983. The results indicate that automobile insurers could significantly improve their forecasts of property damage liability claim costs by adopting econometric models. For bodily injury liability claim costs, the accuracy of the econometric and insurance industry methods is approximately the same, and both outperform the ARIMA models. Overall, a net gain in accuracy could be achieved by adopting econometric models.