Predictive Actuarial Modeling of Health Insurance Claims Costs

Abstract

This study constructs predictive models that can be used in forecasting future health care cost of in-patient diagnoses with high incidence of occurrence for effective health care intervention. Using various forecasting techniques, such as auto-regressive-integrated-moving average and regression, the study aims to find the best fit for data for various health conditions so as to determine the best out-of-sample forecast. The results suggest that while regression modeling was not particularly suitable for the data, Box Jenkins models provide a good fit. The latter models can therefore be used with high degree of accuracy in predicting future costs, thereby facilitating more effective intervention in health care management.