An assessment of the determinants of Mauritian automobile insurance claims using negative binomial and gamma regression models

This paper focuses on the factors that influence the Mauritian automobile insurance claims; a process known as automobile ratemaking. The response variables are the average claim frequency counts and the claim severity. These two components are then combined to provide the pure premium. Generalized Linear Models are used to measure the influence of some predictors on these two response variables. More specifically, the Negative Binomial and Gamma regression models are used to fit the claim frequency and severity respectively. The descriptive and inferential analysis was done using the R Software and was based on robust statistical criterions such as G-Statistics and Akaike Information Criterion. Gender, claim condition, age of the client, accident frequency and servicing frequency are the essential determinants of the claim frequency while claim condition, age of the client and vehicle's make are the significant predictors of the claim severity.