Analysis of Claim Counts Using Hurdle Generalized Poisson Regression Models

This study introduces the functional form of hurdle generalized Poisson (HGP-P) regression model and its applications in insurance risk classification. The HGP-P model generalizes the HGP-1 and HGP-2 (Saffari et al. (2013)). In particular, we analyze the French motor third party liability insurance data using the proposed model, and compare the results with the functional forms of zero-inflated generalized Poisson and the zero-inflated negative binomial regression models. Our analysis shows that the proposed HPG-P model is useful in dealing with insurance loss count data, which has a high proportion of zeros and is over-dispersed.

Reference:

Saffari, S. E., Adnan, R., & Greene, W. (2013). Investigating the impact of excess zeros on hurdle-generalized Poisson regression model with right censored count data. Statistica Neerlandica, 67(1), 67-80.