The accidents in the automobile insurance: an applied econometric analysis

The objective of this paper is to determine what the most important factors for the number of accidents in the automobile insurance are. We try to verify if the theory is right and the highest levels of coverage are positively correlated with a greater number of accidents. Data used in this empirical analysis have been provided by a private insurance company. In order to estimate the number of accidents, we apply a recent approach for considering count data situations, not very used in the automobile insurance: a zero-inflated negative binomial model. So we can make a distinction between the null observations from drivers that do not declare any accidents and the ones recorded when truly no accident happens. We finally show what types of vehicles suffer a greater number of accidents, what use they are devoted to, and how much related are accident variables and others: the age, the policyholders' driving experience, the premiums paid, or the different levels of insurance coverage.