Modelling and Estimating Individual and Firm Effects with Count Panel Data

In this article, we propose a new parametric model for the modelling and estimation of accident distributions for drivers working in fleets of vehicles. The analysis uses panel data and takes into account individual and fleet effects in a non-linear model. Our sample contains more than 456,000 observations of vehicles and 87,000 observations of fleets. Non-observable factors are treated as random effects. The distribution of accidents is affected by both observable and non-observable factors from drivers, vehicles and fleets. Past experience of both individual drivers and individual fleets is very significant to explain road accidents. Unobservable factors are also significant, which means that insurance pricing should take into account both observable and unobservable factors in predicting the rate of road accidents under asymmetric information.