When the fine structures for speeding offences are observed, it is often found that fines depend on speeders' offence history. In this paper we devise two fine structures: a uniform fine, and a fine which depends on offence history. If drivers differ in their expected accident costs, the literature prescribes that the fine for bad drivers should be higher than for good drivers. However, governments do not know the type of driver. We develop a model where the number of previous convictions gives information on the type of driver. We find that the optimal fine structure depends on the probability of detection, and on the strength of the relationship between the type of driver and having a record. We illustrate this by means of a numerical example and show that, for reasonable values for the probability of detection, a uniform fine is preferred. (Author/publisher)
Abstract