Fines, Appeals and Liability in Public Enforcement with Stochastic Damage and Asymmetric Information

This paper studies an enforcement game between a regulator and firms that can cause harmful accidents. The distribution of potential accident damage is private information to the firms, and realized damage can be observed only at the cost of going to court. Under conditions described in the paper, an optimal policy involves the separate assessment of regulatory/settlement fines and court liability. In this optimum, injurers self-select by appealing (or not) to the court process; liability takes a 'threshold' form, assessing maximal liability when damages are high and zero liability otherwise; and, vis-a-vis a first-best, some firms are over-deterred and others under-deterred from having accidents