GAME MODELS FOR STRUCTURING MONITORING AND ENFORCEMENT SYSTEMS

This paper characterizes current enforcement of U.S. Environmental laws as very likely inadequate, while admitting that proof of this proposition would be extremely difficult exactly because there is so little information on what actual behavior of regulated firms and government activities is. Obvious avenues of improvement—large increases in monitoring and enforcement budgets, and credible use of draconian fines that would compensate for low probabilities of detection—are dismissed as politically infeasible. One path to improvement is explored that involves thinking of the problem as a game between pollution source and environmental protection agency. A single‐play game with and without errors of monitoring measurement forms the benchmark for discussion of a multiple‐play game in which the source's past record of discovered violations determines its future probabilities of being monitored. In particular, failing two monitoring visits in a row subjects a source to a long period in which monitoring is frequent enough to make continuous compliance optimal. The prospect of this sentence can be made sufficiently unpleasant that sources will choose to comply after their first failed monitoring visit, even though the likelihood of the next visit remains below that required to make compliance optimal in the single‐play context. In the long run, with proper system design, the bulk of the sources will be in that situation—complying after one failed monitoring visit while subject to quite low probabilities of the next visit. It is shown that this multiple‐play game approach can yield a significant savings in meeting a goal defined as limiting the extent of violation to a particular frequency in the population of polluters. Questions for further research are identified.