Models of the enforcement-compliance relationship have assumed that both the probability and magnitude of fines are independent choice variables of policy makers. These models indicate that it may be optimal to monitor with low frequency but to inflict uniformly maximal penalties for all infractions detected. This article shows that if the judicial system is built on the "reasonable doubt test, " then the penalty and the probability of conviction are not independent. In particular, as the penalty increases, the probability of conviction falls. As a result, uniformly maximal penalties may actually encourage crime rather than deter it. This article shows that optimal fines should rise with the severity of the infraction, that is, the penalty should "fit the crime."
Stored only when both papers are in the corpus — 266 citations exist upstream.