Punishment causes reputational losses in addition to more tangible losses. Lowering the probability of punishment reduces these reputational losses by diluting the informational value of verdicts. These considerations better align the positive as well as normative implications of law enforcement models with intuition and empirics: Crime is more responsive to the certainty rather than the severity of punishment even absent risk-seeking offenders (positive), which causes extreme Beckerian punishments to be inefficient when sanctions are socially costly to impose (normative). Moreover, in some cases optimal enforcement is 'anti-Beckerian': Punishment is symbolic and detection costs are incurred solely to provide reputational incentives.
Claude Fluet : Université Laval, CRREP, CRED
Murat C. Mungan : George Mason University, Antonin Scalia Law School