Cost minimizing sequential punishment policies for repeat offenders

This paper concludes that, when offenders are wealth constrained and the government is resource constrained and can commit to a certain policy throughout the whole planning horizon, cost minimizing deterrence is decreasing, rather than increasing, in the number of offenses. By extending the framework, suggested in Emons (2003), to n-periods setting, we prove that for the agents who may commit an act several times, optimal sanctions are such that the fine for the first crime equals the offender’s entire wealth, and the fines are zero for all the subsequent crimes. This result contradicts the widely prevailing escalating penalties imbedded in many penal codes and sentencing guidelines. In addition, analogous to Emons (2004), this scheme does not appear to be a time consistent (subgame perfect) strategy for the government in an n-periods setting. Moreover, we show that, if the government cannot commit, equal rather than decreasing sanctions will be optimal. JEL-Classification: D82; K41; K42