The Determinants of Punishment

National Bureau Of Economic Research Inc, Cambridge MA

Investigators

Abstract

The economic approach to punishment in criminal cases suggests a calculus where sentence length is determined to minimize costs due to crime (including the costs of punishment). While there is a long literature on the theoretical implications for law enforcement, there is much less research on the extent to which sentencing in practice follows the implications of the Becker (1968) framework. This project tests whether the basic optimal deterrence framework actually explains behavior in homicide sentencing. In particular, it examines whether sentences are higher for types of homicide where probability of apprehension are low and higher for types of homicide where recidivism is high. Our most significant test of the Beckerian framework is to see whether victim characteristics matter as strongly for vehicular homicides as they do for other types of murder. In vehicular homicides, victims are not completely randomly chosen, but they are much more random than in most other forms of murder. Basic deterrence theory suggests that the connection between victim characteristics and sentences should be much weaker in the case of vehicular homicides. Preliminary work suggests that this is not the case and the victim gender and race are just as important with vehicular homicides as they are for other forms of murder. One major objective of this research proposal is to assemble a larger database of vehicular homicides with which to further test this hypothesis. A second advantage of looking at the impact of victim gender and race in vehicular homicides is that it is much less likely that these characteristics proxy for omitted characteristics of the criminal or the crime. Thus, for example, it is possible that men who kill women in standard settings are just generally more dangerous and they receive stiffer because of higher recidivism rates. This argument is harder to make in the case of vehicular homicides. The gender effect in those types of homicides is more likely to represent a general societal inclination to particularly punish people who kill women. A second objective of this research proposal is to test for the existence of agency problems, which may interfere with optimal deterrence. The presence of judge or jury effects in sentencing suggests that optimal punishment may not be happening because individuals are maximizing their own welfare not that of the community as a whole. To the extent to which there is this form of heterogeneity, this information may be useful in considering optimal reform of the legal system. Following in the vein of the positive approach to punishment, this proposal also suggests a second research project examining the rise of incarceration in the late 19th century. A simple model, sketched in the proposal, suggests that the rise of urbanization lead to professional criminals and greater heterogeneity among criminals. As incapacitation becomes more important, relative to deterrence, when heterogeneity among criminals rises, this might explain the rise in jails in the early 19th century.