Reducing Crime by Shrinking the Prison Headcount

The punishment of offenders is intended (among other purposes) to control crime. Yet the current situation in the U.S. combines high crime and high punishment. Large increases in punishment have been followed by comparatively modest decreases in crime: as incarceration has quintupled, crime has merely halved. Since the punitiveness of the system—as measured by the expected value of punishment per crime committed—depends on the ratio between punishment and crime, punitiveness has increased by something like an order of magnitude. The failure of such a drastic increase in punitiveness to produce a corresponding decrease in crime presents a puzzle: if punishment prevents crime, how is it that we find ourselves saddled with so much crime despite handing out so much punishment? Of more practical interest, what can be done to shrink both the crime rate and the prison population? This paper proposes both an analysis of that explanatory question and a set of policies based on that answer and designed to reduce both crime and incarceration, and to do so over a period measured in years rather than decades. We have too much crime and too much punishment because we use punishment badly, in ways that both neglect what we know about how individuals react to punitive threats and also ignore both the risks and the uses of positive feedbacks. The principles of cost-effective crime control can be expressed as “Five C’s”: Certainty and Celerity of punishment (rather than severity); dynamic Concentration of punitive resources on subsets of offenders and offenses; the direct Communication of deterrent threats to those whose problematic behavior we seek to change, and the maintenance of the Credibility of those threats by relentlessly delivering the promised sanctions. It hardly needs saying that the success of the proposed policies would depend not only on the accuracy of the underlying analysis but