May 19, 2009 Deterrence is often a stated goal of criminal sentencing guidelines, but there is debate about whether the threat of punishment actually discourages people from committing crimes. A new study published in the Journal of Political Economy sheds some empirical light on the question of deterrence. Using a recently passed Italian law as a natural experiment, the study found that former prisoners are less likely to return to jail if they expect longer sentences for future crimes.
"This paper contributes to the literature providing evidence that potential criminals do respond to a change in prison sentences," write study authors Francesco Drago (University of Naples Parthenope), Roberto Galbiati (Centre National de la Recherche Scientifique, Paris) and Pietro Vertova (University of Bergamo).
Passed in 2006, Italy's Collective Clemency Bill presents a unique opportunity to study the deterrent effect of prison sentences, the authors say. Crime rates often drop when criminal penalties are increased. But it's often hard to tell if the rates go down because the threat of longer sentences deters potential criminals, or if the drop happens because actual criminals are physically removed from the street for longer periods. This study of the clemency law's effects eliminates the latter scenario, measuring only deterrent effect.
When the clemency bill was passed, it immediately released thousands of prisoners who had three years or less left on their sentences. The remainder of each prisoner's sentence was suspended, but not forgiven. The law stipulated that a former inmate who commits a new crime within five years will have the suspended portion of his sentence reinstated and added to the sentence for the new crime. As a result, a repeat offender can expect extra jail time equal to the suspended portion of his sentence—anywhere from one month to three years.
Using government data, the researchers looked at the recidivism rates of these former inmates for the first seven months after their release. They found that those with longer suspended sentences—and therefore longer expected sentences for new crimes—were less likely to be re-arrested than those with shorter suspended sentences.
"These results corroborate the general theory of deterrence," the authors write. According to their calculations, "increasing the expected sentence by 50 percent should reduce recidivism rates by about 35 percent in seven months."
But even a small increase in the expected sentence was enough to deter recidivism at least a little, the team found. The data suggest that a one month increase in expected sentence resulted in a 1.3 percent lower probability of returning to prison.
The deterrent effect was consistent across age groups, and among men and women, though 95 percent of the sample was male.
"This means that a policy a commuting actual sentences in expected sentences significantly reduces recidivism," Dr. Vertova says. "A mass release of prisoners can be effective in reducing their propensity of re-committing crimes if, when a released individual gets convicted of a new crime, his normal sentence is increased by the time that was pardoned because of the early release."
There was one important exception to the deterrent effect, however. Recidivism rates among those whose original crime was more serious were essentially unaffected by the length of their suspended sentence. That finding suggests that "more dangerous inmates are not deterred," the authors write.
The researchers also caution that their results only measure deterrence on those who have already served time in jail. "Indeed, it is not clear whether these results can be to individuals who have never received prison treatment."
Despite the limitations, however, the study does provide real-world evidence that "individuals vary their criminal activity in response to a change in prison sentences," the authors write.
Other social bookmarking and sharing tools:
- Drago et al. The Deterrent Effects of Prison: Evidence from a Natural Experiment. Journal of Political Economy, 2009; 117 (2): 257 DOI: 10.1086/599286
Note: If no author is given, the source is cited instead.