The Impact of Prison Labor Programs on Recidivism: The Case of Chile

We estimate the effect of prison labor programs on recidivism using a nationwide census-based dataset of all prison inmates released in 2010 in Chile and tracked for two years after release. Because participation in prison labor programs is not random, we use an instrumental variables (IV) regression procedure to address endogeneity and to estimate whether there is a reduced probability of recidivism that can be attributed to participation in prison labor programs. The results indicate that once the endogeneity problem is addressed, participation in prison labor programs does not contribute to a statistical reduction in the odds of recidivism for the overall sample; however, the estimation of heterogeneous effects reported statistically significant effects for specific groups. Length:31 pages