This paper employs a hazard model to analyse the impact of education and two types of prison employment programmes on recidivism over a ten-year period for 4515 prisoners released from Ohio prisons in 1992. Estimations with a Weibull mixture model and propensity score approach provide two means for investigating self-selection bias. Selection bias is detected for participation in the most common prison job programme but has little effect on estimated marginal savings impacts of prison industry and education programmes. Estimates of the cost savings from postponing return to prison due to programme participation are provided. The potential for cost savings through decreasing or delaying return to prison is an important finding given the substantial and increasing cost of incarceration. © The London School of Economics and Political Science 2008.
Author Keywords
Index Keywords
