I study the effects of prison quality on recidivism using individual level data fromColombia. To estimate causal effects, I leverage the quasi-random assignment of in-mates to newer, less crowded and higher service prisons. For inmates assigned tonewer facilities, I find that the probability of returning to prison within one year is36% lower. Criminal capital, access to rehabilitation programs, and negative prisonexperiences—which could trigger changes in intrinsic preferences over illegal occupa-tions, seem to be important mechanisms. The program led to substantial welfare gains,even when assuming a low social cost per crime.

ACCESS PAPER