Prison, Probation, and the Community

This paper examines how incarceration and probation shape community crime rates with 20 years of administrative court records from Milwaukee, Wisconsin. I estimate causal effects by instrumenting for a criminal defendant's sentence with the preferences of a randomly assigned judge. I extend the canonical randomized judge design framework to incorporate effects generated along the intensive margin of sentence durations. This framework allows me to distinguish between effects to the marginally incarcerated defendant from effects to non-marginal defendants. Empirically, I find that sentencing effects are driven by extensions in the duration of incarceration, with one additional year preventing an estimated 17% of defendant reconvictions and an additional 4% of neighbor convictions during the subsequent five years. Heterogeneity analyses reveal that spillover effects are driven by violent felony offenders, indicating that incapacitating high-risk offenders generates additional social benefits in the neighborhood.