Public appeals regarding criminal justice have shifted somewhat from "tough on crime" to "smart justice" that is more lenient when tradeoffs merit it. Among other considerations, smart sentencing policy depends on how sentence severity affects recidivism. Using administrative data on two common non-violent felonies committed by adults in Michigan, we measure the effect of sentences on offenders' future criminal activity. Discontinuities in the legislative guidelines that constrain sentences chosen by Michigan judges provide exogenous variation in sentence severity. Harsher sentences generated by sentencing guidelines significantly reduce recidivism by felony shoplifters but not repeat drunk drivers. Recidivism falls most for young, male offenders from Southeast Michigan and varies non-monotonically with prior criminal record. Because of such heterogeneity, any empirical strategy measures a local average treatment effect relevant to a particular population of offenders. Contrary to our main results, we find no evidence that harsher sentences induced by judge assignments reduce recidivism in our sample. When sentencing guidelines provide the primary practical policy lever, "smart justice" should incorporate directly relevant empirical evidence that accounts for offense-and offender-specific tradeoffs between public safety and the public budget.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.