with Vishal Kamat, Matthew Pecenco

Abstract

The criminal justice system affects millions of Americans through criminal convictions and incarceration. In this paper, we introduce a new method for credibly estimating the effects of both conviction and incarceration using randomly assigned judges as instruments for treatment. Misdemeanor convictions, especially for defendants with a shorter criminal record, cause an increase in the number of new offenses committed over the following five years. Incarceration on more serious felony charges, in contrast, reduces recidivism during the period of incapacitation, but has no effect after release. Our method allows the researcher to isolate specific treatment effects of interest as well as estimate the effect of broader policies; we find that courts could reduce crime by dismissing marginal charges against defendants accused of misdemeanors, with larger reductions among first-time defendants and those facing more serious charges.