Assessing racial disparities in parole release

In a rational choice model of parole release, a color-blind parole board seeking to minimize violations would release all prisoners below a certain risk threshold. To test this prediction, we extend the outcome-test methodology used in assessing discrimination in police searches. We overcome the inframarginality critique by taking advantage of strategic timing of release: within each racial group, violation rates are equalized for a given sentence length. We use the National Corrections Reporting Program data, which record all parole-release decisions in the United States. We find that violation rates are consistently higher for African American parolees, a result not consistent with a parole board bias against African Americans. This conclusion is robust to a variety of tests, including ruling out postrelease discrimination. Evidence on the timing of release suggests a policy aimed at limiting racial disparities in time served rather than in violation rates, which favors fairness over efficiency.