Paper Abstract
90. Emma Harrington & Hannah Shaffer, Discerning Discretion? Estimating Prosecutor Effects at Criminal Sentencing, 05/2020.
Abstract: In criminal courts, prosecutors have discretion over charging decisions and plea-deal offers, which often determine defendants' ultimate sentencing outcomes. These decisions require prosecutors to balance many, potentially competing, objectives. Sentencing laws task prosecutors with reducing criminal re-offense and treating similarly situated defendants similarly. Prosecutors may also weigh the costs of incarcerating the defendant against these legal mandates. Our project aims to evaluate the extent to which different prosecutors arrive at systematically different results and whether progress on one aim comes at the cost of other objectives. This paper focuses on the trade-off between reducing incarceration and criminal re-offense. This tradeoff may seem especially stark given that defendants cannot re-offend while in prison. However, variation across prosecutors in their degree of selective incarceration would attenuate the aggregate relationship between prosecutors' recidivism and incarceration effects.
Using court records from North Carolina Superior Court, we find that prosecutors systematically vary in both their incarceration and recidivism effects. A prosecutor one standard deviation above the mean in sentencing severity imposes prison sentences that are 15% longer than the mean prosecutor (12 more days relative to a mean of 83 days). A prosecutor one standard deviation above the mean in re-offense has a 6% higher rate of 3-year re-offense in their caseload (3.6pp off of a mean of 50%). As expected, prosecutors who impose longer prison sentences achieve lower rates of re-offense. However, we find that the bulk of the systematic variation in re-offense is driven by differences in the degree to which prosecutors selectively incarcerate those defendants most likely to re-offend. The variation in “skill” is similar in magnitude to the variation in unconditional recidivism: a prosecutor one standard deviation above the mean in skill can achieve a 3.1pp lower rate of re-offense.