In the past decade, several U.S. prosecutors, elected on reform-oriented platforms, have implemented policies categorically declining to prosecute certain low-level offenses. These progressive prosecutors argue that their policies alleviate some of the disproportionate and racially biased consequences of broken windows policing. This article uses one case study to examine whether these policies affect the rate at which police make arrests for low-level crimes that prosecutors will decline to prosecute. With arrest and citation data from New York City, this article uses a differences-in-differences design to examine whether marijuana enforcement changed in Brooklyn as a result of District Attorney Ken Thompson’s 2014 decision not to prosecute low-level marijuana possession. This analysis finds a null effect of that policy upon the rates of arrest and citation for low-level and more serious marijuana possession crimes. Arrests, even without prosecution, serve important functions for police and bring weighty consequences for arrestees; this null effect raises questions about how closely linked the purposes of arrests and prosecutions are in the view of patrol officers. Moreover, these findings might prompt questions about the ability of prosecutors to enact effective reforms on their own, given the highly siloed nature of discretion in the U.S. criminal legal system.

Substantial race-based disparities exist in federal criminal sentencing. We analyze 380,000 recent (2006-2019) sentences in the JUSTFAIR database and show that these disparities are large and vary considerably across judges. Judges assign white defendants sentences 13% shorter than Black defendants’ and 19% shorter than Hispanic defendants’, on average, conditional on case characteristics and district. Judges one standard deviation above average in their estimated Black-white disparity give Black defendants sentences 39% conditionally longer than white defendants’, vis-à-vis an average disparity of 13%. Judges one standard deviation above average in their estimated Hispanic-white disparity give Hispanic defendants sentences 49% conditionally longer than white defendants’, compared to the average disparity of 19%.