Contemporary research has examined the relationship between determinate sentencing reforms and unwarranted punishment disparities in states and the federal criminal justice system. Recent investigations suggest that legal developments in federal sentencing—namely, the High Court’s rulings in U.S. v. Booker (2005) and Gall/Kimbrough v. U.S. (2007) which rendered and subsequently reaffirmed the federal guidelines as advisory—have not altered disparities associated with imprisonment outcomes. Punishment disparities following Booker and Gall, particularly racial and ethnic disparities, have been linked to Assistant U.S. Attorneys’ (AUSAs) use of substantial assistance departures. What remains unanswered in the literature is whether the changes in AUSAs’ decision making following the landmark cases has enduring effects and whether the effects are conditioned by defendants’ race/ethnicity and the type of case (guidelines cases or mandatory minimum cases), and whether the use of substantial assistance varies across U.S. District Courts.

Accordingly, these questions are examined using sentencing data from the U.S. Sentencing Commission, coupled with data from the National Judicial Center, U.S. Census Bureau, Uniform Crime Reports, and Interuniversity Consortium for Political and Social Research. This study looks at 465,476 defendants convicted from fiscal year 2001 to fiscal year 2010 across 89 federal districts. A series of multilevel discontinuity regression models are estimated to assess the short-term and long-term effects of the Booker and Gall/Kimbrough decisions on AUSAs’ use of substantial assistance departures, accounting for contextual differences between federal district courts.

The results show that AUSAs are less likely to seek motions for substantial assistance immediately and in the long term in the post-Booker period but are more likely to seek substantial assistance in the long term in the post-Gall/Kimbrough period. These effects, however, are restricted to the models that include all cases and guidelines cases. The interaction models show that Hispanic defendants facing a mandatory minimum sentence are less likely to receive a substantial assistance departure immediately and in the long term following the Court’s Booker decision. Moreover, the use of substantial assistance varies across federal districts. The results are discussed in relation to their implications for theory, courts and sentencing policy, and future research on punishment outcomes.