Excerpted From: Hannah Shaffer, Prosecutors, Race, and the Criminal Pipeline, 90 University of Chicago Law Review 1889 (November, 2023) (253 Footnotes) (Full Document)


HannahShafferIn July of 2017, Cody Helms and Darius Miles were arrested on separate breaking and entering charges in Alamance County, North Carolina. The two had indistinguishable criminal records: a breaking and entering conviction and two priors for felony possession of cocaine. Under the state sentencing guidelines, Cody and Darius were facing eight to ten months in state prison.

In their case files, Cody and Darius looked identical except for their race: Cody was white, and Darius was Black. However, the apparent similarity in their arrests and criminal records may have masked earlier discrimination. Historically, police, prosecutors, and judges may not have treated Black and white people equally. A larger police presence in majority-Black neighborhoods may have led to more arrests of Black people. And the post-arrest process may have disproportionately convicted Black defendants--or simply passed through earlier bias in arrests.

If the prosecutor assigned to both cases believed that Black defendants' criminal records were inflated by past bias, she may have felt that an equal punishment for Cody and Darius was unfair. Ultimately, the court records indicate that the prosecutor reduced Darius's charge to a misdemeanor but retained Cody's felony charge. Cody was sentenced to eight months in state prison, while Darius was sentenced to six months of probation.

Prosecutors can exercise their discretion to compound, pass through, or reduce racial disparities in their cases. This Article introduces a new approach to understanding this prosecutorial power. Rather than attempting to estimate prosecutors' own biases, I examine how prosecutors' beliefs about past biases in the system impact their current decisions. Specifically, I analyze how prosecutors interpret and respond to racial disparities in defendants' criminal records, which reflect any actual racial differences in criminal conduct as well as any accumulated biases from past cases. To do this, I use 336,000 court records from North Carolina Superior Court and an original survey of 203 North Carolina prosecutors.

I find that prosecutors have increasingly used their discretion in recent years to reduce racial disparities by penalizing the prior convictions of white defendants more than Black defendants. When facially similar white and Black defendants have no criminal record, they were equally likely to be incarcerated. However, for defendants with records, white defendants were significantly more likely to be incarcerated than facially similar Black defendants. This empirical finding runs counter to an extensive literature that has either found or assumed that prosecutors increase racial disparities. This finding also poses challenges to scholars and advocates who argue that legislatures should limit prosecutorial discretion--or colorblind prosecutors--in order to reduce disparities. Such proposals may inadvertently increase disparities by neutralizing the offsetting effects of some prosecutors, thereby cementing race gaps generated earlier in the criminal pipeline.

To understand the post-arrest system's aggregate penalty of white versus Black defendants' criminal records, I estimate how the increase in the likelihood of a prison sentence for additional prior convictions differs for white defendants versus Black defendants. I first trace the raw relationship between criminal history and prison rates for white and Black defendants. To better isolate whether Black defendants' prior records are being discounted, I then compare white and Black defendants who appear to be similar in their case files.

I find that the influence of a defendant's rap sheet on his current punishment differs significantly by defendant race. Over the past decade, the sentencing penalty associated with an additional prior conviction has been 25%, or 1.8 percentage points (pp), larger for white than facially similar Black defendants. For each additional prior conviction, Black defendants were 7.2pp more likely to receive a prison sentence, while white defendants with similar arresting charges and criminal records were 9.0pp more likely to receive a prison sentence. Returning to the cases of Cody and Darius, this indicates that if both had an additional prior conviction, Cody's likelihood of incarceration would have increased by 25% more than Darius's. Since Black Americans are four times as likely as white Americans to have felony records, the smaller penalty for Black defendants' prior convictions means that the system offset disparities from the past.

This offset of prior conviction disparities became increasingly pronounced over the course of the decade. If this trend were driven by increasing attention to police bias, one would expect the trend to be concentrated in the disparate treatment of prior convictions initiated by police stops. To test this theory, I compare the trend in the penalty of prior convictions for drug and weapon possession offenses--which more likely are initiated by a police officer's decision to stop and arrest a civilian--to the same trend for prior convictions for violent, sex, and property offenses--which more likely are initiated by a victim's or witness's report to police. I find that the trend is concentrated in shifting penalties for prior convictions likely initiated by the police, suggesting that the change in North Carolina may have stemmed from a shift in perceptions of police bias. Strikingly, there was almost no change over time in the racially disparate penalty of prior convictions likely initiated by victims or witnesses. Instead, the post-arrest system in North Carolina consistently penalized prior convictions for property, violent, and sex offenses less for Black defendants than white defendants.

The post-arrest system's offset of prior conviction disparities could reflect prosecutorial discretion as well as pressures from defense attorneys, judges, or the electorate. To better understand the role of the prosecutor, I link an online, written survey of 203 North Carolina assistant district attorneys to each participants' real-world cases. I developed and fielded this survey with my collaborators from May to November 2020, following an in-person pilot in two additional offices in November 2019.

The survey asked prosecutors about their views on the source of racial disparities in criminal outcomes. Specifically, it asked how much sentencing disparities were driven by disparate conduct (that Black defendants have more severe criminal conduct) versus racial bias (that Black defendants' conduct is perceived to be more serious than the same conduct committed by white defendants). Linking each prosecutor's survey responses to her cases allowed me to test whether a prosecutor's perception of bias predicts the sentencing penalty for Black versus white defendants' prior convictions in that prosecutor's cases. I also leverage the fact that cases are quasi randomly assigned to prosecutors within office crime units in North Carolina in order to estimate the relationship between prosecutors' beliefs and their disparate impacts in their cases.

These linkages reveal that the prosecutors who attribute disparities more to racial bias than to differences in criminal conduct drive the entirety of the post-arrest system's smaller penalty for Black defendants' priors. I hypothesize that these prosecutors likely (consciously or subconsciously) regard the priors of Black defendants as less reliable signals of underlying criminal conduct--or as providing less evidentiary weight about dangerousness or moral culpability. By contrast, prosecutors who attribute racial disparities more to disparate conduct than bias have equal prison rates for Black and white defendants with similar arresting charges and criminal records. These prosecutors likely interpret prior records as unbiased signals that were produced by a colorblind system. While the second group of prosecutors reproduces the disparities from the past, the first group offsets the accumulated disparities embedded in criminal records.

These findings run counter to the widespread view that prosecutorial discretion compounds racial disparities at sentencing. Scholars have advanced a range of theories about prosecutorial dominance and unchecked bias. Some argue that the proliferation of overlapping criminal sanctions empowers prosecutors to charge and punish civilians more extensively and selectively. Others argue that prosecutors' combined powers over investigation and adjudication invite bias and abuse of power. All agree that sentencing guidelines amplify prosecutorial power, and that courts and legislatures fail to safeguard defendants from discriminatory charging.

In the wake of these critiques, calls to limit prosecutorial discretion have been resounding. Some reforms explicitly propose blinding prosecutors to race in order to root out bias. Indeed, several district attorneys have recently implemented “race-blind charging,” which strips incident reports of defendant race and any information that could signal race. Colorblinding prosecutors is also a live debate among state legislatures: In September 2022, the California legislature passed a “Race-Blind Charging” bill that mandates race-blind charging in all district attorney offices in the state by 2025. Other proposals to limit prosecutors are implicitly grounded in the same paradigm of colorblindness and formal equality. Prosecutor guidelines are a case in point. For decades, scholars have argued for prosecutor guidelines to ensure more uniformity and less bias in charging decisions.

This Article's findings suggest that reformers should exercise caution when considering the calls to colorblind prosecutors. In some cases, colorblinding policies may successfully reduce racial disparities by preventing prosecutors from introducing new bias. For instance, in other work that I coauthored with Professor Emma Harrington, we find that prosecutorial discretion increased disparities in the 1990s and early 2000s in North Carolina. Thus, colorblinding prosecutors may have helped to reduce disparities in past decades. However, in other cases, especially in recent years, colorblinding prosecutors may have no effect or, paradoxically, increase race gaps by cementing the impacts of accumulated disparities in arrests, charges, and sentences. Colorblinding could exacerbate race gaps by preventing current prosecutors from counteracting disparities inherited from police and past decision-makers.

To demonstrate the potential downside of colorblinding, this Article evaluates the impacts of a hypothetical law requiring consistent sentencing of defendants with similar charges and criminal records. This hypothetical law would have significantly increased disparities in prison rates relative to actual outcomes in North Carolina, particularly for Black people with prior records. Suppose that the law resulted in Black defendants being incarcerated at the same rate as current, facially similar white defendants. Had this occurred, the system would have incarcerated 696 more Black people with prior felonies in 2019, a 26% increase relative to the actual number that year.

This Article's findings cannot, by themselves, tell us whether prosecutors moved prison outcomes closer to racially equal punishment of similar criminal conduct. Similarly, the findings cannot identify the degree of prosecutorial bias. The results indicate that certain prosecutors reduced disparities relative to disparities inherited from police arrests and past cases. Therefore, the extent of prosecutors' own racial biases--and the normative implications of their decisions--depend on the extent of past bias in the system and racial differences in underlying conduct, neither of which are observable. The smaller penalty for Black defendants' priors could be: (1) a welcome corrective for past bias; (2) an insufficient corrective given the extent of past racism; or (3) an unfair overcorrection to a system that treated Black and white people equally. Although the results cannot directly adjudicate between these interpretations, they do show that the system reduced racial disparities in total levels of incarceration relative to a colorblind system. Despite the system's offset of racial disparities, however, Black North Carolinians are still disproportionately likely to be incarcerated: in 2019, for instance, they were still 3.8 times as likely as white North Carolinians to be sentenced to prison.

To provide additional evidence about whether Black defendants' criminal records are, in fact, inflated by bias, I assess whether prior convictions are stronger predictors of rearrest for white defendants relative to Black defendants. I find that white defendants' priors are indeed stronger predictors of rearrest. This suggests that Black defendants' records are weaker signals of criminal conduct, which, in turn, suggests that prosecutors who discount Black defendants' records are moving the system closer to racially equal punishment of underlying conduct. While future arrest is itself a biased measure of criminal conduct, these results nevertheless provide suggestive evidence of equity and efficiency gains.

This Article makes three contributions to the literature on prosecutorial discretion and racial disparities. First, as a methodological contribution, it combines qualitative survey data with the administrative records of each surveyed prosecutor's past cases. This survey-to-court-record linkage is the first of its kind and enables me to analyze how prosecutors' stated beliefs predict the outcomes in their cases. Notwithstanding the calls to limit prosecutorial discretion, little is known about how prosecutors exercise their discretion in individual cases or how prosecutors think about past disparities in their current decisions. While past work has estimated prosecutors' impacts on sentencing disparities, this Article considers how rank-and-file line prosecutors factor racial bias into their decision-making process.

Second, this Article makes a conceptual contribution. By tracing the racial disparities in prosecutors' responses to prior records, it emphasizes and interprets their impacts in a sequential system, in which a prosecutor's response to past disparities may be as important as her own “internal” biases. Much of the literature has focused on isolating disparate treatment at a particular juncture of the criminal pipeline, which has had the effect of “controlling away” disparities from the past. Viewing discretion in isolation can obscure the fact that treating people similarly who appear similar in their case files may, in fact, run counter to genuine equality under the law. While drilling down on discretion at a single point in the process can offer important insights, this Article takes a step back to consider the impacts of discretion given the significant disparities that already exist when prosecutors first open their case files.

Third, this Article makes a contribution to our understanding of racism in the criminal process. There is widespread agreement that racism has infected the system and that the discretion of individual decision-makers tends to compound disparities. These findings add nuance to the study of endemic racism in the criminal process by showing that some prosecutors use their discretion to offset past disparities.

The Article proceeds in four Parts. Part I lays the foundation for the Article's analysis. Part II presents the core results from the post-arrest system. Part III presents the results from the prosecutor survey. Part IV turns to the policy and normative implications of the findings and briefly discusses the implications for the progressive prosecutor movement.

The Appendix has five Sections. Section A details the construction of prosecutor case identifiers. Section B presents the estimation strategy and the main analyses. Section C tests cases that are randomly assigned to prosecutors within crime units, which is essential to the survey analysis. Section D includes the interfaces for survey questions used in the analysis. It also explains how I used the survey to classify prosecutor beliefs and shows robustness to alternative classifications. Finally, Section E includes all tables and figures not included in the main text. Tables and figures in the main text are numbered, while those in appendices are lettered and numbered (e.g., Table E.3 refers to Figure 3 in Appendix E).

[. . .]

For criminal justice advocates who aim to reduce racial disparities in sentencing outcomes, this Article provides evidence that it may be more effective to harness prosecutorial discretion than to regulate or eliminate it. Especially for those concerned about the disparate impacts of sweeping police powers to stop and arrest civilians, these findings counsel caution in colorblinding prosecutors, which may lock in the impacts of police discretion. As critical race theorists have long emphasized, colorblindness cannot eliminate the deeper structural problems of racism--and eliminating racism today cannot erase disparities from the past that will continue to impact defendants in the future.

The Article's findings also sound a note of optimism for the future. There is new blood in prosecution, perhaps in part due to the rise of the Black Lives Matter movement and a shift in the way we talk about prosecutors in law school classes. These younger prosecutors may be a positive force in reducing racial disparities in future years.

Assistant Professor, Harvard Law School.