Mandatory sentencing and racial disparity: assessing the role of prosecutors and the effects of Booker.

• Author: Starr, Sonja B.
• Position: Introduction through II. Estimating Racial Disparity in Sentencing: A Process-Wide Approach, p. 2-38

ARTICLE CONTENTS INTRODUCTION I. PROSECUTORS, SENTENCING, AND THE "HYDRAULIC DISCRETION" THEORY II. ESTIMATING RACIAL DISPARITY IN SENTENCING: A PROCESS-WIDE APPROACH A. Studies Estimating the Extent of Unwarranted Sentencing Disparities B. Our Dataset C. Our Research on Racial Disparities in Charging and Sentencing: Some Key Findings D. Interpretations and Limitations 1. Possible Unobserved Offense Differences 2. Possible Differences in Offender Characteristics 3. Possible Sources of Disparity that Our Estimates Leave Out 4. Race, Gender, and Their Interaction III. THE BOOKER QUESTION: DOES EXPANDING JUDICIAL DISCRETION INCREASE RACIAL DISPARITY? A. The Changing Yardstick Problem B. The Causal Inference Problem C. Our Method D. Regression Discontinuity Estimates of Booker's Effects 1. Changes to Charging 2. Changes in Plea-Bargaining 3. Changes in Sentencing Fact-Finding and Sentencing Outcomes E. Limitations and Causal Inference Challenges 1. Limitations of the RD Method 2. Blakely and Anticipation of Booker CONCLUSION INTRODUCTION

In the United States, one of every nine black men between the ages of twenty and thirty-four is behind bars, (1) and, in 2003, the Bureau of Justice Statistics projected that one in every three young black men could expect to be incarcerated at some point in his life. (2) These rates far exceed those of any other demographic group--for instance, black males are incarcerated at nearly seven times the rate of white males. (3) The impact of demographically concentrated incarceration rates on offenders, families, and communities is a critical social concern. (4) But why do these gaps exist? Can they be explained by differences in criminal behavior, or by differences in how the criminal justice system treats offenders? If it is the latter, can the process be improved by reforms, such as changes to sentencing law?

These questions are not new. For decades, racial and other "legally unwarranted" disparities in sentencing have been the subject of considerable empirical research, which has in turn helped to shape major policy changes. Most importantly, the U.S. Sentencing Guidelines and their state counterparts were adopted with the goal of reducing such disparities. In 2005, when the Supreme Court's decision in United States v. Booker rendered the formerly mandatory Guidelines merely advisory, Justice Stevens's dissent predicted that "[t]he result is certain to be a return to the same type of sentencing disparities Congress sought to eliminate in 1984." (5) Whether this prediction was accurate is perhaps the foremost empirical question in sentencing policy today. The most prominent study to date, a 20l0 report of the U.S. Sentencing Commission, gave an alarming answer: Booker and its judicial progeny had quadrupled the black-white sentencing gap among otherwise-similar cases, from 5.5% to 23.3%. (6) In January 2013, the Commission issued an update with similar figures (revising the latter figure slightly downward, to 19.5%), this time combined with explicit calls for legislation in effect returning the Guidelines to something fairly dose to their prior binding status. (7)

This Article introduces a new empirical approach and gives a very different answer. The Commission's methods are hobbled by two serious limitations that also pervade the broader empirical literature on sentencing disparity. (8) First, these studies consider the judge's final sentencing derision in isolation, ignoring crucial earlier stages of the justice process. Those earlier stages have important sentencing consequences, and yet these studies exclude the portions of the ultimate sentence gap that result from earlier-stage decision, making from their estimates. Second, studies of changes in disparity after legal changes (like Booker) have failed to disentangle the effects of the legal change from surrounding events and background trends.

This Article develops these two critiques and discusses our own research on racial disparities among federal arrestees, which uses a method that avoids these problems. We first highlight some findings from our recent study showing that while a black-white gap appears to be introduced during the criminal justice process, it appears to stem largely from prosecutors' charging choices, especially decisions to charge defendants with "mandatory minimum" offenses. These findings highlight the importance of taking into account the early parts of the justice process. With that in mind, we then present our new findings on Booker, estimating its effects not only on sentencing, but also on charging, plea-bargaining, and sentencing fact-finding, an analysis no prior studies have performed. Far from finding evidence that judges" use of expanded discretion worsens disparity, we fail to find an increase in disparity and find suggestive evidence cutting in the opposite direction. (9)

Our research seeks to close a surprisingly wide gap that separates two bodies of scholarship: the theoretical and qualitative literature on how the criminal justice system functions (which uniformly recognizes the critical role of prosecutors) and empirical research on sentencing disparities (which effectively ignores that role). The modern criminal justice process is prosecutor-dominated. Prosecutors have broad charging and plea-bargaining discretion, and their choices have a huge impact on sentences. A central claim made by critics of mandatory sentencing is that restricting judicial discretion further empowers prosecutors, who tend to exercise that power in ways that perpetuate or worsen disparity. This "hydraulic discretion" theory has been described as a near-consensus view of sentencing scholars. (10)

Yet the empirical research on sentencing disparity has not tested these claims and fails to account for the role of prosecutorial discretion. Researchers typically estimate sentencing disparities in federal and other courts subject to sentencing guidelines after controlling for (among other things) the recommended guidelines sentence. But the guidelines recommendation is itself the end product of charging, plea-bargaining, and sentencing fact-finding. Controlling for it filters disparities in those processes out of the sentencing-disparity estimates and gives an incomplete view of the scope and sources of sentencing disparity. (11) In effect, the existing literature focuses on disparities in compliance with the sentencing guidelines. While this is an important piece of the sentence-disparity picture, it is far from the only piece, because decisions made throughout the process ultimately affect the sentence. Moreover, sentencing-stage disparities might either offset or exacerbate disparities arising earlier, making it hard to interpret them in isolation.

We accordingly take a broader, process-wide approach, constructing a dataset that links records from four different federal agencies and allows us to trace criminal cases from arrest through sentencing. We focus on the gap between black men and white men in non-immigration cases. Instead of controlling for the Guidelines sentence, we control for the arrest offense and other characteristics that are fixed at the beginning of the justice process. The arrest offense is an imperfect proxy for underlying criminal behavior, but we believe it is the best proxy available for this purpose. Our method allows us to assess aggregate disparities introduced throughout the post-arrest justice process, from charging through sentencing. Further, it also allows us to analyze the contribution of each procedural stage (as well as underlying case differences) to the total black-white gap.

The problem with the prevailing method is not merely an academic concern. In Part II of this Article, we highlight and discuss key findings of our analyses of charging and sentencing in federal criminal cases from 2007 to 2009. (12) That research shows that after controlling for the arrest offense, criminal history, and other prior characteristics, there remains a black-white sentence-length gap of about 10%. But judges' choices do not appear to be principally responsible. Instead, between half and the entire gap can be explained by the prosecutor's initial charging decision-specifically, the decision to bring a charge carrying a "mandatory minimum." After controlling for pre-charge case characteristics, prosecutors in our sample were nearly twice as likely to bring such a charge against black defendants. (13) In other words, studies that focus only on the judicial sentencing decision exclude what appears to be the most important procedural source of disparity in sentences.

A proper analysis of Booker's effects on disparity, then, should take the whole justice process into account, to the extent possible. In Part III, we present the results of such an analysis. We begin that inquiry with a simple linear time-trend analysis, which shows that, when one measures sentence disparity in the broader way that we recommend, unexplained black-white disparity did not grow between 2003 and 2009, the period in which the Sentencing Commission found that it quadrupled. Indeed, our estimate of the disparity trend is negative, although imprecise. That is, the gap in sentences for similar black and white arrestees was, if anything, slightly smaller by the end of 2009 than it was just before Booker. The Commission's claim that disparity grew over that same period is an artifact of its flawed way of measuring disparity.

Beyond the question of whether disparity has changed during the period surrounding Booker, we must further ask whether it has changed because of Booker. The two questions are not the same, but they are too often confused. In addition to the disparity-measurement question, a second serious flaw pervades the empirical literature on sentencing-law changes: the failure to provide a sound basis for causal inferences. This second problem is exemplified by the Sentencing Commission's analysis. The...