INTRODUCTION I. A BRIEF HISTORY OF FEDERAL SENTENCING REFORM A. Inter-Judge Sentencing Disparity Before Booker 1. The Sentencing Reform Act of 1984 2. Mandatory Sentencing Guidelines (1987-2004) 3. PROTECT Act (2003) B. The Booker Revolution, 2005-2007 1. Booker, Kimbrough, and Gall 2. Average sentence length and guideline sentencing 3. Inter-judge sentencing disparity II. THE EMPIRICAL STUDY OF INTER-JUDGE SENTENCING DISPARITY A. Data and Methods 1. Judge-specific data 2. Natural experiment method 3. Measures of inter-judge disparity 4. Why Massachusetts? B. Results 1. Sentence length 2. Guideline sentencing patterns III. IMPLICATIONS A. Conventional Explanations for Within-Range Sentencing 1. Inertia 2. Risk aversion 3. Anchoring 4. Strategic behavior 5. Laziness B. Alternative Explanations for Within-Range Sentencing 1. Agreement with the Guidelines' recommendations 2. Institutional considerations CONCLUSION APPENDIX A. Methodological Details 1. Period selection 2. Case matching 3. Random distribution 4. Discretionary sentences B. Detailed Results 1. Regression models 2. Alternative time periods INTRODUCTION
A central purpose of the Sentencing Reform Act of 1984 was to reduce inter-judge sentencing disparity. Congress was concerned that similarly situated defendants were receiving widely divergent sentences based on the philosophy, politics, and biases of the sentencing judge. The Federal Sentencing Guidelines, promulgated by the United States Sentencing Commission, were designed to minimize that form of unwarranted disparity by designating a mandatory sentencing range, applicable to all judges, based on the circumstances of the offense and characteristics of the offender.
But in a series of decisions from 2005 to 2007, the Supreme Court upended the federal sentencing regime. In United States v. Booker, (1) the Court resolved a constitutional defect in the design of the Guidelines by rendering them "effectively advisory," leaving judges free to impose any reasonable sentence consistent with the broad purposes of punishment outlined by Congress. (2) Three years later, in Gall v. United States, (3) the Court directed appellate courts to review sentencing decisions under a "deferential abuse-of-discretion standard." (4) And on the same day, in Kimbrough v. United States, (5) the Court indicated that district courts are now free to sentence outside the guideline range "based solely on policy considerations, including disagreements with the Guidelines." (6)
In the wake of those decisions, the Commission has received extensive anecdotal reports of a surge in inter-judge sentencing disparity. The Department of Justice reported in a June 2010 memorandum that "[m]ore and more, we are receiving reports from our prosecutors that in many federal courts, a defendant's sentence will largely be determined by the judicial assignment of the case; i.e., which judge in the courthouse will conduct the sentencing." (7) Attorney General Eric Holder, in a June 2009 speech on sentencing policy, issued a call for research into whether post-Booker sentencing practices "show an increase in unwarranted sentencing disparities" based on "differences in judicial philosophy among judges working in the same courthouse. " (8) Prosecutors around the country echoed those concerns at the Commission's 2009-2010 regional hearings. (9) Patrick Fitzgerald, the U.S. Attorney for the Northern District of Illinois, warned that Booker has "re-introduced into federal sentencing both substantial district-to-district variations and substantial judge-to-judge variations." (10) Prosecutors have reported a similar spike in inter-judge disparity in "nearly all districts" in the Ninth Circuit. (11) Frank Bowman calls the Supreme Court's decisions a "debacle," (12) and warns that in white-collar cases, "we're back to a pre-guidelines era" marked by "disparity and the most potential for disparity." (13)
Those reports, if accurate, deserve urgent attention because they implicate Congress's core objective in reforming federal sentencing. Inter-judge sentencing disparity, in the view of sentencing reformers, offends important rule-of-law principles, erodes respect for the courts, and undermines the deterrent effect of the criminal law. Congress, if it wishes, has several options available to address the problem by altering the Sentencing Guidelines to resolve the constitutional defects identified by the Supreme Court.
To date, however, the evidence of an uptick in inter-judge disparity has been strictly anecdotal. This Article addresses a critical gap in the research, offering the first empirical account of inter-judge sentencing disparity since the Supreme Court's shake-up of federal sentencing. It does so by drawing on an original new dataset of sentences from the District of Massachusetts, the only district that makes key sentencing documents available to the public. The records allow, for the first time, a study of how individual judges have responded to the federal sentencing revolution.
Analysis of those sentences reveals a clear increase in inter-judge disparity, both in sentence length and in guideline sentencing patterns. Following the Supreme Court's decisions in Booker, Kimbrough, and Gall, the effect of the judge on sentence length has doubled in strength. (14) In cases not governed by a mandatory minimum, the court's three most lenient judges have imposed average sentences of 25.5 months or less, while its two most severe judges have imposed average sentences of 51.4 months or more. That stark difference translates to an average of more than two years in prison, depending on which of those judges is assigned the case. (15)
Similarly, the Boston data reveal that some judges have taken advantage of their enhanced discretion to depart from the Guidelines to a far greater extent than others. Two judges (call them "business as usual" judges) continue to impose below-guideline sentences at essentially the same rate as before Booker, as little as 16% of the time. But four other judges (call them "free at last" judges) now sentence below the guideline range at triple or quadruple their pre-Booker rates, as much as 53% of the time. (16) In addition, the effect of the judge on how far sentences fall from the guideline range has more than doubled in the wake of Booker, Kimbrough, and Gall. (17)
These results tend to corroborate the anecdotal reports of an increase in inter-judge sentencing disparity. Yet they are necessarily tentative. As with any study of a single district court, there is a risk that the results are not representative of sentencing trends nationwide. And because inter-judge disparity is but one factor to consider in evaluating a sentencing system, the results do not compel any judgment about whether the Supreme Court's decisions, on balance, have improved or worsened federal sentencing. Nonetheless, the Boston data offer an unprecedented look at how individual judges have responded to the Supreme Court's decisions.
The Article proceeds in three parts. Part I explains the importance of inter-judge sentencing disparity to Congress's reform efforts and describes the trio of Supreme Court decisions that reshaped federal sentencing between 2005 and 2007. Despite anecdotal reports of a surge in inter-judge disparity, neither the Commission nor other researchers have examined the effects of Booker, Kimbrough, and Gall on the sentencing patterns of individual judges.
Part II of the Article reports the empirical study. Part II.A describes the Article's unique dataset of sentences linked to individual judges. It also summarizes the Article's methods, which build on "natural experiment" studies of inter-judge disparity after the promulgation of the Guidelines. Part II.B reports the results of the study. Details of the data and methods, as well as full reports of the regression models, appear in the Appendix.
Part III considers possible explanations for the Article's key finding of a spike in inter-judge sentencing disparity. It casts doubt on the conventional theories that persistent within-guideline sentencing is the product of inertia, fear of reversal, "anchoring" effects, strategic behavior, or simple laziness. Instead, it proposes two alternative explanations: some judges might actually agree with the Guidelines' recommendations, or may elect to impose within-range sentences for institutional reasons.
A BRIEF HISTORY OF FEDERAL SENTENCING REFORM
Before describing the nuts and bolts of the empirical study, a brief history of federal sentencing reform is needed, both to demonstrate the importance of inter-judge disparity to sentencing reform, and to describe the Supreme Court decisions that radically altered federal sentencing law from 2005 to 2007.
Inter-Judge Sentencing Disparity Before Booker
The Sentencing Reform Act of 1984
Until the early 1980s, criminal sentencing in the federal system was "indeterminate." Federal judges enjoyed almost entirely unfettered discretion in choosing the type and severity of sentence. (18) Criminal statutes generally designated high maximum penalties and no minimum penalties, leaving judges free to impose a term of probation or imprisonment of any length within a broad range. (19) Judges were under no obligation to give reasons for the sentence imposed, (20) and appellate review of sentencing decisions was virtually nonexistent. (21) The theory was that judges should "individualize" sentences to serve the rehabilitative needs of criminal defendants, "almost like a doctor or social worker exercising clinical judgment." (22)
In practice, however, indeterminate sentencing gave judges so much discretion that criminal defendants faced starkly different levels of punishment depending on which judge happened to draw the case. For prominent scholars, the evidence of "inter-judge sentencing disparity"--differences in sentencing outcomes caused by the judge, rather than by legitimate differences between...