Recalibrating the Federal Economic Crime Guideline: An Admiring Rejoinder to Judge Bennett and Friends

I.     Introduction 

I have read with the greatest pleasure the article on federal white-collar crime sentencing by U.S. District Judge Mark Bennett and Professors Justin Levinson and Koichi Hioki.1 They review the history of fraud sentencing in the Sentencing Guidelines era, offer a persuasive critique of some deficiencies in the current regime, present the results of their own survey of judicial attitudes toward sentencing a representative fraud case, and propose a series of useful prescriptions for change. Inasmuch as they are kind enough to cite my own work approvingly throughout the article, it will come as no surprise that I agree with most of their diagnoses and virtually all of their prescriptions. That said, there are a few points on which we are not entirely in accord, and I think that the data from their very interesting survey may point in slightly different directions than they suggest. Hence, I offer this admiring rejoinder.

II.     Points of Agreement

Let me begin by noting the points on which I agree with Judge Bennett and his colleagues (who, for the sake of brevity and with no intent to stint on my admiration of Professors Levinson and Hioki, I will customarily refer to simply as “Judge Bennett”).

Judge Bennett is correct that the Federal Sentencing Guidelines (“Guidelines”) promulgated in 1987 consciously altered the previous paradigm for white-collar sentencing by prescribing short, but definite, terms of incarceration for economic criminals who would, under the prior regime of total judicial discretion, almost invariably have received probation.2 He is also correct that, over the next decade and a half, the combination of a scattershot accretion of annual amendments, the 2001 Economic Crime Package restructuring of the economic crime guidelines,3 and the Sarbanes–Oxley Act of 20024 and responsive guideline amendments pushed the fraud sentences called for by the Guidelines to unprecedented levels of severity, particularly for crimes involving large pecuniary losses.5

Of course, the fact that something is unprecedented does not necessarily make it bad. In my view, the transformative insight of the 1987 Guidelines—that economic crime is as morally blameworthy and socially harmful as crimes of other sorts, and thus calls for meaningful incarcerative punishment—was entirely correct. Moreover, at least some of the post-1987 increases in guideline sentences for economic crime were justified, or at the least not manifestly unreasonable. As I will discuss below, there is solid evidence that until roughly 2001–2003 the federal judiciary was in broad agreement with this assessment.6 The problem is that the continual upward ratchet of economic crime sentencing levels got out of hand. Accordingly, I fully concur with Judge Bennett that there is a subset of economic-crime defendants for whom the current Guidelines prescribe sentences that incontestably exceed the bounds of reason.7 The trickier question is correctly identifying that subset.

There is nearly universal agreement that the fraud guideline, section 2B1.1, now prescribes unreasonable sentences for many high-loss defendants.8 There is also a body of opinion, centered principally in the defense bar, but including some federal judges and academic commentators, which holds that section 2B1.1 is pitched too high even for many defendants who inflicted losses in the middle ranges of the fraud-loss table.9 Judge Bennett seems to be in this camp.10 And even among those who think the primary problem is with high-loss defendants, there is considerable divergence of opinion over where, which is to say at what amount of loss, the troublesome high-loss category begins. Reading the work of Judge Bennett and his colleagues has provided me with food for thought on this point and stimulated some new reflections that I share below.11

I also heartily agree with Judge Bennett’s assessment that the Sentencing Commission’s 2015 amendments to the fraud guideline were wholly inadequate to address its deficiencies, and I have enumerated my disappointments with those amendments in testimony to the Commission12 and in subsequent scholarship.13 Moreover, I concur in principle, and sometimes in detail, with four of Judge Bennett’s five suggestions for improving section 2B1.1 (and indeed have made the same or similar proposals in the past myself). In particular, I agree that the loss table should be reworked (“trimmed” or even given a “buzz-cut” as Judge Bennett puts it),14 the number of specific offense characteristics in section 2B1.1 should be reduced,15 the victim table should be revised,16 and the sophisticated means enhancement should be eliminated.17 I will offer a few brief comments on the details of each of these proposals below, and explain why I am unenthusiastic about Judge Bennett’s fifth suggestion—to add a downward departure for lack of pecuniary gain.18

III.     When Did the Fraud Guideline Jump the Shark?

As noted above, there exists a near-universal consensus that the current iteration of section 2B1.1 prescribes insupportably severe sentences for a subset of high-end defendants. This was not always true. Even though the 1987 Guidelines governing economic offenses embodied a paradigm shift in routinely prescribing prison for significant white-collar defendants and the sentences prescribed edged steadily upwards throughout the 1990s, the principal complaints about the Guidelines were: (1) the two guidelines that then covered economic crime—section 2F1.1 for fraud and section 2B1.1 for theft—were duplicative and confusing;19 (2) the definition of “loss,” the measurement central to both the theft and fraud guideline, was poorly conceived and often hard to apply;20 (3) the Guidelines prescribed sentences that were too long for some low-level offenders;21 and (4) the Guidelines prescribed sentences that were too short for some serious offenders.22

Accordingly, when the Sentencing Commission launched its six-year-long project to revise the economic crime guidelines, the project proceeded on two parallel tracks. One part focused on the consolidation of sections 2B1.1 and 2F1.1 into a unitary economic crime guideline and on the redefinition of “loss.”23 The second part focused on measures, primarily adjustments to the loss table, section 2B1.1(b)(1), that sought to make more low-loss defendants eligible for non-incarcerative punishments while at the same time increasing prison sentences for high-loss defendants.24 The result was the so-called Economic Crime Package of 2001, which consolidated and restructured sections 2B1.1 and 2F1.1 into current section 2B1.1, redefined “loss,” lowered sentences for some low-loss defendants, and raised sentences for some mid-to-high-loss defendants.25

As an active participant throughout the process that produced the 2001 Economic Crime Package,26 my assessment is that consolidation made sense and the redefinition of “loss” was fundamentally sound and has largely stood the test of time.27 However, we failed to account properly for the interaction of the enhanced 2001 loss table with the many aggravating specific offense characteristics retained in the consolidated economic crime guideline.28 As a result, sentences prescribed by consolidated section 2B1.1 were pitched too high for some segment of the high-loss offender population.29 This deficiency was compounded by the passage of the Sarbanes–Oxley Act of 2002 and a series of guidelines amendments enacted in response to it, with results widely disparaged today.30

I have told this story at length elsewhere.31 I repeat it here in abbreviated form only because, if one goal of current reform efforts is to recalibrate fraud sentences to levels broadly acceptable by the judiciary, it would be useful to identify a period during which judges generally found them to be so. If such a period existed, it would also be useful to know when, and if possible why, judges became discontented. While researching these remarks, I came across some data that, in combination with data emphasized by Judge Bennett, provides at least hints of answers to these questions.

Consider the following charts. The first is copied from Judge Bennett’s article.32 The second was recreated from a figure that was originally illustrated in a 2006 U.S. Sentencing Commission report.33


Figure 1. Average Sentence Length and Average Guideline Minimum for Section 2B1.1 Offenders (2003–2012)

Screen Shot 2017 03 29 at 9.28.38 AM 


Figure 2. Average Sentence Length and Average Guideline Minimum Quarterly Data for Section 2B1.1 Offenders (Fiscal Years 2002–2006)

Screen Shot 2017 03 29 at 9.29.40 AM 


These charts plot the difference between the average sentence actually imposed by judges on defendants sentenced under section 2B1.1 and the average guideline minimum sentence calculated for such defendants. Both charts reveal an interesting fact about federal economic crime cases over the past decade: namely, that the severity of federal fraud cases as measured by the Guidelines (and thus in significant measure by loss amount34) has steadily increased. But Judge Bennett focuses on a different point: beginning in 2004, judges, on average, sentenced defendants below the bottom of the average guideline minimum and the divergence between the sentence imposed and the average guideline minimum has continued to increase.35 He and his colleagues intimate that this growing gap is evidence of an increasing judicial unwillingness to impose the sentences prescribed by section 2B1.1, at least in some significant fraction of cases, even as the severity of cases as measured by loss has edged steadily upward.36

This is a fair reading of the data, at least for the last few years. But Figure 2, which extends the trend lines back before 2003 suggests a more nuanced story. In short, Figure 2 and other data suggest that, in FY 2002, judges were generally content with the guideline sentences for economic offenses and indeed gave signs of feeling that guideline economic crime sentences were too low in some fraction of cases.37 But FY 2002 represents a key inflection point because, although it was the statistical year during which the Economic Crime Package became legally effective, due to ex post facto considerations the sentencing-enhancing features of the Package would not begin to manifest themselves statistically for some period thereafter.38 The data further suggests that, as the enhanced guideline sentencing levels for medium-to-high-loss offenders dictated by the 2001 Economic Crime Package and the 2003 post-Sarbanes–Oxley amendments began to kick in, judges grew less content and manifested their unease in increasingly high rates of below-guideline sentences, a trend that accelerated after Booker rendered the Guidelines advisory.39

To understand the data bearing on this issue requires a little background. To begin, both before and after Booker, a significant fraction of cases in every offense type have been sentenced below the guideline range due to a formal “departure” or, after Booker, a “variance.”40 For example, in 2002, judges sentenced 34.2% of all defendants below the applicable range, a figure that increased to 50.5% by 2015.41 As for defendants sentenced within range, judges have imposed sentences at the bottom of the applicable range or in its lower half in the overwhelming majority of cases, regardless of offense type. For example, in 2002, 74.6% of all defendants sentenced within-range were either sentenced to the guideline minimum (59.8%) or within the lower half of the range (14.8%).42 In 2015, the percentage was 71.2%.43 The combination of below-range sentences in a third or more of all cases and the overwhelming prevalence of bottom-of-the-range sentences for the remainder would lead one to expect that the average sentence imposed on all defendants in an entire offense category would be lower than the average guideline minimum for those defendants. And that is precisely what the statistics show for every major crime type except economic crime since 2001, the year in which the Commission began publishing such data. For all the other major federal crime types—drugs, guns, and immigration—judges have, on average, consistently imposed sentences lower than the average guideline minimum.44

To me, the most remarkable aspect of the data in Figure 2 is that, in 2002, the year the Economic Crime Package began to take effect, judges were, on average, imposing economic crime sentences higher than the guideline minimum. In 2002, 27.7% of fraud cases were sentenced below the applicable range45 and 54.8% of the remainder were sentenced at the guideline minimum,46 meaning that roughly 66.7% of all fraud defendants were sentenced at or below the guideline minimum.47 Nonetheless, the number of above-guideline-minimum sentences and the magnitude of those sentences combined to completely offset and slightly overmatch all the downward departures so that the average sentence imposed was about one month higher than the average guideline minimum. As a result, unlike sentences for drugs, guns, and immigration offenses, the average sentence imposed by federal judges for economic crimes was within the guideline range.

It is also noteworthy that in the period before and immediately after the effective date of the Economic Crime Package, the rate of downward departures in economic crime cases awarded by judges without a substantial assistance motion48 was strikingly lower than the average for other case types. In 2001, the rate of non-substantial-assistance downward departures for all cases was 18.3%, while the rate for fraud was 11.0% and for larceny 7.3%.49 In 2002, the rate of non-substantial-assistance downward departures for all cases was 16.8%, while the rate for fraud was 9.9% and for larceny 6.2%.50 By comparison, in 2015, judges imposed below-range sentences without a government motion in 30.1% of fraud cases and 28.2% of larceny cases, at a time when the average below-range rate for all cases was 21.3%.51

Taken together, these facts suggest that from 2001 to 2002, the interval just before the Economic Crime Package and the post-Sarbanes–Oxley amendments began to bite, the Guidelines were prescribing economic crime sentences with which judges were generally comfortable. This conclusion may seem at odds with the undoubted fact that the push to increase sentences for high-to-very-high-loss defendants that produced changes in the 2001 loss table was championed by an important contingent of federal judges.52 However, the statistics cited above are consistent with the conclusion that judges were generally satisfied with guideline sentences in the mine run of economic crime cases, but were troubled by the perceived inadequacy of guideline sentences for some serious offenders.

The mere fact that for some period the average sentence imposed on economic crime defendants was above the average guideline minimum remains strikingly discordant with the pattern in other offense types. Anyone who has practiced under the Guidelines recognizes that the essential directionality of plea negotiations and sentencing arguments is down, by which I mean that the default assumption in virtually all plea negotiations is that the bottom of whatever range emerges from a negotiated plea will be sufficiently punitive to satisfy the government. Usually the only remaining question is whether some further diminution of the sentence should be awarded, either with or without the government’s imprimatur.53 In this environment, the fact that, in economic crime cases before 2003, sentences above the bottom of the range more than offset downward departures suggests that this default assumption was not operating for a substantial number of economic crime defendants. Put simply, the data suggest that prior to about 2003 judges felt the guideline minimum was not severe enough in an unusually large fraction of economic crime cases.

This does not necessarily mean that before the Economic Crime Package the economic crime guidelines were set too low across the board. Rather, it suggests that they were set at a level which encouraged judges to use the full guideline range more often than was customary for other offense types, which should be a considered a good thing. After all, a system that designates guideline ranges presumably intends for those ranges to represent realistic sentencing options in a large fraction of cases.

In addition, before 2003, judges were more disposed to depart upwards in economic crime cases than in other types. This inference is supported by statistics on upward departures in the years before and during the transition to the new sentencing levels dictated by the 2001 Economic Crime Package and the post-Sarbanes–Oxley amendments. Upward departures have always been rare (another indicator of the downward directionality of plea negotiations and sentencing practice in the Guidelines era), but as Figures 354 and 455 below indicate, from 2000 to 2004, upward departures in the core economic crime categories of fraud and larceny were consistently more common than average.


Figure 3. Percentage of Upward Departures (Fiscal Years 2000–2004)

Screen Shot 2017 03 29 at 9.31.03 AM  


Figure 4. Percentage All Cases That Are Fraud or Larceny vs. Percentage of All Upward Departures Awarded in Fraud Cases

Screen Shot 2017 03 29 at 9.31.50 AM 

In sum, the data suggests that, in the period just before the 2001 Economic Crime Package and 2003 post-Sarbanes–Oxley amendments began to take effect, the economic crime guidelines were set at a severity level judges found broadly acceptable, if perhaps a smidgeon on the low side in some really serious cases. This conclusion, if correct, has important ramifications for the ongoing debate about the severity level of sentences currently prescribed under section 2B1.1. In a nutshell, if we assume that guidelines ought to be calibrated at levels judges are generally comfortable with imposing, it appears that this objective could be achieved for economic crime by recalibrating sentences prescribed by the Guidelines to roughly
pre-Economic Crime Package levels. Note that this recalibration would not require abandoning either the 2001 consolidation of the formerly separate fraud and theft guidelines into current section 2B1.1 or the 2001 redefinition of “loss.” All that would be necessary would be readjustment of the loss table and reconsideration of some specific offense characteristics.56

Of course, the assumption that Guidelines should hew closely to judicial preferences or comfort levels is fairly debatable. In the economic crime realm in particular, there is an often-stated and I think reasonable fear that, left to their own devices, judges will accord undue leniency to white-collar thieves from their own social class.57 For precisely this reason, the original 1987 fraud and theft guidelines were purposely set higher than prevailing norms to counteract the judicial preference for giving very low sentences to white-collar criminals.58 Nonetheless, it is not unreasonable to conclude that nearly 30 years under the Guidelines have entrenched a new, and more punitive, norm that would not be entirely erased by a modest recalibration to roughly pre-November 2001 levels.

But even if we were to agree that reversion to roughly pre-Economic Crime Package sentencing levels would, on average, be more in line with judicial preferences, that would not tell us exactly what changes we ought to make in the current guidelines. The data on average sentences is, at best, a very rough gauge of the views of judges about any one of the many different types and classes of economic offender. For example, these averages standing alone are consistent with at least two quite different views of pre-2001 economic crime sentencing: (a) that the severity of sentences under the
pre-2001 regime was about right for all types and grades of offenders; or
(b) that, as was the common wisdom among the stakeholders in the Economic Crime Package debate, high-loss sentences were sometimes too low and low-loss sentences were sometimes too high. The averages cannot tell us whether to revert to pre-2001 sentencing levels for all offenders or to accept some of the changes made during 2001–2003, while rejecting others.

III.     How Should Section 2B1.1 Be Recalibrated?

A.     The Bennett Study and Recalibrating Section 2B1.1

Among the most interesting and thought-provoking features of the article by Judge Bennett and his colleagues is its study of how 180 federal district court judges, federal magistrate judges, and state judges responded to a sentencing case study.59 The judges were presented with a hypothetical case in which the defendant, a corporate insider, committed securities fraud against his own company causing a loss of approximately $7 million.60 The judges were told that the defendant had entered a plea pursuant to Federal Rule of Criminal Procedure 11(c)(1)(C) obliging the court to sentence within a range of 151 to 240 months (12 ½ to 20 years) which roughly corresponded to the sentencing guideline range for such an offense.61 The judges were asked to choose a sentence within this range, but not allowed to sentence outside of it, and were also asked questions about their personal characteristics and sentencing philosophy.62

The results reported by Judge Bennett illuminate a number of issues relating to judicial sentencing practice, particularly questions about how political affiliation, age, gender, sentencing philosophy, and other personal characteristics influence a judge’s sentencing decisions.63 However, I will focus here only on the extent to which the study assists in deciding how to recalibrate section 2B1.1. Despite its many virtues, the Bennett study is not very useful in that regard. Indeed, I fear that, on this point at least, the study’s authors may be over-interpreting their data.

The study reports that 75.0% of federal district court judges, 55.0% of federal magistrate judges, and 27.1% of state judges sentenced the hypothetical defendant to the minimum available sentence of 151 months.64 The authors interpret these figures as demonstrating dissatisfaction by the federal district judges with the 151-month minimum sentence, and by extension with the severity of the guidelines that purportedly dictated this minimum.65 Although I agree that there is considerable dissatisfaction among members of the federal district court bench with guideline sentences in many high-loss cases, I cannot agree that this study does much to prove the existence of such dissatisfaction. Nor do I think it helps very much in pinpointing the set of cases judges find most troubling. Moreover, the response of state court judges to the study’s scenario significantly undercuts the notion that sentences in the range of 12 to 15 years for serious economic crime are self-evidently unreasonable.

Consider first the federal district judges. To begin, we know that, in real life, federal judges have complete discretion to sentence below the applicable guideline range. We also know that, in real life, when they sentence within range, they overwhelmingly sentence at the bottom of the range.66 But because the study artificially precluded judges from sentencing below the prescribed range, it impossible to say how many of the 75% who sentenced at the bottom of the range thought that 151 months was too severe a punishment for this defendant. For all we know, most or even all such district judges thought that 151 months was entirely appropriate.

Moreover, the model presentence investigation report (“PSI”) provided to the study participants is strikingly devoid of the kinds of mitigating circumstances and humanizing personal characteristics that are standard fare in real white-collar cases of this type. I have no doubt that Judge Bennett and his colleagues structured the PSI in this way to avoid the confounding effects of such information. But the result is that the experienced federal district judges in the study were presented with what would appear to them as an atypical white-collar PSI, one from which they might well conclude that the defendant merited a sentence at the bottom of the substantial prescribed range, but no lower.

It is only by looking outside of the Bennett study and comparing the 75% figure with what we know from the Commission’s data on actual sentences that we can conclude that at least some of the judges who awarded the guideline minimum would very probably have sentenced below 151 months if permitted by the parameters of the study. In 2014 over one-quarter of all fraud cases were sentenced below range without a government motion;67 it is thus fair to surmise that some of the study judges would have sentenced below range if allowed. But we cannot know how many judges would have done so.68

One point on which the Bennett study is of no material assistance is whether federal district judges consider their hypothetical guidelines sentencing range for a $7 million securities fraud to be more or less appropriate than the guideline sentences for other types and sizes of fraud. Since the study participants were offered no hypothetical cases with different loss amounts or other facts, one cannot tell whether the judges would have been more or less disposed to impose low-end sentences in such cases.

In short, the most one can say of the Bennett study is that: (a) it is consistent with the general pattern of federal district judges sentencing disproportionately at the low end of the applicable guideline range in all case types, including white-collar crime; and (b) it is not inconsistent with other data showing that federal district judges sentence below the applicable guideline range in many white-collar cases, particularly those with large loss amounts.

The most interesting aspect of this section of the study is the markedly different results reported for federal magistrate judges and state judges. Forty-five percent of magistrate judges and 73% of state judges sentenced the hypothetical defendant above the prescribed 151-month minimum.69 Judge Bennett reads the marked difference between the reactions of these judges and the federal district judges who imposed an above-minimum sentence only 25% of the time as signaling a high level of “resistance to a recommended sentence” among federal district judges, perhaps arising from “experience with (or rejection of) overly harsh sentencing guidelines.”70 This is not an implausible hypothesis. But, as noted above, the limitation of the Bennett study to a single scenario and its prohibition of downward departures from the stipulated sentencing range diminishes the study’s usefulness in determining how many participating federal district judges disapproved of the stipulated range on normative grounds.

Far more intriguing is the fact that the majority of federal magistrate judges71 and the overwhelming majority of state judges seemed perfectly comfortable with the stipulated range of 151 to 240 months. Indeed, the mean sentence imposed by state judges was 174.81 months or 14 ½ years, almost two years above the stipulated minimum.72 The limited scope of the Bennett study prevents drawing any definitive conclusions about this difference in perspectives. But two hypotheses suggest themselves.

First, large fraud cases are much more common in federal than state court and federal district judges are more likely than state court judges to have sentenced a number of such cases or talked with colleagues who have done so.73 Consequently, as intimated by Judge Bennett, they are likely to be familiar with the foibles of the guidelines that generate lengthy guideline ranges for such defendants,74 and they will have encountered defendants in such cases as real persons rather than hypothetical constructs. By contrast, federal magistrate judges do not sentence anyone. And state judges, though they sentence people regularly, rarely encounter a multi-million-dollar fraud case. A $7 million fraud of the sort employed in the study is a pretty big deal, even in federal court. For example, in FY 2012, only 291 of 8501 federal economic-crime defendants, or 3.4%, were sentenced for frauds with losses exceeding $7 million.75 Although I am unaware of any definitive data on the point,76 my experience as a both a state and federal prosecutor suggests that the number of such cases in state courts is vanishingly small. It is not unreasonable to suppose that state judges were willing to impose long sentences in a $7 million fraud because the defendant had committed a crime far more severe by any ordinary reckoning than any non-violent case they see.

Second, although I advance this second hypothesis far more tentatively, it may be worth observing that federal district judges are better paid than magistrate judges and virtually all state trial court judges,77 enjoy higher social status, and not uncommonly come to the bench from well-compensated private practice or federal government practice backgrounds.78 To the extent that there may be a class bias in favor of white-collar defendants, federal district judges are at least arguably more likely to share the class characteristics of white-collar defendants than magistrate judges or state judges.

Finally, the comfort expressed by state court judges with sentences in the 12 to 15 year range for a large, but not extraordinary, fraud suggests that sentences of this magnitude are not self-evidently unreasonable for such a case. After all, a group of experienced, and perfectly reasonable, judicial officers approved of them. The difference between federal district court and state trial court judges may be proof, not that state judges are unduly punitive, but that federal judges can become unduly blasé about serious economic crime.

As noted, I join Judge Bennett and his colleagues in the broad consensus that the Guidelines as now configured dictate unreasonable sentences for certain economic offenders, particularly in cases where a large loss combines with other aggravating specific offense characteristics. But the Bennett study does not help very much in identifying a level or zone in which the current Guidelines begin to generate unreasonably punitive sentences. My personal sense is that the range of truly problematic sentences for serious white-collar offenses begins right around the low end of the range in the Bennett study. Or to put it another way, my sense is that, except in an extraordinary case, judges will strongly resist sentencing a non-recidivist white-collar offender to any term higher than the 10 to 15 year range. The Bennett study is at least consistent with this notion, and further evidence for such resistance can be found in Sentencing Commission data that correlates loss amount with sentences imposed.

Figure 5 below, reproduced from the Bennett article and employing data published by the Sentencing Commission staff,79 shows that the average sentence imposed in FY 2012 in large-loss cases (those exceeding $2.5 million) ranged from just under five years to just under 12 years (the sole exception being a pair of cases in the $100 to $200 million range where the average sentence was 210 months or 17.5 years). Of course, the average sentence imposed in such cases is pulled down by many substantial assistance departures and other government-sponsored accommodations. Moreover, many defendants were sentenced above the reported averages and some well above it, but it is plain that the norm in such cases is a sentence far below the much higher levels prescribed by the Guidelines,80 and sentences above 10 to 15 years are an anomaly even in the largest fraud cases.

Figure 5. Average Sentence Length and Average Guideline Minimum for Each Section 2B1.1 Loss Table Category (2012)

Screen Shot 2017 03 29 at 9.33.34 AM


B.     Look to 2001 for a Reset Point for Section 2B1.1

If the Bennett study offers little help in determining how to recalibrate section 2B1.1, where else might one look? In testimony to the Sentencing Commission, I have suggested that section 2B1.1 should be recalibrated by “set[ting] a ceiling on the punishment government can justifiably mete out for the most serious financial crimes, and then . . . scal[ing] the punishments for lesser forms down from that maximum.”81 Although I continue to think this would be the theoretically optimum approach, cold-bloodedly specifying a maximum sentence for the most serious financial crimes would be both conceptually and politically difficult.

The conceptual problem is twofold. First, one must identify the combination of factors that make up the most serious type(s) of financial crime. Second, using some rationally justifiable criteria, one must assign a maximum punishment to that grade of crime. You might say that the Guidelines have already performed the first task of identifying the most serious crime types by placing loss amount at the center of section 2B1.1 and then adding aggravating specific offense characteristics. By this logic, whatever class of offenses generates the highest offense level under section 2B1.1 is the most serious financial crime. The problem is that the offenses and offenders who score highest under the Guidelines do so not because of any considered comparison of their crimes to other types, but simply because of the higgledy-piggledy accretion of specific offense characteristics over the past three decades. It might well be that the defendants who now receive the highest offense level scores under the Guidelines are indeed the class of defendants we as a society think are most deserving of punishment, but no one has ever actually made that determination. I would be pleased if the Sentencing Commission undertook a serious normative evaluation of this problem, but it would be a tough task.

If the Commission performed that task, it would still face the second conceptual problem: assigning a number—a particular term of years—as the most severe punishment that ought to be imposed for a financial crime (absent the most extraordinary circumstances, which would justify an upward departure or variance). This may be an even tougher task that identifying the most serious financial offense type. After all, correlating the commission of a specific crime with a specific increment of judicially prescribed pain is more an exercise in metaphysics than logic. In practice, judges and sentencing commissions rarely, if ever, confront that philosophical conundrum, but look instead to markers like past sentencing patterns. Accordingly, if the Commission could identify the most serious types of economic crime, it could analyze sentencing data from federal and state courts to determine a rough consensus about the maximum sentence for such offenses that would provide the top anchor from which the remainder of the guidelines would be scaled downward. To describe this process is to reveal its inherent uncertainty.

Even if conceptual obstacles could be overcome, a completely rational redesign of section 2B1.1 starting with a designated maximum guideline sentence for the worst economic crimes would face political obstacles, as well. As presently configured, section 2B1.1 prescribes sentences of 30 years or even life imprisonment plus multiple additional offense levels in a fair number of fraud cases each year.82 Except for once-in-a-blue-moon instances like Bernie Madoff, who received 150 years,83 judges will not impose such sentences. Nonetheless, any effort to scale back guideline fraud sentences by baldly declaring that the maximum guideline sentence for all economic crime cases will henceforth be, say, 20 or 25 years, is likely to be vigorously opposed as “soft on crime” by some in Congress and some elements of the Justice Department (“DOJ”). Opponents will inevitably point to Madoff and a few other high-profile bad actors as evidence that the selected guideline maximum is not enough punishment for these heinous poster children, while ignoring the reality that the advisory guidelines system leaves judges at perfect liberty to sentence far above that maximum in truly extraordinary cases. Thus, rather than identifying a single specific maximum guideline sentence on which opponents can fixate, it may be easier to achieve scaled-back fraud sentences by arguing for reversion to a previous systemic norm for which there was demonstrable judicial support.

Which is where the data concerning the 2001 inflection point comes in. We know that, at least on average, sentences for economic crimes in 2001 were broadly acceptable to judges.84 If our present aim is to recalibrate section 2B1.1 in a way that judges find useful and satisfactory, it makes sense to use pre-2001 sentencing levels as a reference point. This does not, of course, mean that the Commission should simply void every amendment to the economic crime guidelines from November 2001 forward. It merely suggests that 2001 sentencing levels provide a helpful frame of reference to be considered together with other data.

IV.     Specific Remedies

The remainder of this Article will discuss the remedies prescribed by Judge Bennett and colleagues for the ailing economic crime guideline.

A.     The Loss Table

Critics of section 2B1.1 often complain that “loss,” the reasonably foreseeable pecuniary harm caused or intended by the defendant,85 has too large an effect on guideline sentences. Although the current definition of “loss” has its detractors,86 the core criticism of section 2B1.1 is that loss, however defined, adds too many offense levels to defendants’ guideline calculations and thus increases sentence length by unduly large amounts. As I noted recently:

[T]he loss table of the November 2015 version of Section 2B1.1 has fifteen two-level steps pursuant to which loss amount can add from two to thirty offense levels. A first-time offender convicted of mail or wire fraud now begins with a base offense level of 7, meaning that, loss and other specific offense characteristics aside, his or her starting sentencing range is 0–6 months. Accordingly, to say that loss can add two to thirty offense levels means that loss amount alone can now raise such a defendant’s offense level to 37, and the guideline range all the way to 210–262 months (17.5–21.8 years).87

As will be discussed in the next section, the impact of so large an upward adjustment for loss alone is compounded when multiple section 2B1.1 specific offense characteristics and Chapter 3 role and victim adjustments are added.

Judge Bennett recommends that the section 2B1.1 loss table be given a “buzz-cut.”88 I concur in the sentiment, but, as I am sure Judge Bennett would agree, the “buzz-cut” imagery with its suggestion of simply whacking a bunch of offense levels off the top understates the degree of tonsorial artistry that will be required to reconfigure the loss table properly. Reconfiguration of the table requires addressing two problems: first, what is the maximum influence even the largest loss amount should have on an economic crime sentence? Or putting it more concretely, what is the maximum number of offense levels that any defendant should be assigned for loss, however large? Second, should the Commission simply eliminate some number of levels at the top of the loss table, leaving everything below that level the same? Or should it also recalibrate the amounts of loss associated with each remaining two-offense-level increment?

To illustrate the interaction of these two problems, I reproduce below the 2015 version of the section 2B1.1 loss table.89 Suppose, one wanted to reduce the maximum number of offense levels chargeable to a defendant based on “loss” from the 30 levels possible today to the 20 levels available under the 2000 guidelines. The simplest way to achieve that end would be to eliminate the top five increments of the loss table (as I have indicated by striking through them in Figure 6). Although this might be a beneficial step in itself, it would leave a table that: (a) made no distinction between a $9.5 million fraud and a $550 million fraud; and (b) provided no reduction or modification of current guideline sentences for any defendant with a loss less than $9.5 million. On the former point, some may be of the view that judges should simply employ their post-Booker discretion to differentiate between very big and stupendously large frauds. That may be so, but the difference between stealing $9.5 million and stealing half-a-billion dollars is so large that perhaps the Guidelines should formally recognize some intermediate interval(s) in that range. On the latter point, given that the primary complaint about the current guideline is its tendency to over-punish high-loss offenders, one would think that any redesign of the loss table should give some attention to cases below the $9 million range. In any case, Judge Bennett intimates,90 and I agree, that the Commission would need to give careful thought both to the aggregate number of offense levels available for loss amount and to resetting the loss amount triggers for each two-offense-level bump.


Figure 6. The Values of the 2015 Loss Table

Screen Shot 2017 03 29 at 9.35.54 AM


I will add only a few more thoughts about how the loss table ought to be restructured. First, as Judge Bennett notes, I advised the Commission in 2015 to eliminate the top four steps on the existing offense table.91 This would make the maximum amount on the loss table $25 million and the maximum enhancement for loss 22 levels. Judge Bennett views this suggestion as inadequate, primarily because the loss enhancement of 22 offense levels for the largest cases would remain too high, and perhaps also because he views the class of cases requiring sentencing relief as beginning with losses lower than $25 million.92 I do not entirely disagree on either point. However, my proposal to the Commission was made with political realities in mind and it seemed to me that a four-level haircut was the absolute maximum the Commission might consider. Even that assessment proved over-optimistic since the Commission cut no levels from the table. If permitted to leave political calculations to one side, I would recommend that the maximum number of steps on the loss table be reduced by five (bringing the total number of steps on the table from 16 to 11). This would reduce the maximum number of offense levels attributable to loss from 30 to 20. The result would be a return to the number of loss table steps and loss-based offense levels in effect prior to the 2000 pre-Economic Crime Package level.93

Of course, a fair argument could be made that the 2001 and 2003 revisions of the loss table were responsive to judicial and DOJ concern over unduly low sentences for serious fraud offenders and to fairly clear congressional directives to increase sentences for high-loss defendants.94 If the Commission is unwilling to roll back the loss table all the way to 2000 levels, it should nonetheless consider eliminating only the top four steps on the table, making the maximum loss adjustment 22 offense levels. That would represent real progress toward rationalizing high-end sentencing without openly repudiating congressional directives and stakeholder views.

Second, if some number of loss table steps were removed, I do not have much detailed advice about how to recalibrate the surviving 20 or 22 steps. Assuming the primary discontent with the current table is large-loss cases, the first question will be what constitutes a “large” loss and thus at what loss table step and corresponding loss amount the Commission should begin its revisions. I have created in Figure 7 below a Model Loss Table by way of illustration. It consists of only 11 steps and modifies the top four remaining loss steps by increasing the loss amount that correlates with each two-offense-level increase. The numbers I have inserted are largely notional.95 An actual recalibration should be based on careful study of the distribution of cases, loss amounts, and sentences in the overall high-loss category.


Figure 7. Model Loss Table

 Screen Shot 2017 03 29 at 9.36.28 AM


B.     Specific Offense Characteristics (“SOCs”)

Judge Bennett and friends also recommend a “buzz-cut” to the large number of specific offense characteristics (“SOCs”) in section 2B1.1.96 I have argued for some years that, particularly in high-loss cases, the interaction of the current loss table with multiple SOCs, some of which are highly correlated with large losses, can produce insupportably high sentences.97 And I agree with Judge Bennett that some of the SOCs in section 2B1.1 are of relatively little value in distinguishing relative offense severity.98

That said, I think Judge Bennett somewhat mischaracterizes the problem with section 2B1.1 SOCs. He begins by quoting the observation of Professor Kate Stith and Judge Jose Cabranes that “the Commission has never explained the rationale underlying any of its identified specific offense characteristics, why it has elected to identify certain characteristics and not others, or the weights it has chosen to assign to each identified characteristic.”99 He implies that the Stith-Cabranes statement applies to section 2B1.1 SOCs. This is, at best, only half-right.

The rationales for virtually all of the Guidelines’ economic crime SOCs have been extensively explored by the Commission, its staff, and system stakeholders and observers, and the Commission’s rationales have customarily been formally articulated in published “Reasons for Amendment.”100 It is true that the Commission’s amendment process was formerly more opaque, but it has improved the transparency of its decisional process over the years and, at least since the period leading up to the 2001 Economic Crime Package, all the Commission’s decisions about economic crime sentencing, including the SOCs of section 2B1.1, have been the subject of robust and open debate capped by Commission decisions with explanations.101 One can disagree with the Commission’s choices on economic crime over the past 15 years, but one cannot fairly claim that the Commission has failed to explain them.

Moreover, the rationales for virtually all the section 2B1.1 SOCs are self-evident. For example, section 2B1.1(b)(2)(A) prescribes a two-level increase for committing an offense involving ten or more victims, employing mass marketing, or causing substantial financial hardship to one or more victims. Stealing from multiple victims is, all else being equal, worse than stealing from one person. Employing mass marketing risks causing harm to larger numbers of persons than less sophisticated techniques. And stealing so much from a single person that it endangers that person’s solvency is worse than stealing an amount that is merely inconvenient.

Judge Bennett also states that most economic crime SOCs have “no empirical basis.”102 This has become a fairly common trope in post-Booker critiques of particular guideline provisions, but I have never known quite what to make of it. Any guidelines system is the result of a melding of three processes: (1) an effort to rank the relative seriousness of different kinds and degrees of criminal behavior; (2) an effort to assess the social dangerousness and personal culpability of individual defendants; and (3) an effort to correlate offense seriousness rankings and personal culpability assessments with ranges of punishment.

Empiricism may be of some assistance in the latter two processes. For example, if a guidelines system enhances punishment based on a defendant’s criminal history on the theory that prior conduct is predictive of recidivism and thus of enhanced social dangerousness, that theory can be empirically tested. Similarly, if one purpose of punishment is to reduce crime through deterrence or promoting rehabilitation, empirical studies might be able to assist the process of setting the range and type of punishment most likely to reduce recidivism for particular classes of offender. But even the processes of assigning significance to prior criminal history and setting punishment levels are heavily driven by non-quantifiable considerations. For example, if it were proven empirically that repeat offenders present no increased recidivism risk, we might conclude that repeat offenders are more morally culpable than first-time offenders and deserve greater punishment on that ground. Punishments for many crimes are driven by considerations of moral desert, rather than crime prevention. We sentence murderers to long terms or execute them, even though as a class they have a lower recidivism risk than most other offenders.103

In any case, the process of ranking offense seriousness—which is what Chapter Two of the Guidelines and all the many SOCs therein are about—has little to do with empirically falsifiable determinations. Offense seriousness ranking is driven primarily by moral judgment and social policy. The determination that premeditated murder is a more serious offense than a negligent homicide is not empirically testable. Nor is the determination that criminal homicide of any type is usually, but not always, a more serious offense than robbery.104 Both are essentially moral judgments. The same is true of the conclusion that stealing from multiple victims is worse than stealing from one, or that stealing so much from one person that his financial security is endangered is worse than stealing the same amount from another person who could more easily afford the loss. Relatedly, when the Guidelines rank stealing a lot of money from a federal health care program as a more serious offense than stealing from other entities,105 that ranking represents a social policy judgment that fraud against such programs is a major problem that the federal criminal law should be employed to prevent. The scope of health insurance fraud may be empirically measurable, but the decision to rank health care fraud as a more serious, and more severely punishable, class of fraud does not rest on any empirically testable proposition.106

The problem with the section 2B1.1 SOCs is not that, taken individually, they are irrational or lack empirical support. The problem, and here I concur entirely with Judge Bennett, lies in the second half of the Stith-Cabranes complaint, namely that the Commission has failed to give adequate consideration to the weight accorded these SOCs, and in particular to their interaction with each other and with the loss table. As I wrote when commenting on the Commission’s 2015 amendments:

The cumulative effect of the evolution of the economic crime guidelines from 1987 to 2000, the 2001 Economic Crime Package, and the 2003 Sarbanes-Oxley-driven amendments has been: (1) to give loss amount ever-increasing weight, while (2) progressively teasing out a long list of factors that are highly correlated with big-dollar frauds—and for which, therefore, the large enhancements in the Loss Table are already a proxy—and giving those factors independent, cumulative, logarithmic weight in setting a sentence.107

Accordingly, the current challenge is figuring out how to reduce the cumulative effect of economic crime SOCs, particularly in high-loss cases. Judge Bennett recommends a comprehensive review of section 2B1.1 SOCs conducted with a “presumption that most should be removed unless there is solid empirical data to support [them].”108 As noted, I doubt that empiricism has much to say about SOCs.109 But I agree that a thorough review of section 2B1.1 SOCs would be desirable and that a good many could be either removed altogether or transformed into factors judges are encouraged to consider when setting sentences within range or for purposes of departure or variance.

That said, there are likely to be several practical impediments to this project. First, given the extreme modesty of the Commission’s 2015 amendments, which were adopted as the long-awaited culmination of a multi-year study of fraud sentencing in which all the critiques Judge Bennett and I share were thoroughly aired, one is disposed to doubt that the Commission has the appetite for so detailed an undertaking. Perhaps, as several commissioners have intimated, they may revisit fraud sentencing as part of a general overhaul of the entire Guidelines structure,110 but for the near term, I am not holding my breath.

Second, as Judge Bennett acknowledges,111 a good many of the section 2B1.1 SOCs were enacted by the Commission in response to congressional directives, and the Commission has customarily declined to undo amendments of this sort.112 One might argue that the Commission’s unwillingness to revisit such SOCs is unduly timid. Except in the rarest of instances, the so-called “congressional directives” did not legally require the Commission to enact any particular guideline change. Rather, they customarily took the form of language requiring the Commission to review the existing Guidelines and “ensure” the adequacy of sentences for crimes of a specified type.113 There is no denying the strength of the hint embodied in such language, but the Commission could have concluded that the then-current sentences were perfectly adequate and declined to take the hint. Likewise, there is nothing in the so-called directives that bars the Commission from revisiting their earlier choices and concluding that some or all of the congressionally-stimulated increases have proven excessive or counterproductive and should now be revisited. However, there is little indication that the Commission is any more disposed to boldness now than it has ever been.

It was in recognition of the Commission’s practice of treating any SOC enacted in response to Congress as effectively inviolable that in 2015 I recommended placing a numerical cap on the cumulative effect of SOCs, rather than trying to winnow them out of the Guidelines individually.114 Absent a wholesale rewrite of the section 2B1.1, this still seems the most promising approach to the problem of sentence inflation by SOC.

C.     The Victim Table

Judge Bennett maintains that both the current multiple-victim enhancement of section 2B.1.1(b)(2) and its pre-2015 predecessor are deficient and unduly inflate sentences in certain classes of fraud cases.115 I agree. I have discussed my concerns elsewhere and will not repeat them here.116 I also agree with Judge Bennett that the ABA’s proposal to rank “victim impact” seems highly subjective and unlikely to prove useful.117 I will only add that I am not convinced that Judge Bennett’s alternate proposal, to substitute assessments of victims’ “emotional distress,”118 would adequately address the problem. Not only does assessment of emotional state seem highly subjective, but I am not sure that it is desirable to impose different punishments on defendants based on differing emotional responses by victims to otherwise identical defendant behavior. Moreover, I am not sure that the Bennett alternative accounts for two of the primary issues the current guideline is designed (however imperfectly) to address—the fact that stealing from multiple victims is, ceteris paribus, worse than stealing from one victim, and the fact that the same quantum of monetary loss can have differential economic impacts on different persons.

In the end, I tend to think that the Guidelines should eliminate both victim number and assessments of special financial impact as SOCs triggering specific offense level enhancements. Both factors would be better addressed as considerations in setting a sentence within or outside of the guideline range. At a minimum, the size of the existing victim enhancements under section 2B1.1(b)(2) should be markedly decreased. It is now possible to receive a six-level increase for causing substantial financial hardship to 25 or more victims.119 Because of the logarithmic character of the Guidelines’ sentencing table, this six-level bump roughly doubles a defendant’s guideline sentence. Given that causing this much harm to 25 people would necessarily involve a very significant pecuniary loss, with a correspondingly large offense-level increase from the loss table, doubling the resultant guideline range for victim hardship seems gratuitously punitive.

D.     Sophisticated Means

Judge Bennett urges elimination of the sophisticated means enhancement.120 I agree.121

 E.     Departure for Lack of Pecuniary Gain

Judge Bennett recommends addition of an encouraged downward departure in cases where “the offender’s gain is substantially lower than the loss the offender caused.”122 The notion that a defendant’s punishment should be lower when his personal gain is less than the victim’s loss has been around for a long time. A departure on this ground was considered and rejected as part of the 2001 Economic Crime Package,123 but variations of the idea continue to surface.124 I cannot endorse the suggestion.

As I have explained in detail elsewhere, offense seriousness rankings are based on three factors: the magnitude of the harm caused by the crime, the magnitude of the harm risked by the crime (sometimes even when the harm did not occur), and the defendant’s culpable mental state.125 “Loss,” when actual, is a direct measurement of harm and a decent, if imperfect, proxy for mental state.126 Loss, when unrealized but intended, is a direct measurement of culpable mental state and a decent, if imperfect, proxy for harm risked.127 By contrast, the personal gain of any single defendant is not a reliable direct measurement of harm, risk of harm, or mental state.

Gain equates to harm only insofar as the defendant’s gain derives from the victim’s loss. But there are many cases in which the gain of the defendant being sentenced is only a fraction of the loss to the victim and it is difficult to see why this fact should affect sentencing. To take the simplest case, if three crooks swindle me out of $1 million and then split the take equally, the gain to each defendant will be only one-third of my loss. Why should these three crooks receive a lower sentence than if only one of them had inflicted the same amount of harm? Or suppose in the same case that one of the three crooks, after swindling me, gives his confederates only $10,000 apiece and sticks to the rest himself. Why should the two who were foolish enough to believe in honor among thieves receive a lower sentence? Given that the three joined together to harm me, why should the later perfidy of one co-conspirator have any effect on the sentence of the others?

Numerous other examples could be offered. The imperfect correlation between harm and gain is indicated by the fact that the guidelines allow gain as a measurement of loss (the primary measurement of harm in economic crime) only where loss itself cannot be reliably determined.128

The disposition to compare gain and loss may stem from an imperfectly articulated intuition that gain relates to culpable mental state. The idea may be that securing all or most of the victim’s loss for oneself signifies greed, which may be thought to be a more blameworthy mental state than mere intention or willingness to cause a loss in the first place. I confess that I cannot follow this logic. Greed in the sense of a desire for material wealth is a constant of the human condition. It only becomes criminal when it is coupled with a choice to break the law and harm others to secure wealth. Consider a securities fraud case in which corporate insiders lie about the financial status of their company in order to inflate the value of their own stock options, or just to fool the board of directors into retaining them in their positions and continuing their salaries. The loss to investors in such cases can run to tens or hundreds of millions of dollars, but the “gain” to the defendants may be only tens or hundreds of thousands. Or suppose that I commit insurance fraud by hiring you to set a fire to my house. If I pay you $10,000 for the arson and later collect $500,000 from the insurance company, should you be considered less responsible for the destruction your arson caused because you got a smaller payout? As a general matter, why should a defendant’s willingness to impose large harms on another in order to achieve only modest gains for himself count as a mitigating factor?

Perhaps the best explanations for the persistent interest in comparisons of gain and loss are these: first, in some multi-defendant cases, a particular defendant’s small gain relative to overall victim loss may signify that the small-gain defendant played a relatively minor role as compared to other defendants who got more money. Second, sometimes the relatively small size of the defendant’s gain in comparison to the victim’s loss may indicate that the defendant set out to commit a crime of modest financial benefit to himself, but ended up causing large, but foreseeable, financial harms to others. In some such cases, Guidelines critics think the definition of actual loss as “reasonably foreseeable pecuniary harm that resulted from the offense”129 over-inflates loss and over-punishes defendants. I understand both concerns, but a downward departure provision triggered merely by a disproportion between gain and loss is too crude a tool to alleviate these specific concerns. If they are to be addressed in the Guidelines themselves, language particularly directed at each should be inserted.

V.     Conclusion

I close by reiterating my admiration for the work of Judge Bennett and Professors Levinson and Hioki. Their article is a significant addition to the literature on federal economic crime sentencing. We agree on all the major points. I hope their work is read and carefully considered by the Sentencing Commission and other stakeholders in the ongoing conversation about how economic crime sentencing can be improved. I offer the foregoing comments only to sharpen that conversation regarding some potentially important details and I hope they will prove of some use to that end. 


  1. [1].     Mark W. Bennett et al., Judging Federal White-Collar Fraud Sentencing: An Empirical Study Revealing the Need for Further Reform, 102 Iowa L. Rev. 939 (2017).

  2. [2]. Id. at 948–49; see also Frank O. Bowman, III, Pour Encourager les Autres? The Curious History and Distressing Implications of the Criminal Provisions of the Sarbanes–Oxley Act and the Sentencing Guidelines Amendments that Followed, 1 Ohio St. J. Crim. L. 373, 385 (2004).

  3. [3]. See generally Frank O. Bowman, III, The 2001 Federal Economic Crime Sentencing Reforms: An Analysis and Legislative History, 35 Ind. L. Rev. 5 (2001) (describing and analyzing the 2001 Economic Crime Package restructuring of the federal economic crime sentencing guidelines).

  4. [4]. Sarbanes–Oxley Act of 2002, Pub. L. No. 107–204, 116 Stat. 745 (codified in scattered sections of 15 U.S.C.).

  5. [5]. Bennett et al., supra note 1, at 951–52; see Bowman, supra note 2, at 387–89, 405–06; see also Frank O. Bowman III, Sentencing High-Loss Corporate Insider Frauds After Booker, 20 Fed. Sent’g Rep. 167, 168 (2008).

  6. [6]. See infra notes 33–51 and the accompanying text.

  7. [7]. Compare Bennett et al., supra note 1, at 973 (referring to “long-known shortcomings of the [economic-crime provisions of the Sentencing Guidelines], especially as they relate to the exceedingly harsh punishments of high-value economic criminals”), with Frank O. Bowman, III, Damp Squib: The Disappointing Denouement of the Sentencing Commission’s Economic Crime Project (and What They Should Do Now), 27 Fed. Sent’g Rep. 270, 270 (2015) (contending “that for high-loss defendants the fraud guideline is... ‘fundamentally broken’” (quoting Patti B. Saris, Chair, U.S. Sentencing Comm’n, Remarks for Public Meeting 2 (Jan. 9, 2015),

  8. [8]. See Bennett et al., supra note 1, at 973–74; Bowman, supra note 7, at 270 (noting the widespread dissatisfaction with section 2B1.1 in high-loss cases and the concession by U.S. Sentencing Commission Chair Patti B. Saris that the guideline does not appear to work for cases where the loss exceeds one million dollars).

  9. [9]. See, e.g., James E. Felman, Chairman, Am. Bar Ass’n Criminal Justice Section, Testimony on Behalf of the American Bar Association Before the United States Sentencing Commission for the Hearing on Proposed Amendments to the Federal Sentencing Guidelines Regarding Economic Crimes 9 (Mar. 12, 2015), (urging the Commission to “address the overstated impact of loss on the guidelines at all points in the loss table”).

  10. [10]. See Bennett et al., supra note 1, at 973–75.

  11. [11]. See infra notes 69–80 and the accompanying text.

  12. [12]. See generally Frank O. Bowman, III, Comment on Proposed Amendments to Economic Crime Guideline §2B1.1 (2015),; U.S. Sent’g Comm’n, Public Hearing on Proposed Amendments to Federal Sentencing Guidelines 172–82, 189–93, 197–98, 200–02 (2015), (comments of Frank O. Bowman, III).

  13. [13]. Bowman, supra note 7, at 276.

  14. [14]. Bennett et al., supra note 1, at 982–84.

  15. [15]. See id. at 985–86.

  16. [16]. Id. at 986–87.

  17. [17]. Id. at 988.

  18. [18]. Id.

  19. [19]. Bowman, supra note 3, at 23–25.

  20. [20]. Id. at 25–29.

  21. [21]. Id. at 29–30.

  22. [22]. Id.

  23. [23]. Bowman, supra note 2, at 387.

  24. [24]. Id. at 387–88.

  25. [25]. See generally Bowman, supra note 3 (describing and analyzing the 2001 Economic
    Crime Package).

  26. [26]. See generally, e.g., Bowman, supra note 3; Frank O. Bowman, III, A Judicious Solution: The Criminal Law Committee Draft Redefinition of the “Loss” Concept in Economic Crime Sentencing, 9 GEO. MASON L. REV. 451 (2000) (describing and explaining the work of the Sentencing Guidelines Subcommittee of the Criminal Law Committee of the Federal Judicial Conference, to which I was the academic advisor, on the 2001 Economic Crime Package); Frank O. Bowman, III, Back to Basics: Helping the Commission Solve the “Loss” Mess With Old Familiar Tools, 10 Fed. Sent’g Rep. 115 (1997) (outlining a proposed redefinition of “loss” that emphasizes the foreseeability of pecuniary harm caused by the defendant); Frank O. Bowman, III, Coping With “Loss”: A Re-Examination of Sentencing Federal Economic Crimes Under the Guidelines, 51 Vand. L. Rev. 461 (1998) (setting out and explaining the provisions of a proposed consolidated economic crime guideline).

  27. [27]. See generally Frank O. Bowman, III, “Loss” Revisited: A Guarded Defense of the Centerpiece of Federal Economic Crime Sentencing Guideline, 82 Mo. L. Rev. (forthcoming 2017) (discussing critiques of the definition of loss adopted as part of 2011 Economic Crime Package and concluding that it remains fundamentally sound, even though some technical improvements might be possible).

  28. [28]. See Bowman, supra note 12, at 5–6 (making this point).

  29. [29]. Id.

  30. [30]. See, e.g., Felman, supra note 9; Michael Caruso, Fed. Pub. Def., S. Dist. Fla., Statement Before the United States Sentencing Commission Public Hearing on Economic Crime and Inflationary Adjustments 1–5 (Mar. 12, 2015),

  31. [31]. See generally Bowman, supra note 2.

  32. [32]. Bennett et al., supra note 1, at 960 fig.2.

  33. [33]. U.S. Sentencing Comm’n, Final Quarterly Data Report 32 fig.D (2006), (“Figure includes only cases with a primary sentencing guideline of USSG § 2B1.1. Additionally, cases with an amendment year prior to 2001 were excluded from this figure because prior to this time fraud cases were reported separately as USSG § 2F1.1. Cases with guideline minimums of life or probation were included in the guideline minimum average computations as 470 months and zero months, respectively. In turn, cases with sentences of 470 months or greater (including life) or probation were included in the sentence average computations as 470 months and zero months, respectively. In addition, the information presented in this table includes time of confinement as defined in USSG § 5C1.1. Guideline minimums account for applicable statutory mandatory penalties.”).

  34. [34]. Research by the Sentencing Commission’s staff has confirmed that the average amount of loss in fraud cases has indeed increased over the past few years. Compare U.S. Sentencing Comm’n, Overview of Federal Criminal Cases: Fiscal Year 2015, at 11 (2016),
    .pdf (reporting a median loss amount of $213,831), with U.S. Sentencing Comm’n, Overview of Federal Criminal Cases: Fiscal Year 2013, at 12 (2014),
    _Criminal_Cases.pdf (reporting a median loss amount of $195,649).

  35. [35]. Bennett et al., supra note 1, at 960.

  36. [36]. Id. at 972.

  37. [37]. Catharine Goodwin, The Case for a New Loss Table, 13 Fed. Sent’g Rep. 7, 8–9 (2000) (describing the joint effort of the Justice Department and the Criminal Law Committee of the Federal Judicial Conference to increase sentences for some classes of economic offenders); Letter from Sim Lake, U.S. Dist. Judge, Chair, Sentencing Guidelines Subcomm., Criminal Law Comm., U.S. Judicial Conference, to Diana E. Murphy, U.S. Dist. Judge, Chairman, U.S. Sentencing Comm’n (March 5, 2001), in 13 Fed. Sent’g Rep. 21 (2000) (expressing support for revision of loss table that would raise sentences for many economic offenders). I can attest to the influence of the judges on this point from personal experience during the six-year-long process that produced the 2001 Economic Crime Package, experience that included serving as academic advisor to the Sentencing Subcommittee of the Criminal Law Committee of the Federal Judicial Conference.

  38. [38]. The Guidelines rules in effect at the time of sentencing govern, except in the case of rules enacted after the date on which an offense was committed that would have the effect of increasing the guideline sentence. U.S. Sentencing Guidelines Manual § 1B1.11(b)(1) (U.S. Sentencing Comm’n 2016). In such a case, the Ex Post Facto Clause of the Constitution bars retroactive application of the sentencing-enhancing rule. U.S. Const. art. I, §§ 9–10. In consequence, only cases involving crimes committed on or after November 1, 2001 would unequivocally be subject to the new rules. Cases in which the crime occurred prior to November 1, 2001 would not be subject to the new rules if they would increase the guideline sentence. Given the time lag in investigating and prosecuting economic offenses to the sentencing stage, the effect of the 2001 rules would have phased in slowly over a period of years. The same is true of the 2003 post-Sarbanes–Oxley amendments.

  39. [39]. See United States v. Booker, 543 U.S. 220, 245 (2005).

  40. [40]. Prior to Booker, the only mechanism available to a judge who wished to impose a sentence outside of (i.e., either above or below) the calculated guideline range was a “departure,” which the judge was obliged to explain in terms of factors like the presence of substantial assistance to the government, § 5K1.1, or the presence of aggravating or mitigating factors of a type or degree not accounted for by the Guidelines themselves, §5K2.0. The Guidelines had, and continue to have, a great many rules restricting or discouraging departures. See generally U.S. Sentencing Guidelines Manual §§ 5K1.1–5K3.1 (U.S. Sentencing Comm’n 2004). And prior to Booker, appellate courts had both the power and, frequently, the disposition to vacate departures they thought inappropriate. See, e.g., Frank O. Bowman, III & Michael Heise, Quiet Rebellion? Explaining Nearly a Decade of Declining Federal Drug Sentences, 86 Iowa L. Rev. 1043,
    1114–16 (2001) (discussing the effect of Koon v. United States, 518 U.S. 81 (1996), on appellate review of district court departures). After Booker, the departure mechanisms remained in place, but are substantially devalued inasmuch as there are now virtually no meaningful restrictions on sentencing judges’ power to vary from the guideline range.

  41. [41]. Compare U.S. Sentencing Comm’n, 2002 Sourcebook of Federal Sentencing Statistics 53 tbl.26 (2003), (reporting 10,203 cases of “substantial assistance departure” and 9,865 cases of “other downward departure” out of 58,684 total cases) [hereinafter U.S. Sentencing Comm’n, 2002 Sourcebook], with U.S. Sentencing Comm’n, 2015 Sourcebook of Federal Sentencing Statistics tbl.26 (2016), (reporting 8,470 cases of “§5K1.1 substantial assistance,” 6,294 cases of “§5K3.1 early disposition,” 5,224 cases that were “other gov[ernmen]t sponsored,” 1,521 cases of “downward departure,” 593 cases of “downward departure w[ith] Booker,” 12,032 cases of “below range w[ith] Booker,” and 378 cases that “remain[ed] below range” out of 68,270 total cases) [hereinafter U.S. Sentencing Comm’n, 2015 Sourcebook]. These figures combine all sentences imposed below the applicable range, whether they resulted from a formal departure (on substantial assistance or other grounds) or from a post-Booker variance.

  42. [42]. U.S. Sentencing Comm’n, 2002 Sourcebook, supra note 41, at 59 tbl.29.

  43. [43]. U.S. Sentencing Comm’n, 2015 Sourcebook, supra note 41, at tbl.29 (reporting that 48% of all defendants sentenced with range in FY 2015 were sentenced to the guideline minimum and an additional 23.2% were sentenced within the lower half of the range). The notable drop from 2002 to 2015 in percentage of defendants sentenced to the absolute bottom of the range is probably accounted for by the large increase in defendants sentenced, without government motion, below the range altogether. In 2002, only 9.9% of fraud defendants were sentenced below-range without a government motion, while in 2015 the percentage jumped to 30.1%. Compare U.S. Sentencing Comm’n, 2002 Sourcebook, supra note 41, at 56 tbl.27, with U.S. Sentencing Comm’n, 2015 Sourcebook, supra note 41, at tbl.27A. A fair inference is that many defendants who in 2002 might have been sentenced at the guideline minimum due to the absence of a plausible ground for a downward “departure” are now being sentenced below-range based on Booker-authorized “variances.”

  44. [44]. U.S. Sentencing Comm’n, Final Quarterly Data Report: Fiscal Year 2010, at 34–37 figs.E–H (2011), (showing that the average length of sentence imposed compared to average guideline minimum for firearms, immigration, and drug offenses from 2005–2010);

  45. [45]. U.S. Sentencing Comm’n , 2002 Sourcebook, supra note 41, at 56 tbl.27 (reporting a “substantial assistance departure” in 17.8% of fraud cases and an “other downward departure” in 9.9% of fraud cases).

  46. [46]. Id. at 59 tbl.29.

  47. [47]. This figure is something of an approximation because it combines figures from two tables with different N values. But it is a fair estimate. The total number of fraud defendants for which we have departure data is reflected in Table 27: 6,378. Id. at 56 tbl.27. Of those, 17.8% or 1,135 got substantial assistance departures, and another 9.9% or 633 got other departures. Id. To figure out what percentage of those who did not get downward departures, but were sentenced in range or at the bottom of that range, one must turn to Table 29, which says that 54.8% of fraud defendants are sentenced at the guideline minimum. Id. at 59 tbl.29. The computational problem is that in composing Table 29, the Commission had to exclude quite a few defendants for whom they had insufficient data on in-range sentence position. Id. at 59 tbl.29 n.1. So the total number of fraud defendants considered in Tables 27 and 29 is different. Accordingly, I took the 54.8% figure from Table 29 and assumed that it reasonably approximated the percentage across all fraud defendants, and then multiplied 54.8% times the total number of fraud defendants reported in Table 27 to produce an estimate of 2487 defendants sentenced at the bottom of the range. I then added the 2,487 defendants sentenced at the bottom of the range to the 1,135 defendants with substantial assistance departures and the 633 defendants that got other departures to arrive at a total of 4,255 defendants that were sentenced at or below the guideline minimum.

  48. [48]. A substantial assistance motion is a motion by the government pursuant to 18 U.S.C. §3553(e) and/or U.S. Sentencing Guidelines Manual §5K1.1 (U.S. Sentencing Comm’n 2016), requesting that the defendant receive a sentence below the applicable guideline minimum or statutory minimum sentence based on his or her cooperation with the government in the investigation or prosecution of another person.

  49. [49]. U.S. Sentencing Comm’n, 2001 Sourcebook of Federal Sentencing Statistics 56 tbl.27 (2002), [hereinafter U.S. Sentencing Comm’n, 2001 Sourcebook].

  50. [50]. U.S. Sentencing Comm’n, 2002 Sourcebook, supra note 41, at 56 tbl.27.

  51. [51]. U.S. Sentencing Comm’n, 2015 Sourcebook, supra note 41, at tbl.27A.

  52. [52]. See supra note 38 and the accompanying text.

  53. [53]. Of course, this is not true in every case. Sometimes the parties will either agree that a sentence above the guideline minimum is appropriate, or will agree that the government is free to argue for an above-minimum sentence or even an upward departure above the range. However, in my experience as both prosecutor and defense counsel, and based on review of data and conversations with other practitioners, such cases are a distinct minority.

  54. [54]. The data in Figure 3 is drawn from the U.S. Sentencing Commission’s annually published Sourcebook of Federal Sentencing Statistics for the years 2000–2004, and found in Table 27 in each respective Sourcebook. U.S. Sentencing Comm’n, 2004 Sourcebook of Federal Sentencing Statistics tbl.27 (2005),
    /archive/sourcebook-2004; U.S. Sentencing Comm’n, 2003 Sourcebook of Federal Sentencing Statistics tbl.27 (2004),
    /archive/sourcebook-2003; U.S. Sentencing Comm’n, 2002 Sourcebook, supra note 41, at 56 tbl.27; U.S. Sentencing Comm’n, 2001 Sourcebook, supra note 49, at 56 tbl.27; U.S. Sentencing Comm’n, 2000 Sourcebook of Federal Sentencing Statistics 56 tbl.27 (2001), In 2004, the Commission reported data for the period before the Supreme Court’s June 24, 2004 decision in Blakely v. Washington, 542 U.S. 296 (2004), separately from the period following that decision. The 2004 data in Figure 3 is from the pre-Blakely period.

  55. [55]. The data in Figure 4 is drawn from the U.S. Sentencing Commission’s annually published Sourcebook of Federal Sentencing Statistics for the years 2000–2004, and found in Table 27 in each respective Sourcebook. See supra note 54. The 2004 data in Figure 4 is also from the pre-Blakely period.

  56. [56]. I recognize, of course, that some readers may find this sort of incremental approach insufficient. Given that more than 10 years have now elapsed since the Booker opinion rendered the Sentencing Guidelines advisory, it is not unreasonable to think that the Sentencing Commission should by now have wholly reconfigured the Guidelines to conform to the new constitutional reality. As I and others have observed, the Guidelines are now a bizarre and unwieldy anomaly—a highly complex, fact-driven set of rules designed to narrowly constrain judicial discretion that judges still must adhere to procedurally, but can now effectively ignore substantively. See generally Frank O. Bowman, III, Nothing Is Not Enough: Fix the Absurd Post-Booker Federal Sentencing System, 24 Fed. Sent’g. Rep. 356 (2012). The Sentencing Commission surely should redesign the entire Guidelines structure in light of Booker (and while doing so recalibrate sentencing levels that are normatively suspect). But I am skeptical that such a redesign is either within the institutional capabilities of the Commission or politically possible given today’s toxic congressional environment. Incremental improvements to the current structure seem more possible.

  57. [57]. Judge Bennett alludes to this concern, employing a revealing turn of phrase. He writes: “The statistical evidence on sentencing discretion, then, raises the question of whether, after three decades of sentencing reform, judges still empathize with economic criminals (as they were reported to do in the 1980s), or if the sentencing guidelines simply failed to approximate judges’ proper estimates of the harm done.” Bennett et al., supra note 1, at 964 (emphasis added) (footnote omitted). Note the confident assumption that the capacity for arriving at “proper estimates of the harm done” by economic crime rests with judges, rather than legislatures or other system actors. I certainly concur that the actual sentencing decisions of judges, particularly if considered collectively, are one important indicator of the appropriateness of sentencing levels, but I disagree with the implication that judges should be the sole arbiters of this question. If that were the case, the customary federal sentence for even very large frauds would still be probation, as it was in the pre-Guidelines era. Before the Guidelines, judges were demonstrably and unreasonably sympathetic to economic criminals who, not coincidentally, tended to be of the judges’ own race and class. Marvin E. Frankel, Sentencing Guidelines: A Need for Creative Collaboration, 101 Yale L.J. 2043, 2047 (1992). Their collective judgment was wrong then. The Guidelines provided a much-needed corrective. The question with which we wrestle now is whether, and if so to what extent, the Guidelines over-corrected for the judges’ previous under‑estimation of the harmfulness and blameworthiness of economic criminals.

  58. [58]. Justice Stephen Breyer who before his elevation to the Supreme Court served as one of the original members of the U.S. Sentencing Commission said of white-collar offenders: “A pre-Guidelines sentence imposed on these criminals would likely take the form of straight probationary sentences.” Stephen Breyer, The Federal Sentencing Guidelines and the Key Compromises Upon Which They Rest, 17 Hofstra L. Rev. 1, 7 n.49 (1988). He reported that the Commission concluded that the Guidelines’ objectives would be better served by the imposition of “short but certain terms of confinement for many white-collar offenders.” Id. at 20.

  59. [59]. Bennett et al., supra note 1, at 965.

  60. [60]. Id.

  61. [61]. Id. at 966.

  62. [62]. Id. at 966–67.

  63. [63]. See id. at 968–71.

  64. [64]. Id. at 968.

  65. [65]. Id. at 972.

  66. [66]. See supra notes 49–51 and accompanying text.

  67. [67]. U.S. Sentencing Comm’n, 2014 Sourcebook of Federal Sentencing Statistics tbl.27A (2015), (reporting that in FY 2014, 28.3% of fraud defendants were sentenced below the guideline minimum without a government motion).

  68. [68]. Even so, the meaning of the 75% low-end study result for federal district judges is muddied by the fact that, at least in my opinion, the range of 151 to 235 months represented in the study as being a fair approximation of the guideline range in a real case with these facts substantially overstates the likely guideline range. See Bennett et al., supra note 1, at 966, 985. In particular, the guidelines calculation employed by Judge Bennett and his colleagues includes a four-level increase for causing substantial financial hardship to five or more victims. Id. at
    990–94 app.A. However, there are no facts in the pre-sentence investigation report provided to the study participants that would support this enhancement. Id. at 995–1000 app.B. Absent those four levels, the guideline range for this defendant would be 97 to 121 months (about 8 to 10 years). Judge Bennett calculates the hypothetical defendant’s offense level as 34 and his Criminal History Category as I, producing a guideline range of 151 to 88 months. Id. at 991–92. If one subtracts the four offense levels for causing substantial financial hardship, the total offense level becomes 30 and the guideline range becomes 97 to 121 months. U.S. Sentencing Guidelines Manual § 5A (U.S. Sentencing Comm’n 2016) (sentencing table). Any district court judge who consulted the guidelines would realize that the study calculation was markedly too high for this defendant. Whether any judges performed this calculation and, if so, whether such a calculation affected such judges’ decision about how to sentence this case is unknown, but the implications cannot be disregarded.

  69. [69]. Bennett et al., supra note 1, at 968.

  70. [70]. Id. at 972.

  71. [71]. Forty-five percent of federal magistrate judges sentenced above the minimum of the stipulated sentencing range. Id. at 968. It is reasonable to assume that at least some significant fraction of those who sentenced at the minimum were not dissatisfied with that minimum as a normative matter. Hence, it is reasonable to conclude that a majority of the federal magistrate judges in the study were normatively satisfied with the stipulated range.

  72. [72]. Id.

  73. [73]. See id. at 964.

  74. [74]. See id.

  75. [75]. U.S. Sentencing Comm’n, Economic Crime Public Data Briefing 11 fig.6 (2015),

  76. [76]. A survey by the Association of Certified Fraud Examiners found that the median loss in frauds against privately held businesses in 2014 ranged between $160,000 and $231,000 between 2010–2014. Ass’n of Certified Fraud Exam’rs, Report to the Nations on Occupational Fraud and Abuse: 2014 Global Fraud Study 24 (2014), The survey states that 60.9% of these frauds were referred for criminal prosecution in 2014, but due to the global nature of the survey, it does not divide those that were referred between federal and state referrals. Id. at 64.

  77. [77]. In 2015, the salary of a federal district judge was $201,100. Judicial Compensation, U.S. Courts, (last visited Mar. 19, 2017). A federal magistrate judge may receive no more than 92% of the salary of a federal district judge. History of the Federal Judiciary, Fed. Jud. Ctr.,
    /page/js_6.html (last visited Mar. 19, 2017). By contrast, in 2015, the median salary of general-jurisdiction state-trial judges was $146,803. Knowledge & Info. Servs. Office, Nat’l Ctr. for State Courts, Survey of Judicial Salaries 1 (2016),

  78. [78]. See generally Russell Wheeler, Changing Backgrounds of U.S. District Judges: Likely Causes and Possible Implications, 93 Judicature 140 (2010) (describing the pre-appointment backgrounds of federal district judges).

  79. [79]. Bennett et al., supra note 1, at 963 fig.3A.

  80. [80]. Moreover, as I have argued elsewhere, the average guideline minimum sentences reflected in Figure 5 are quite probably the product of plea bargaining arrangements that produced far lower guideline ranges than honest application of the Guidelines’ rules would generate. See Bowman, supra note 7, at 274–75.

  81. [81]. Bowman, supra note 12, at 14; see Bowman, supra note 7, at 277–78.

  82. [82]. See, e.g., United States v. Adelson, 441 F. Supp. 2d 506, 511 (S.D.N.Y. 2006) (scoring defendant at offense level 46); United States v. Parris, 573 F. Supp. 2d 744, 747–48 (E.D.N.Y. 2008) (scoring defendant at offense level 42, with a guideline range of 30 years to life imprisonment); see also Bowman, supra note 5, at 168 (describing the effect of current guideline rules on sentences prescribed for corporate fraud); Bowman, note 7, at 273–75 (describing how parties to high-loss fraud cases appear to be reducing defendant guideline ranges through factor bargaining).

  83. [83]. Diana B. Henriques, Madoff Is Sentenced to 150 Years for Ponzi Scheme, N.Y. Times (June 29, 2009),

  84. [84]. See supra notes 47–52 and the accompanying text.

  85. [85]. U.S. Sentencing Guidelines Manual § 2B1.1 cmt. n.3(A)(i) (U.S. Sentencing Comm’n 2016).

  86. [86]. For an extended discussion of the loss definition and common criticisms of it, see generally Bowman, supra note 27.

  87. [87]. Id. at 5 (footnotes omitted).

  88. [88]. Bennett et al, supra note 1, at 983.

  89. [89]. Figure 6 reproduces the values in the 2015 loss table, see U.S. Sentencing Guidelines Manual § 2B1.1(b)(1) (U.S. Sentencing Comm’n 2015) (sentencing table). The modifications are recommended by the author, see infra text accompanying notes 92–94.

  90. [90]. Bennett et al, supra note 1, at 985–86.

  91. [91]. Bowman, supra note 7, at 278–79; see also Bennett et al., supra note 1, at 983.

  92. [92]. Bennett et al., supra note 1, at 983.

  93. [93]. U.S. Sentencing Guidelines Manual § 2B1.1(b)(1).

  94. [94]. Bowman, supra note 2, at 411–15.

  95. [95]. I did pick the maximum loss amount in the Model Table, $50 million, with reference to the fact that in FY 2012 there were at least 56 cases with a loss of more than $50 million. U.S. Sentencing Comm’n, supra note 75, 9 fig.4. By contrast, in 2012, there were only 12 cases in the top two steps of the then-existing loss table. Id. Even the top end of the loss table should apply to some appreciable and recurring number of cases. If not, the handful of cases falling into the top level(s) should be dealt with through upward variances. In addition, in this Model Table, $50 million loss amount associated with a 20 offense-level increase is lower than the $80 million required for a 20-level increase in 2000. This would allow a reversion to the pre-Economic Crime Package maximum of 20 offense levels for loss, while expanding the applicability of that 20-level maximum. This would be another way of reforming the table, while acknowledging
    Sarbanes–Oxley era congressional directives to increase sentences for high-loss offenders.

  96. [96]. Bennett et al., supra note 1, at 985.

  97. [97]. See Bowman, supra note 5 at 168–69; Bowman, supra note 7, at 271–73.

  98. [98]. See Bennett et al., supra note 1, at 985–86.

  99. [99]. Id. at 986 (quoting Kate Stith & José A. Cabranes, Fear of Judging: Sentencing Guidelines in the Federal Courts 69 (1998)).

  100. [100]. Space does not permit an exhaustive demonstration of this point in this Article, but the extensive discussion of the Commission’s 2015 amendments to section 2B1.1 before, during, and after its March 12, 2015, public hearing on those amendments illustrates the basic process. In January 2015, the Commission published proposed amendments to section 2B1.1, including alterations of several of its SOCs. Proposed Amendments to Sentencing Guidelines, 80 Fed. Reg. 2570, 2586–90 (proposed Jan. 16, 2015). Each proposed amendment was accompanied by an explanation of the problem the amendment was seeking to address and by questions on which the Commission sought input. Id. Thereafter, the Commission received extensive commentary on the proposed amendments and held a public hearing. See Public Hearing on Proposed Amendments to the Federal Sentencing Guidelines, U.S. Sent’g Commission (March 12, 2015), The Commission then made some changes to its original proposals and promulgated final amendments in April 2015, which were accompanied by the Commission’s explanations of its decisions. U.S. Sentencing Comm’n, Amendments to the Sentencing Guidelines 24–34 (2015), 20150430_RF_Amendments.pdf.

  101. [101]. See Bowman, supra note 3, at 81–82 (discussing the 2001 elimination of the former more-than-minimal planning SOC and the addition of the sophisticated means and multiple victim SOCs).

  102. [102]. Bennett et al., supra note 1, at 986.

  103. [103]. See Bureau of Justice Statistics, U.S. Dep’t of Justice, Recidivism of Prisoners Released in 30 States in 2005: Patterns from 2005 to 2010, at 8 tbl.8 (2014),

  104. [104]. For example, criminally negligent homicide is commonly a less serious offense than the most serious grades of robbery. Compare Mo. Rev. Stat. § 565.027 (2016) (stating that criminal negligent involuntary manslaughter is a Class E felony punishable by up to four years of imprisonment), with Mo. Rev. Stat. § 570.023 (2017) (stating that first degree robbery is a Class A felony punishable by 10–30 years of imprisonment or life imprisonment). The Missouri statute specifying sentence lengths for different classes of crime is Mo. Rev. Stat. § 558.011 (2016).

  105. [105]. U.S. Sentencing Guidelines Manual § 2B1.1(b)(7) (U.S. Sentencing Comm’n 2016).

  106. [106]. I suppose one could argue that the deterrent effect of an additional increment of guideline severity could be tested. The difficulty is that, even assuming that deterrent effect could be accurately measured, deterrence is not the only legitimate purpose of punishment. Even if no additional deterrent effect could be found, one could rationally justify an increase in punishment for health care fraud on the ground that stealing from programs designed to help the sick is particularly morally despicable.

  107. [107]. Bowman, supra note 7, at 272–73 (footnote omitted).

  108. [108]. Bennett et al., supra note 1, at 986.

  109. [109]. This is not to say that the question of which SOCs to keep, which to discard, and which to modify should be answered without careful thought informed by certain kinds of data. For example, the Commission should carefully consider data on questions like the degree to which a given SOC correlates with the presence of some other SOC such that imposing enhancements for both risks double punishment. However, we should foreswear the illusion that data is going to yield answers to questions that primarily concern values and priorities.

  110. [110]. U.S. Sentencing Comm’n, supra note 12, at 194–95, 207 (comments of Commissioner Rachel Barkow and Commissioner William H. Pryor, Jr.).

  111. [111]. Bennett et al., supra note 1, at 986 n.221.

  112. [112]. See U.S. Sentencing Comm’n, supra note 12, at 190–91 (comments of Commissioner Patti B. Saris expressing concern about effect of congressional directives on possible guidelines amendments).

  113. [113]. See generally Bowman, supra note 2, at 405–11 (discussing the congressional directives to the Sentencing Commission in the Sarbanes–Oxley Act of 2002).

  114. [114]. Bowman, supra note 7, at 279–80.

  115. [115]. Bennett et al., supra note 1, at 986–87.

  116. [116]. Bowman, supra note 7, at 276–77.

  117. [117]. Bennett et al., supra note 1, at 986–87.

  118. [118]. Id. at 982, 987.

  119. [119]. U.S. Sentencing Guidelines Manual § 2B1.1(b)(2)(C) (U.S. Sentencing Comm’n 2016).

  120. [120]. Bennett et al., supra note 1, at 988.

  121. [121]. Bowman, supra note 7, at 280.

  122. [122]. Bennett et al., supra note 1, at 988.

  123. [123]. See Bowman, supra note 3, at 74–75.

  124. [124]. See, e.g., Eric A. Tirschell, Vice-Chair, Practitioners Advisory Grp., Testimony Before the United States Sentencing Commission 1 (March 12, 2015), (urging the Sentencing Commission to devalue loss as a measure of offense seriousness and to increase the importance of factors such as “the extent to which the offender personally profited or intended to gain from the crime”).

  125. [125]. Bowman, supra note 27, at 8–11.

  126. [126]. Id. at 12–13. Because actual loss is defined by section 2B1.1 n.3(A)(i) as “foreseeable pecuniary harm,” it also contains elements of risk analysis. Compare U.S. Sentencing Guidelines Manual § 2B1.1 cmt. n.3(A)(i), with Bowman, supra note 27, at 13.

  127. [127]. Bowman, supra note 27, at 18–19.

  128. [128]. U.S. Sentencing Guidelines Manual § 2B1.1 cmt. n.3(B) (U.S. Sentencing Comm’n 2016) (“The court shall use gain that resulted from the offense as an alternative measure of loss only if there is a loss but it reasonably cannot be determined.”).

  129. [129]. U.S. Sentencing Guidelines Manual § 2B1.1 cmt. n.3(A)(i).


Floyd R. Gibson Missouri Endowed Professor of Law, University of Missouri School of Law.