Sentencing Authority and the Separation of Powers: "Who Decides Sentences at the Front End?"

Testimony Courts

Sentencing Authority and the Separation of Powers: "Who Decides Sentences at the Front End?"

November 12, 2003 24 min read
Paul Rosenzweig
Paul Rosenzweig
Visiting Fellow, The Heritage Foundation
Visiting Fellow, The Heritage Foundation

Testimony before the American Bar Association in the Justice Kennedy Commission on November 12, 2003

Good morning Mr. Chairman and Members of the Committee. Thank you for the opportunity to testify before you today on the topic the proper allocation of sentencing authority amongst the three coordinate branches of government.

I am a Senior Legal Research Fellow in the Center for Legal and Judicial Studies at The Heritage Foundation, a nonpartisan research and educational organization. I am also an Adjunct Professor of Law at George Mason University where I teach Criminal Procedure and an advanced seminar on White Collar and Corporate Crime. I am a graduate of the University of Chicago Law School and a former law clerk to Judge Anderson of the U.S. Court of Appeals for the Eleventh Circuit. For much of the first 15 years of my career I served as a prosecutor in the Department of Justice and elsewhere, prosecuting white-collar offenses. During the two years immediately prior to joining The Heritage Foundation, I was in private practice representing principally white-collar criminal defendants.

The issue addressed by this particular panel -the proper allocation of sentencing authority amongst the three branches of government -is, of course, principally a normative one. As such, answers will vary depending especially upon one's conception of the proper role of each branch; one's vision of the American Founding; and one's sense of the "best"policy to be followed. This debate is a good and useful one and I will offer some thoughts on the normative question in a moment -generally suggesting that legislative primacy is both the historically correct and politically appropriate answer. But before speaking to the "what should be"question, I thought I'd offer some descriptive information on "what is."

Sentencing - Some Empirical Data:  Let me start by addressing a seemingly simple issue -what controls sentencing at the federal level more, Congressional acts, the Sentencing Guidelines, or judicial discretion? As with so many things statistics are susceptible of varying interpretations, but the ones below strongly support the instinct that, I'm sure, animated Justice Kennedy's recent speech: Most of the determination is done by Congressional action and the Sentencing Commission -much less by the courts. The data presented below updates a short study I presented to the United States Senate Subcommittee on Crime in 2002. I present the statistics first and then provide some rough interpretations.

In Fiscal Year 2001, the most recent year for which we have statistics, according to the United States Sentencing Commission, federal courts entered convictions for 58,897 individuals. For those where detailed data is available (58,589 individuals), an overwhelming percentage of those who were sentenced for traditional crimes received sentences requiring terms of imprisonment.[1] For example, 93.8% of those convicted of drug trafficking were sentenced to prison. 97.5% of those convicted of robbery were imprisoned, as were (a bit surprisingly) only 85.1% of those convicted of arson, and (much less surprisingly) 97.7% of those convicted of murder. By contrast only 58.3% of those convicted of fraud and 34.9% of those convicted of embezzlement were sentenced to prison. And, using a blended rate, those convicted of technical regulatory offenses (that is white-collar crime involving malum prohibitum rather than malum in se crimes) were incarcerated only 30% of the time. I should add that these figures (and all the others presented in this section) are virtually identical to those I examined in Fiscal Year 2000 last year - in other words they have proven over the past two years to be stable.

At first blush it looks like a disparity in sentencing exists, which suggests the possibility that, for example, judicial discretion controls sentencing to a significant degree. If judicial discretion were controlling sentencing it would appear, one suspects, precisely where it seems to exist - in a lower sentencing rate for white-collar crimes than for common law offenses because of a presumed favoritism by the bench for more well-considered defendants. But if we look deeper into the statistics we see data that substantially challenge this perception.


In truth the data quoted are skewed because of the mandatory sentencing nature of many of our drug and other street crime statutes. If we change the question and ask, what percentage of those who are eligible under law for non-prison sentences wind up getting jail terms, we see a different picture. In other words, the data tell a different story if we examine sentencing rates but eliminate those cases where Congress has removed the discretion from the district court judge and look only at those cases where a district judge has a legal choice to make between incarceration and some non-jail alternative (community service, probation, home detention, or some other form of punishment not involving a jail term) available. According to the Sentencing Commission, the following were the national rates of incarceration for federal cases in which there were non-jail alternatives (some 10,137 individuals in Fiscal Year 2001, down from 11,137 in Fiscal Year 2000 - or roughly 1/6th of the entire total):


Crime Type Rate of Imprisonment (%)
Fraud 36.4
Larceny 19.2
Immigration 82.2
Embezzlement 35.7
Drugs - Trafficking 42.7
Drugs - Simple Possession 38.6
Firearms 46.1
Forgery/Counterfeiting 27.1
Other Miscellaneous Offenses 23.6


As you can see, if we exclude the immigration category (for which there are probably some exogenous explanations), when courts have discretion much of the disparity in sentencing rates disappears. White-collar frauds, for example, are incarcerated at rates greater than those for defendants who possess drugs or firearms. This uniformity in rates of incarceration suggests that what discretion courts retain (notably in only 1 in 6 cases) to choose a non-incarceration sentence is exercised with relative uniformity across the range of potential crimes. This consistency in the exercise of sentencing discretion suggests that courts, to the extent they have the discretion exercise it within a relatively constant bound. It also suggests (in a way, perhaps, that looking at Guideline departure rates does not), that most of the discretion about the most fundamental choice - whether to jail or not - is exercised at the definitional level by Congress.


The second prism through which to attempt to assess the question of sentencing discretion lies, of course, not in imprisonment rates but in the length of imprisonment. Here the mandatory nature of certain drug offenses again is reflected in the data:

Crime Type Mean Sentence (in months) Median Sentence(in months)
Robbery 95.7 70.0
Drugs -- Trafficking 72.7 51.0
Drugs - Possession 15.7 6.0
Manslaughter 34.1 21.0
Larceny 15.7 12.0
Fraud 18.7 14.0
Embezzlement 7.9 5.0
Bribery 19.0 12.0
Tax Offenses 16.6 12.0
Money Laundering 47.7 37.0
Environmental/Wildlife 13.7 10.0
Antitrust 8.4 6.0
Food & Drug 20.0 15.0

But this, of course, does not tell the whole story. As we have seen already in connection with incarceration rates, the courts are often constrained by statutory requirements. So too with the length of terms of imprisonment imposed. 

To review (though I am sure the Commission members are well aware of this): As a general rule, the length of a sentence is determined either by statute or, of course, by the operation of the sentencing guidelines. [The guidelines themselves are statutorily mandated, yet substantively developed through regulation; they are, thus, ultimately derived from statute]. It is useful therefore to ask whether the sentences reflected in the data are of the lengths they are because they are required to be that long by the sentencing guidelines or if they are the product of disparate departures from those guidelines by the courts - i.e. who is exercising the discretion in setting sentencing levels, Congress (though the Commission) or the courts? The answer is that the courts appear to depart from the Guidelines and exercise discretion at roughly equal rates for all cases thus suggesting that statute does a greater job of controlling sentence length than judicial discretion across the board. Consider the following data (which exclude departures for substantial assistance to the authorities):

Crime Type Rate of Departure (%)
Robbery 12.1
Drug Trafficking 17.9
Firearms 11.8
Larceny 7.3
Fraud 11.0
Embezzlement 14.6
Immigration 35.7
Other Miscellaneous 14.0

Once again, immigration offenses are unusual. Beyond that, the rates of departure from the guidelines are roughly consistent for all offenses and there is even some suggestion that serious offenses such as robbery and firearms are more likely to have judges depart from the guidelines than regulatory. Again, the drug trafficking offenses are a possible exception to the general rule.

There are several tentative conclusions that can be drawn from this data. First and foremost, whatever sentencing differences exist are principally the product of the actions of Congress. Median and mean sentences vary by type of crime, but insofar as we can tell, when offered a discretionary choice among offenders the courts do not impose incarceration in a unpredictable or biased manner. Rather, the rates of exercise of discretion are more or less uniform in the courts irrespective of offense type. Even drug trafficking offenders are, in the midst of the war on drugs, incarcerated less than 50% of the time when the courts are given the opportunity to choose whether to impose a sentence of imprisonment or not.


Moreover, the lengths of sentences flow almost exclusively and directly from either statutory requirements (mandatory minimums, and the like) or indirectly from statutes through the sentencing guidelines adopted by the U.S. Sentencing Commission. With the possible exception of drug trafficking charges there appears to be little difference, generally, in the way judges treat offenders before them. They get sentences less than what the guidelines would call for with the same approximate frequency.


Finally, insofar as the data are susceptible to analysis, other than serious personal offenses (such as robbery) and offenses relating to drug trafficking (including money laundering) most offenses are treated relatively similarly, with typical sentences falling in a fairly narrow range of from 1-2 years. Even manslaughter sentences do not vary appreciably from this seeming norm. One might almost suspect that we have reached a general consensus on the subject as a society and identified 1‑2 years as the appropriate just punishment for most criminal offenses.


This is not terribly surprising. Recall, if you will, how it is that the Sentencing Guidelines were initially developed. The Commission chose to take the tack of historical analysis, looking to past practice around the nation, and attempting to carry that historical practice forward into the guidelines, while evening out disparities between regions and districts. In doing this, the Commission collected data on more than 40,000 cases.


Interestingly, the one area where the Commission chose to depart from this historical base was in the area of economic or regulatory crime. There, the historical data reflected that "economic crime[s] [were punished] less severely than other apparently equivalent behavior." Consequently, the guidelines as initially proposed in 1987 and as in use today make an effort to upgrade the penalties for regulatory and economic, white-collar offenses. I think the success of that effort is reflected in the data presented. With the exception of drug offenses - a sui generis topic on which Congress has often legislated - we have reached a fairly consistent point of equilibrium.


The question then is whether that equilibrium is the right place to be. And, whether the other branches of government have anything to say about the placement of that equilibrium.

Normative Issues - As a theoretical matter, this seems a relatively easy question to answer. The courts have (more or less) determined that they have little or nothing to say about either legislative exercise of sentencing discretion or about prosecutorial exercise of discretion in charging (which, as we know, in an almost straight line effects sentencing issues). Outside of the capital sentence context there are few judicially imposed limits on legislative sentencing.[2] The "three strikes"cases from last Term demonstrate that, at least for now, neither Due Process nor the Eight Amendment constrains sentencing choices by Congress.[3] Similarly, except where the prosecutor appears to be exercising his discretionary choices in an invidious discriminatory manner in violation of the Fifth or Fourteenth Amendments, the courts have determined that they will not look behind the exercise of discretion or control it in any way.[4]

In effect, then, the courts have deliberately chosen a limited, almost self-abnegating role in constraining the use of criminal sanctions. As it stands today, no effective judicial constraint currently limits the extent to which individual conduct may be criminalized or punished. And, as an initial cut in the sentencing context this is the right answer, from a Constitutional perspective. [I will, below, suggest an alternative analysis that chooses a different approach to constraining legislative choice]. Ours is a government of separated powers - Congress may create the laws and the Executive is charged with applying them. So long as they do so within the limits of Constitutional restrictions, there is little role for the courts. We should be deeply skeptical of a Due Process rule that would, for example, hold that every defendant is entitled to individual consideration of his sentence - such a rule would be ahistorical and, ultimately, subversive of legislative authority to set sentences. Indeed, if anything, the trend seems to be in the opposite direction even within the judicial branch - towards less judicial authority and more responsibility for sentencing held by juries.[5] Similarly, the Eight Amendment's limitation on "cruel and unusual" punishment would, it seems, be substantially disfigured if it were stretched to require judicial sentencing discretion.

We should not be unaware of the consequences of this view. Indeed, the consequences of this are two-fold: a pathological legislative approach to criminal law and an excess of prosecutorial discretion.

As Professor William Stuntz has noted, American criminal law "covers far more conduct than any jurisdiction could possibly punish."[6] This wide span of American law is the product of institutional pressures that draw legislators to laws with broader liability rules and harsher sentences. The reason is the dynamic of legislative consideration: When a legislator is faced with a choice on how to draw a new criminal statute (either narrowly and potentially underinclusive or broadly and potentially overinclusive) or how to establish the level of a new penalty (either lenient or draconian), the politics of the situation naturally cause the legislator to be overinclusive and draconian. Few, if any, groups regularly lobby legislators regarding criminal law and those that do more commonly seek harsher penalties and more criminal laws, rather than less. The political dynamic is exacerbated by the consideration (usually implicit) of the costs associated with the criminal justice system. Broad and overlapping statutes with minimum obstacles to criminalization and harsh penalties are easier to administer and reduce the costs of the legal system. They induce guilty pleas and produce high conviction rates, minimizing the costs of the cumbersome jury system and producing outcomes popular with the public.

The final piece of the equation is legislative reliance on the existence of prosecutorial discretion. Broader and harsher statutes may produce bad outcomes that the public dislikes, but blame for those outcomes will lie with prosecutors who exercise their discretion poorly, not the legislators who passed the underlying statute. As a consequence, every incentive exists for criminal legislation to be as expansive as possible.

And in the absence of any judicial check on this legislative trend, the result is a wholesale transfer of power from elected legislative officials to prosecutors who, in many instances, are unelected and not responsible to the public. Where once the law had strict limits on the capacity of the government to criminalize conduct or , those limits have now evaporated. Society has come, instead to rely on the "conscience and circumspection in prosecuting officers."[7] Or, as the Supreme Court said in Dotterweich, Americans are obliged to rely only on "the good sense of prosecutors, the wise guidance of trial judges, and the ultimate judgment of juries" to determine criminal conduct.[8] In effect, the legislative branch has transferred a substantial fraction of its authority to regulate American social and economic conduct to those who have no expertise in the matter: prosecutors, trial judges, and jurors who make decisions on criminalizing conduct without any ability to consider the broader societal impacts of their decisions.

As you will understand, from my conservative perspective this dynamic warps the intended constitutional structure. Rather than condemning, therefore, the return of Congressional engagement in sentencing policy, we should welcome it. Though some in the public may be dismayed by the resurgence of Congressional application to sentencing issues that return affords them the perfect opportunity to respond if they so desire - by exercising the franchise at the ballot box. It is good when we experience a return to legislative accountability for the criminal law, not bad. 

And in that regard, the new "top charge" policy that Attorney General Ashcroft has announced may actually be a good thing.[9] It is, in effect, a requirement that federal prosecutors do less plea-bargaining - much less. Absent unusual circumstances, prosecutors are now supposed to only accept a plea to the "top" charge - the most serious charge possible under the facts of a case. No longer, for example, can a murder charge be pled out as a manslaughter crime.

In the short term the effects of this change will, no doubt, be dramatic. There will be fewer plea bargains and more trials. And, when combined with the federal sentencing guidelines and mandatory minimum sentencing rules there will, assuredly, be longer prison terms for criminals.

Some will condemn this trend. Others see a link between increased punishment and the historic reduction in crime rates that we have experienced in America over the past 20 years. I am quite sure this Committee will hear reams of data on this debate over the term of its existence.

But wherever one comes down in that debate, it should not obscure a far more important point: that the new Ashcroft charging policy may go a long way toward preventing Congress from avoiding accountability for its decisions. The lack of accountability is wrong - and it is wrong in a fundamentally important way. The elected legislator, not the un-elected prosecutor is, ultimately, the public's principal means of checking prosecutorial excess. Whether the public vision is that drug sentences are too harsh or too light, or that financial crimes are inadequately or too stringently punished, it is Congress, in the end, that defines what is a crime and how severely an offender must be treated.

The problem (as I've said) is that prosecutorial discretion breaks the feedback loop. Its existence without practical limit confuses the public as to who bears the ultimate authority for criminal law and it allows craven politicians to pander to public fear without any risk of responsibility for the consequences of their decisions.

And therein lies the promise of the new Ashcroft guidelines - though perhaps a promise the Attorney General did not intend. By severely limiting prosecutorial discretion, the new guidelines will lift the veil that protects legislators. As criminal sentences necessarily rise, the public will have the opportunity to more directly understand what the legislature has done. And when they ask "how can these sentences be so high?" answer now will be "because that's what Congress has commanded it," and no longer "because that's what the prosecutor chose." And, that is a good thing indeed.

A Paradox - Finally, I cannot let this opportunity pass without suggesting, at least gently, that we are, paradoxically, looking at the wrong thing. Sentencing is the "back end" of the process. The right place to look, from my perspective, is the "front end" - that is, at the creation of federal crimes in the first instance. "Who says A must say B" and once the broad federal power to criminalize in the first instance is granted, it seems to me that the discussion of who gets to set sentencing levels is a relatively trivial legal question. Thus, I would urge the Commission to broaden its lens and pick up on the questions left open by the ABA's 1998 study on the Federalization of Criminal Law. That study was a descriptive one - it is now time to ask the normative question.

Allow me to present a bit more data. Estimates of the current size of the body of federal criminal law vary. The American Bar Association reported in 1998 that there were in excess of 3,300 separate criminal offenses.[10] More than 40 percent of these laws have been enacted in just the past 30 years, as part of the growth of the regulatory state. And these laws are scattered in over 50 titles of the United States Code, encompassing roughly 27,000 pages. The statutory code sections often incorporate, by reference, the provisions and sanctions of administrative regulations promulgated by various regulatory agencies under congressional authorization. Estimates of how many such regulations exist are even less well settled, but the ABA thinks there are "[n]early 10,000."

Nor is the growth in the number of federal criminal statutes merely an academic question, without real world effects. To the contrary, between March 2001 and March 2002 (the latest year for which data are available), federal prosecutors commenced 62,957 cases, involving 83,809 individual defendants.[11] More than 3,100 of these defendants were charged with crimes categorized as violations of "federal statutes"-a category broadly (though not precisely) congruent with charges reflecting violations of a regulatory program. This number exceeds the number of federal prosecutions during the same year for a host of common law offense categories, including murder, robbery, embezzlement, forgery, and sex offenses. Put another way, more federal prosecutorial resources are invested in regulatory prosecutions than in the prosecution of forgery charges. All categories pale, however, in comparison to the principal area of federal effort-the prosecution of drug offenses, which resulted in more than 32,000 individuals being charged in 2002.

I would respectfully suggest that for those concerned with federal sentencing policy, the right place to look for limits is not in sentencing itself, but rather at the inception of the law - in the decision to criminalize. One is struck by the apparent paradox here. Those same jurists, for example, who are troubled by the legislative determination to enhance sentences with a "three strikes" policy seem to have no hesitancy in approving Congressional authority to criminalize. The traditional conservative view is the opposite - that the structural limitations found in the Constitution's Commerce Clause, for example, provide a much better means of constraining legislative excess than the more "floppy" Eighth Amendment limitations others see as vital. Thus, the right approach here is that taken in Lopez - a limitation on Congressional authority to criminalize directly rather than indirectly through a limitation on punishment options.[12]

And the trend toward increased criminalization has adverse consequences. The growth in federal criminal authority is a direct cause of the disappearance of the distinction between tort and crime in American law.[13] The use of the criminal law to address social goals enlists the criminal law as an agent of social regulation and change. Tort law has been, historically, a private mechanism for compensating for injuries. Affirmative civil enforcement by the government has been seen as a means of enforcing compliance with social norms through administrative procedures or civil litigation-the latter even having a component of punishment by virtue of the proliferation of punitive damages. These systems have been thought, in the past, to suffice in requiring economic actors to internalize the costs of their conduct and avoid imposing those same costs on unwitting external actors.

Now, however, the criminal law is being used in an avowedly instrumental capacity. Identically phrased statutes are often applicable to the same conduct-one authorizing a civil penalty and the other a criminal sanction.[14] In effect, the criminal law has become a tool of socialization, losing its historic character as a system for addressing wrongful conduct. Criminal sanctions for conduct affecting the public welfare have become a reflex answer. The result is a substitution of criminal law for more traditional tort and civil law: There is a "more pervasive use of the criminal sanction, a use that intrudes further into the mainstream of American life and into the everyday life of its citizens than has ever been attempted before."[15]

We should recognize that the courts cannot be the instrument of change in this limiting this trend. Rather, public attention combined with market reality are the more likely constraints on Congressional action. We live in a world of limited resources - one where, increasingly, federal attention will rightly be devoted to matters of national security. Thus, as the Judicial Conference of the United States put it in its Long Range Plan, criminal activity is appropriately the focus of federal concern only when federal interests are paramount. When federal resources are devoted to non-violent criminal conduct or regulatory offenses that are localized and with only an attenuated impact on interstate commerce those efforts are misdirected and contribute to an under-deterrence of real crimes through the diversion of resources to other areas.

I know that this last set of thoughts is, strictly speaking, outside the remit of this Committee. But I am constrained to say that if the Committee does not broaden its focus - if it looks only at sentencing issues without looking at criminalization in the first instance - it will be looking in the wrong place. For once we decide there are few limits on how Congress may declare an act criminal, it seems to me that it necessarily follows that there are few limits on how Congress may punish the acts it criminalizes.

Mr. Chairman, thank you for the opportunity to testify before the Committee. I look forward to answering any questions you might have.

The Heritage Foundation is a public policy, research, and educational organization recognized as exempt under section 501(c)(3) of the Internal Revenue Code. It is privately supported and receives no funds from any government at any level, nor does it perform any government or other contract work.

The Heritage Foundation is the most broadly supported think tank in the United States. During 2013, it had nearly 600,000 individual, foundation, and corporate supporters representing every state in the U.S. Its 2013 income came from the following sources:

Individuals 80%

Foundations 17%

Corporations 3%

The top five corporate givers provided The Heritage Foundation with 2% of its 2013 income. The Heritage Foundation’s books are audited annually by the national accounting firm of McGladrey, LLP.

Members of The Heritage Foundation staff testify as individuals discussing their own independent research. The views expressed are their own and do not reflect an institutional position for The Heritage Foundation or its board of trustees.



[1] All the data cited are from the US Sentencing Commission, Office of Policy Analysis, 2001 Datafile, OPAFY01, available at

[2] See Coker v. Georgia, 433 U.S. 584 (1977) (no capital punishment for rape); Enmund v. Florida, 458 U.S. 782 (1982) (no capital punishment if death does not result or is not attempted).

[3] See Lockyer v. Andrade, 538 U.S. 63 (2003); Ewing v. California, 538 U.S. 11 (2003).

[4] E.g. Wayte v. United States, 470 U.S. 578 (1985).

[5] E.g. Apprendi v. New Jersey, 530 U.S. 466 (2000); Ring v. Arizona, 536 U.S. 584 (2002); but see Harris v. United States, 536 U.S. 545 (2002) (allowing, in effect, continued use of sentencing guidelines and judicial fact-finding).

[6] Much of this analysis is derived from an analysis by Professor Stuntz to which I commend the Commission's attention. See William J. Stuntz, "The Pathological Politics of Criminal Law," 100 Mich. L. Rev. 505 (2001).

[7] Nash v. United States, 229 U.S. 373, 378 (1913).

[8] United States v. Dotterweich, 320 U.S. 277, 285 (1943).

[9] Memorandum to All Federal Prosecutors from Attorney General John Ashcroft, "Department Policy Concerning Charging Criminal Offenses, Disposition of Charges, and Sentencing" (Sept. 2003).

[10] American Bar Association , "The Federalization of Criminal Law" (Washington, DC; ABA, 1998), Appendix C. The data cited here are from this study and also from Ronald L. Gainer, "Federal Criminal Code Reform: Past and Future," 2 Buff. Crim. L. Rev. 46, 53 (1998).

[11] Administrative Office of United States Courts, "Federal Judicial Caseload Statistics," Table D-2, at caseload2002/contents.html (accessed April 3, 2003).

[12] United States v. Lopez, 514 U.S. 589 (1995).

[13] John C. Coffee, Jr. "Does 'Unlawful' Mean 'Criminal'?: Reflections on the Disappearing Tort/Crime Distinction in American Law," 71 B.U.L. Rev. 193 (1991).

[14] See, e.g., United States v. Ward, 448 U.S. 242, 249-51 (1980) (permitting imposition of civil penalty even though language of statute was virtually identical to longstanding criminal statute). Ward has been interpreted to mean that the legislature is free to choose to characterize misconduct as civil or criminal, thereby giving enforcement officials the option of choosing which sanction to impose. Examples of regulatory structures that allow the discretionary imposition of administrative, civil, and criminal sanctions for virtually identical conduct abound. Compare e.g., 33 U.S.C. § 1319(g) (authorizing administrative penalties for violations of Clean Water Act); id. §§ 1319(b), (d) (civil penalties); id. § 1319(c) (criminal penalties).

[15] Coffee, "Tort/Crime Distinction," supra, at 220.


Paul Rosenzweig
Paul Rosenzweig

Visiting Fellow, The Heritage Foundation