Sentencing Law and Policy

Just Sentencing # 7: The problem of disparate racial impact; concluding remarks

Richard Frase, guest-blogging on Just Sentencing: Principles and Procedures for a Workable System (Oxford 2012):

In my last post I examined one of the most difficult and important problems of punishment theory and practice: sentencing enhancements for repeat offending (prior and/or multiple current convictions).  In this final post, I’ll examine a second (but related) difficult and important problem: the disparate racial impact of seemingly race-neutral sentencing policies, and the resulting stark racial and ethnic disproportionality of prison and jail populations.  I address this topic in greater detail in the final chapter of my book.  

One of the most serious collateral consequences of punishment, and one that is so pervasive that I believe it requires recognition of a separate sentencing principle and accompanying procedures, is the tendency of severe criminal penalties to compound and worsen the plight of already disadvantaged minority groups, thus reinforcing and perpetuating social inequality.  Like my model’s desert-based limits, the social equality principle sets upper and occasionally lower limits on sanction severity, and may also affect decisions about the form of punishment.

The need to forthrightly confront this problem is clear.  Notwithstanding gains in civil rights and social status for many nonwhite Americans, there remains a core of multiply disadvantaged, predominately black citizens who seem condemned to an unending cycle of poverty, heightened risk of becoming an offender (and/or crime victim), incarceration, and release to worsened disadvantage followed by repeated crime and further (usually, enhanced) punishment.  The results of this repeating cycle can be seen in the substantial and persistent racial disproportionality of inmate populations.  Although some white Americans are victims of similar long-term, multidimensional disadvantage, the disadvantage suffered by nonwhites, especially blacks, is of a different order of magnitude — more profound, harder to escape, and more likely to persist across multiple generations.

A model sentencing system must protect disadvantaged minority citizens both as potential victims (providing an additional ground for minimum sanction severity) and as offenders (limiting maximum severity or the form of punishment, even if in the short run greater severity or a different form of punishment might help protect minority or other victims).  Unfortunately, recent American sentencing practices, combined with lack of metropolitan regional planning and inadequate social welfare programs, seem to have worsened social and racial inequalities. Such inequalities breed crime, and undermine respect for and cooperation with law enforcement.  Of course, the criminal justice system cannot solve deeply entrenched social inequalities, but it can, and must, avoid making them worse.

The inequality-perpetuating effects of punishment can be attacked on both the systemic and the individual-case levels.  At the systemic level, guidelines commissions are well-equipped to tackle these issues.  Such commissions have and should make use of their capacity to conduct racial-impact assessments, identify the racially disparate effects of particular sentencing policies, and propose guidelines rules and needed legislative changes to reduce those effects.  For example, guidelines and statutes could be devised or modified to give limited weight to the offender’s prior conviction record, in light of the strongly disparate impact of this factor.  It is particularly important to reduce or eliminate criminal history score components that have little if any validity as predictors of recidivism risk.  It may also be appropriate to give lesser or “capped” weight to certain types of prior record with especially strong disparate impact, for example: 1) traffic, drug, and public-order convictions that reflect disparities in law enforcement practices; and 2) a series of prior crimes (especially: drug possession and low-level sale) that reflects addiction, disadvantaged circumstances, or other factors that would cause the offender’s unadjusted “record” to overstate his culpability and (if his condition could be or has been treated) his recidivism risk. My model also recommends that sentencing commissions identify and make recommendations for reducing the disparate impacts of collateral consequences of conviction, since these further perpetuate social disadvantage, and cause more crime.

Addressing inequality issues is more problematic at the case level, given the difficulty of making reliable and consistent assessments on an individualized basis. But there may be room for courts to tailor at least the form and sometimes the severity of the sentence, to reflect cases where an executed prison term would seriously worsen the offender’s disadvantage, or where his or her deprived background and lack of good options to avoid criminal involvement strongly suggests diminished culpability.  Of course, some of these offenders do pose a higher risk, especially to other disadvantaged persons.  But total confinement in prison or jail is not the only way to manage that risk. Such offenders can be released under close supervision, GPS monitoring, halfway-house or other periodic community confinement, in- or outpatient treatment, and other intermediate sanctions that avoid the damaging effects of incarceration on already disadvantaged offenders.  Although resources to support these alternatives are limited in most jurisdictions, is it both fair and a better long-term crime-control strategy to give disadvantaged, high-risk offenders priority access to those resources.

But, a critic might ask — isn’t my approach modeled on the sentencing guidelines system in Minnesota?  And doesn’t Minnesota have one of the highest rates of racial disparity in its prison and jail populations? Chapter 5 of my book (building on an earlier Crime and Justice article), examines this issue in some detail. Although Minnesota’s inmate racial disparities are not as high as they once were, they are still above the national average for states.  But most of the disparity is present from the point of arrest, and seems to reflect racial differences in offending which in turn result from Minnesota’s unusually high rates of socio-economic racial disparity. When I traced disparity rates through later stages of criminal processing (focusing on black and non-Hispanic white offenders), I found a substantial increase at the point at which the sentencing guidelines are applied: recommended-prison rates are almost 50 percent more disparate than felony-conviction rates.  And well over half of this substantial jump-up in disparity results from the strong and racially disparate impact of offender criminal history in the sentencing of crimes of medium and low severity.  That strong disparate impact is not an inherent feature of guidelines sentencing; it could easily be changed -- and would be, under my model.  

I have not made as detailed a study of criminal history’s contribution to racial disparity in other jurisdictions.  But the available data, and nationwide averages, indicate very substantial racial differences in criminal history; the more weight a jurisdiction gives to this sentencing factor, the more disparate its sentences will be.

Mandatory-minimum sentence laws are also known to have disparate impacts, and the same is undoubtedly true for other seemingly race-neutral sentencing policies (the most notorious example being crack versus powder cocaine penalties).  We need to identify those policies, and make sure they are well-grounded and parsimonious. Otherwise, racial minorities and all of society will continue to incur substantial and unnecessarily high costs, both human and economic.  Chapter 5 of my book catalogues the wide variety of ways in which criminal conviction, and especially incarceration, compounds disadvantage, and thus also the risk of crime, for offenders, their families, and their communities.

This is my last post in this series.  Thanks again to Doug Berman for letting me guest-blog the past two weeks, and thanks to readers for your interest.  I'm sure that everyone will find some things in my model they disagree with, and other things they like.  In a way, that's the main point of the book: the need to find a hybrid approach that most people can live with, balancing competing sentencing purposes and procedural values within a structural framework that has proven workable in practice.  I hope you've been stimulated by what you’ve read, and that you'll consider taking a look at the book.

Just Sentencing # 6: The problem of sentencing enhancements for prior and multiple current convictions

Richard Frase, guest-blogging on Just Sentencing: Principles and Procedures for a Workable System (Oxford 2012):

In my first five guest-blog posts I discussed the need for a new hybrid sentencing theory, limiting retributive and other core principles underlying my model, the reasons why presumptive guidelines are the best choice among existing sentencing structures, and how my model incorporates that structure and other procedures.  My last two posts will examine two of the most difficult and important problems of punishment theory and practice, each of which is the subject a chapter in my book. 

Today’s post addresses sentencing enhancements based on the offender’s prior and/or multiple current convictions.  Although separate (and inconsistent) rules often apply to prior convictions and multiple current offenses, these can also be seen as simply two forms of repeated offending.  Repeat offending is generally deemed to justify greater sentencing severity to facilitate incapacitation, other risk management measures, and deterrence. Some writers have also offered retributive justifications for prior-record enhancements.  Existing laws and practices reflect these various theories: prior-record enhancements are found in virtually all modern sentencing systems (but with widely-varying formulas). And these enhancements apply frequently — the majority of offenders have at least some prior record.  As for multiple current offenses, most systems authorize consecutive or other enhanced sentencing. 

The rationales for and permissible degree of both types of repeat-offender enhancements are greatly in need of clarification and limitation.  Here is how the sentencing model in my book resolves these issues.

Prior convictions. Retributive theorists have failed to articulate convincing or even consistent rationales to justify and limit prior-record enhancements. Some retributive theorists reject all such enhancements; others argue that an offender’s blameworthiness steadily increases, the more serious his prior conviction record; a third group claims that offenders with little or no prior record are not fully blameworthy for their latest crime(s).  Utilitarian theories provide more consistent and plausible rationales; in particular, it seems reasonable to view the extent of an offender’s prior record as a useful proxy for his degree of recidivism risk.  But there has been almost no empirical validation of the effectiveness and efficiency of existing criminal history scores, and score components, as predictors of recidivism. Substantial risk-based enhancements also undercut the important retributive (and utilitarian) goal of making sentence severity proportional to current-offense severity.  

Chapter 4 of my book examines various rationales for prior record enhancements, concludes that heightened offender risk provides the most satisfactory rationale, and proposes a structure to ensure that such enhancements operate within offense-proportionality limits.  Specifically, I argue that guidelines offense severity levels (grid “rows”) should be substantially non-overlapping.  (These are typical-offense sentencing ranges; aggravating or mitigating circumstances of the current offense(s) permit departure and imposition of a higher or lower sentence.)  With this simple structural change,  offender prior record will have less weight, in determining recommended non-departure sentences, than under many existing guidelines systems. The strongly overlapped severity levels found in those systems mean that high-record offenders have recommended custody terms as much as thirty times longer than first offenders, and more severe than sentences recommended for low-record offenders who have committed much more serious crimes. Reducing the weight given to prior convictions will thus better achieve limiting-retributive and expressive (offense seriousness) purposes than under many existing guidelines systems. (However, almost all guidelines already partially accept the priority of offense seriousness: prior-record enhancements are “capped” — at some point additional priors have no further effect on recommended-sentence severity.)  Another important reason to limit the sentencing weight of prior convictions is that this factor tends to have a much greater impact on racial and ethnic-minority offenders, thus substantially increasing the racial disproportionality of inmate populations, compounding the post-release disadvantage of these offenders, and threatening social equality goals.

My book also recommends that guidelines systems develop empirically validated risk-assessment tools to supplement or replace criminal history scoring as the mechanism to determine recommended sentences within each offense-severity level. Such tools would only employ normatively acceptable factors (including some components of prior record) that are shown to add risk-predictive power without unacceptably high racial disparate impact. It seems likely that a major low-risk factor will be the offender’s age; criminological research has repeatedly shown that many offenders in their 40s and older are past their peak offending years or likely to soon desist entirely.  Yet these are often the offenders with the most extensive criminal history scores; almost all guidelines systems apply such scores without regard to offender age, thus locking up large numbers of aging, low-risk offenders (most of whom have been convicted of medium- or low-severity crimes).

Multiple current offenses.  Retributive and utilitarian theory is even more inadequate — indeed, almost completely silent — on the question of which multiple offenses should be sentenced consecutively; perhaps for that reason, most sentencing systems simply leave this matter up to the court’s discretion.

I argue that there should be a general presumption in favor of concurrent sentencing. The guidelines in Minnesota and several other states include a similar presumption, but with exceptions whose scope and rationales are often unclear. (I also argue in favor of several forms of “concurrent plus” sentencing found in several existing systems, in particular: counting multiple current offenses as prior convictions.)  When the presumption is overcome and consecutive sentences are imposed, there should be a further presumption that, except in very exceptional cases, the total prison term will not exceed twice the highest recommended (top-of-cell) term applicable to the most serious of the multiple crimes being sentenced.  A similar presumptive doubling limit is found in several state guidelines systems and in the revised Model Penal Code. The (now-advisory) federal guidelines formula is even stricter, limiting multiple-count sentencing enhancements to about 50 percent (above the most-serious-crime sentence).

The two presumptions set out above serve the goal of parsimony and lessen the risk that many defendants with multiple current offenses will receive sentences in excess of desert and valid utilitarian punishment needs. Although such needs and desert do justify consecutive sentencing in some cases, in other cases they do not.  For example, multiple current offenses can reflect short-term or situational factors such as a lost job that suggest lower culpability and dangerousness.  And in multiple-current-offense cases we can never be sure that the offender would have gone on to commit the later crime(s) if he had been caught, punished, and perhaps rehabilitated after the first one. The presumptions in favor of concurrent sentencing and against more-than-doubling give these offenders the benefit of the doubt, in the absence of contrary evidence justifying departure from one or both presumptions.

A further reason to limit the size of consecutive-sentence (and also prior-conviction) enhancements is that eligibility for such enhancements is subject to the vagaries of prosecutorial charging and plea-bargaining discretion — prosecutors can maximize, or minimize, the number of current charges, and thus also the offender’s future criminal history score.

That’s it for now.  My last guest-blog post will address another difficult and important problem of punishment theory and practice: racial disproportionality in prison and jail inmate populations.  I’ll also offer a few concluding remarks to wrap up this series of posts.

Prior posts in series so far:

• Overview
• Existing sentencing theories are inadequate
• A restated and expanded limiting retributive model
• Existing sentencing structures – Which are the best?
• My sentencing model's structural features and procedures

Just Sentencing # 5: My sentencing model's structural features and procedures

Richard Frase, guest-blogging on Just Sentencing: Principles and Procedures for a Workable System (Oxford 2012):

In my last guest-blog post I examined existing American sentencing structures and the need for a new model that builds on the best of them.  In this post I’ll summarize the structure and most important procedural features of the model proposed in my book.  That model was inspired by the sentencing guidelines systems in Minnesota, Washington, Oregon, Kansas, and North Carolina.  Most of the features of my model have been implemented and proven workable in those five states (and are also often found in other guidelines states).  My model also proposes modest improvements in these systems to better achieve the limiting-retributive and other principles identified in my earlier post (#3). 

State sentencing guidelines as the structural model. Existing limiting retributive theory needs further specification, in particular: how wide are the sentencing ranges? My model resolves this critical ambiguity by adopting a structure based on the presumptive guidelines found in the five states listed above. In these systems recommended sentencing ranges substantially constrain case-level discretion and its corresponding potential for disparity, while still leaving considerable room (within the range, or by regulated departure) for individualization of the sentence.

A permanent, independent, and adequately-funded sentencing commission. As explained in my previous post (#4), presumptive guidelines implemented and monitored by an independent sentencing commission improve the consistency and rationality of sentencing policy, ensure that such policy is informed by data on sentencing practices, and help to insulate policy from short-term political and media pressures.  Such commissions are especially valuable as a tool of resource management; they help the legislature prioritize the use of expensive correctional resources, prevent prison overcrowding, and avoid piecemeal legislation in response to the latest appeal to raise this penalty or that one.

Required resource- and demographic-impact assessments. Under my model (as in the five states) the resource management tasks noted above are greatly facilitated by assessments of the expected impact of existing and proposed policies on correctional resources.  Such projections are more accurate under presumptive, parole-abolition guidelines because sentencing in such a system is more predictable and the sentencing commission can collect the needed data and develop expertise in making projections.  Demographic-impact assessments likewise work better in such a system; they can be used to reduce, or at least not make worse, racially disparate impacts of sentencing policies.

Legally binding guidelines enforced by active but not overly intrusive appellate review.  Most of the recommended sentences under my model are presumed to be correct in the absence of unusual circumstances, and departures are subject to appellate review to correct errors and develop sentencing law. But the standard of review leaves trial courts with sufficient discretion to take into account particular case facts and tailor the sentence accordingly. In certain cases my model permits judges to vary from guidelines recommendations by stating reasons but without that variance being deemed a departure that would attract closer appellate scrutiny and require compliance with Blakely procedures.

Required statements of reasons for departures, and for some non-departures. A statement of reasons for departure or variance from guidelines recommendations is a feature of almost all guidelines systems, and serves to clarify the trial court’s analysis and facilitate appellate review.

Limited sentence enhancements for prior convictions and multiple current offenses. Repeat offending is deemed an aggravating factor in almost all sentencing systems, but the justification for such enhancements needs clarification and limitation. My model bases prior-record enhancements solely on assumed greater risk, not increased desert, and specifies that such enhancements must operate within substantially non-overlapping offense-severity levels (guidelines grid “rows”), thus increasing current-offense proportionality.  For multiple current offenses there is a presumption against consecutive sentencing and, in case of departure, a further presumption that the total should not exceed twice the maximum presumptive sentence for the most serious offense.  These limitations promote parsimony, reduce disparate impacts on non-white offenders, and lessen the impact of charging discretion.

Frequent use of suspended-execution sentences.  Suspended sentences to prison, jail, or a fine are frequently used in most sentencing systems.  My model encourages such suspensions because they are less costly and harmful to offenders and their families than an executed sentence; at the same time, they have useful expressive value (conveying the degree of seriousness of the offender’s crimes), give offenders a strong incentive to comply with required conditions, and leave substantial room for later tightening of sanctions in case of noncooperation or new evidence of offender risk.

Encouraged intermediate sanctions with maximum but no minimum severity limits.  Most offenders do not need to be sent to state prison, and many will be made more crime-prone if they are, yet risk management or expressive goals (re: offense seriousness) may require sanctions more restrictive than unsupervised probation.  Frequent, offense-proportionate use of community-based punishments is encouraged by: state-paid subsidies to local corrections programs for each offender in specified categories who is kept out of state prison; sanction equivalency scales to validate the punishment value of noncustodial sanctions; and presumptive limits on the aggregate severity of those sanctions.  But, consistent with parsimony, there should be no presumptive minimum-required severity of intermediate sanctions.  Such minimums are rarely found in state guidelines, would further complicate sentencing decisions, and are not likely to be regularly or consistently enforced.

Limits on sentence modifications via revocation or changed conditions of release. In many states revocations of probation and post-prison release account for a high proportion of prison admissions. Sentencing commissions should discourage revocations by developing rules that: 1) place upper limits on the severity of release conditions (making technical violations less likely); 2) encourage a graduated response to violations of conditions, not immediate revocation; 3) specify types of violations that do and don’t ordinarily merit custodial sanctions; and 4) set presumptive upper limits on the severity of such sanctions.

Abolition of parole-release discretion, with supervised release unrelated to good time. In the five presumptive-guidelines states listed earlier, defendants serve the entire prison sentence imposed by the judge, subject only to reductions for good conduct. Parole-release discretion should be rejected for several reasons: time-of-release decisions are often inconsistent and unreliable; parole discretion is not needed to control prison crowding (guidelines permit that to be done on the front end, see above); such discretion lacks transparency, violates victim needs for “truth in sentencing,” and permits administrators to exercise sentencing power instead of judges; and parole discretion makes it harder for legislatures and commissions to accurately predict future prison populations (to set priorities in prison use and avoid overcrowding).  Most offenders released from prison do need a period of parole-like supervised release, but its duration should be based on the severity of the conviction offense, not on the offender’s remaining prison term or earned good-time credits.

Inmate good-conduct credits that are moderate in amount, but reflect program participation. Parole-abolition states generally grant good-conduct credits to maintain prison order and give inmates an incentive to participate in prison programming. The 33 percent credit in my model falls about in the middle of the range of state credit formulas.  But offenders must in any case be granted initial release when they have served 85 percent, with the remaining sentence reserved to sanction release violations.

My last two posts will examine two of the most difficult and important problems of punishment theory and practice: sentencing of offenders with prior and/or multiple current convictions; and racial disproportionality in prison and jail inmate populations.

Prior posts in series so far:

• Overview
• Existing sentencing theories are inadequate
• A restated and expanded limiting retributive model
• Existing sentencing structures – Which are the best?

Just Sentencing # 4: Existing sentencing structures – Which are the best?

Richard Frase, guest-blogging on Just Sentencing: Principles and Procedures for a Workable System (Oxford 2012):

In my last two guest-blog posts I discussed existing sentencing theories, the need for a new (or restated) hybrid model (here), and the punishment principles embodied in my book’s sentencing reform model (here).  In this post I’ll examine existing sentencing structures and the need for a new model that builds on the best of them; the next post will summarize the specific procedural features of my proposed model.

Today American sentencing is fragmented; state and federal systems have adopted widely divergent structural models.  About half of the states retain traditional indeterminate sentencing systems emphasizing judicial and parole discretion; the remaining states and the federal system have shifted to some form of sentencing guidelines (with or without parole discretion) or a statutory determinate (“presumptive”) sentencing regime.  In addition, every American sentencing system, determinate or indeterminate, has at least some hyper-determinate, mandatory-sentence laws.

Guidelines and statutory determinate systems vary widely, and this diversity of approaches gives legislators a variety of options.  I argue, based on decades of experience in multiple jurisdictions, that the best option is the kind of presumptive guidelines system adopted in Minnesota and several other states.  This structural model has proven workable in practice, and has been endorsed by the American Bar Association and in the American Law Institute’s revised Model Penal Code sentencing provisions.  Presumptive guidelines also provide the best vehicle for implementing the limiting-retributive and other principles identified in my previous post (#3).  In the remainder of this post I will survey the major current types of sentencing structure in the U.S., and the advantages and disadvantages of each.

Indeterminate sentencing regimes.  In these systems, unless a mandatory-sentence law applies judges have discretion to sentence within a broad range – from unsupervised probation with minimal conditions, to an executed prison term of the statutory-maximum duration.  For offenders sent to prison, the parole board likewise has discretion to determine how much of the imposed prison term the offender will actually have to serve.  These broad discretionary powers have been justified by a belief that the primary purposes of sentencing are rehabilitation and protection of the public from higher-risk offenders.  But in the 1970s this model began to fall out of favor, for three reasons: (1) such offender-based sentencing ignores competing sentencing goals that require greater uniformity and proportionality to the offender’s crime (retribution, general deterrence, expressive goals); (2) it seems doubtful that highly-individualized assessments of offender risk, treatment needs, and treatment progress can be made accurately and consistently, particularly when based on in-prison behavior; and (3) such unfettered judicial and parole discretion is inconsistent with the rule of law — indeed, is there any other domain in which American officials make decisions of such immense consequence with so little guidance or restraint?  Other defects of indeterminate sentencing include: the difficulty of accurately predicting resource needs and setting front-end priorities in the use of prison resources; the dramatic differences between prison terms imposed and the much shorter terms actually served by most offenders; and the effective dominance of sentencing by administrative officials rather than judges.

Mandatory-sentence laws.  These laws go too far in the other direction, eliminating any discretion to mitigate punishment.  If fully enforced, such laws guarantee that some and perhaps many offenders subject to the law will receive punishment in excess of their desert, and/or in excess of what is needed and cost-effective to achieve adequate crime control.  Studies have repeatedly shown, however, that these laws are not fully enforced; often they serve primarily to coerce offenders to plead guilty in return for dismissal of the mandatory-penalty charge.  But the result is to turn effective sentencing power over to case-level prosecutors (while also burdening constitutionally-protected trial rights and excessively punishing the few offenders who go to trial).   If unjustified leniency by judges and parole boards is deemed to be a problem (despite the fact that most of these officials are elected or politically appointed), the answer is to apply legally-binding sentencing guidelines, requiring reasons for mitigated sentences and giving prosecutors the power to appeal.   Such presumptive-sentence laws avoid the injustices of mandatory-sentence laws, and strike a much better balance between rule and discretion, and between judicial, executive, and legislative power.

Statutory presumptive sentencing systems.  In the late 1970s California and several other states adopted a form of statutory guidelines, narrowing judicial discretion and, in most of these states, replacing broad parole-release discretion with good-time credits and a fixed period of post-prison supervision.  A half dozen states still have such systems in place.  But the fact that no new system of this type has been adopted for several decades casts considerable doubt on the viability of this reform option.  One of its most important defects is the lack of a sentencing commission to perform the functions further discussed below.

Advisory guidelines.  About a dozen states have advisory sentencing guidelines developed and monitored by a permanent sentencing commission. This approach has several advantages over purely statutory guidelines (or a system like that in the several states which disbanded their commissions but kept their guidelines in effect).  Like other specialized administrative agencies, such a commission can conduct research, develop and apply expertise, and take a comprehensive (all crimes), long-term, evidence-based view of complex sentencing issues.  It is easier to address such issues with a degree of insulation from short-term political pressures, although the legislature of course retains ultimate override authority.  But because advisory guidelines are not even presumptively binding, substantial sentencing disparities are likely to persist, and it remains difficult to accurately predict resource needs, set system-wide priorities, and control prison overcrowding.  These problems are made worse in the several states that retain parole release discretion.  In addition, the absence of even modest appellate review prevents courts from clarifying and developing sentencing law through the common law process.

Presumptive (legally binding) guidelines.  Five states — Minnesota, Washington, Oregon, Kansas, and North Carolina — have, for periods of 18 to 32 years, employed presumptive, parole-abolition sentencing guidelines systems that avoid the problems addressed above.  Prior to Booker, presumptive guidelines were also found in the federal system.  But as explained in my book, that system was not nearly as well designed; the current advisory regime is better (but also better than an indeterminate system).

That’s all for now.  My next post (#5) will show how key features found in these five state guidelines systems, along with a few new or improved procedures, are incorporated in my book’s proposed model.

Prior posts in series so far:

• Overview
• Existing sentencing theories are inadequate
• A restated and expanded limiting retributive model 

Just Sentencing # 3: A restated and expanded limiting retributive model

Richard Frase, guest-blogging on Just Sentencing: Principles and Procedures for a Workable System (Oxford 2012):

In my last guest-blog post I discussed existing sentencing theories and the need for a new (or restated) hybrid model.  In this post I’ll summarize the punishment purposes, limitations, and other core principles embodied in my book’s sentencing reform model. (The next two posts will examine procedural issues.)

As I explained last time, traditional punishment purposes often yield conflicting results when applied to particular cases.  For example, the goals of efficiency and public safety are well served when medium- and low-risk offenders are given reduced penalties.  Such reductions conserve scarce correctional resources (as well as court resources, if more of these offenders plead guilty), take advantage of community-based supports and treatment options, and reduce the risk of worsening these offenders’ life chances and thus making them more crime-prone.  Reduced custodial penalties also leave more room for community service and other restorative justice measures.  But when high–risk offenders who have committed the same crimes receive more severe punishments, the resulting disparities violate traditional retributive uniformity and proportionality principles; a low-risk offender’s reduced penalty also risks depreciating the seriousness of his offense, thus undercutting expressive punishment goals.

The solutions to conflicts like these that have been proposed by retributive theorists are inadequate.  To accommodate variations in offender risk and other non-retributive considerations, strict retributivists propose to merely vary the form of punishment, not its severity.  But loading up lower-risk offenders with onerous probation conditions is expensive and virtually guarantees high failure rates; such failures must then be sanctioned, resulting in more expense, and potentially imposing undeserved severity (since, by hypothesis, the onerous probation conditions have already given the offender his just deserts, and many probation violations are not blameworthy enough to justify much if any added punishment).

Limiting retributivism avoids these problems by positing a range of retributively-permissible penalties, within which risk management and other non-retributive goals may be pursued.  This is a better solution, and it has recently been adopted as the theory underlying the revised Model Penal Code sentencing provisions.  Norval Morris is the best-known proponent of this theory but, as I show in my book, many other writers, as well as most modern sentencing systems, have adopted some version of it.

The hybrid set of sentencing principles proposed in my book represents a restated and expanded version of Morris’s limiting retributive theory.  My model is designed to clarify critical ambiguities in Morris’s theory, incorporate additional important principles that he did not address, and make the restated version of his theory more consistent with the normative premises adopted or implicit in the best existing sentencing systems (thus encouraging the preservation and expansion of those systems).  The remainder of this post summarizes the core principles of my proposed model.

Definite but Asymmetric Desert Limits. Some writers, including Norval Morris, have argued that assessments of deserved punishment are inherently imprecise, even in relative terms (this offender versus that one).  Other writers assume that judgments of relative degrees of desert can be made, at least for typical cases of each crime, but argue that desert limits should be viewed as asymmetric, setting definite upper limits on sanction severity and more flexible lower limits — offenders may not be punished more than they deserve, but they can and often should be given less than their full deserts in order to conserve resources and better achieve other sentencing goals.  My book agrees with the latter view.  And as will be explained in a later post, when state guidelines systems have adopted a desert-based sentencing system the structure and especially the operation of these systems has been more consistent with a definite-asymmetric conception of desert limits.  On the issue of definiteness, I argue that we must try to state upper desert limits as precisely as possible because imprecise limits provide scant protection against pressure to escalate penalties.  Definite typical-case desert assessments also provide a better starting point for judges, and convey more effective expressive messages: “this is how seriously we view crimes of this kind.”  As for asymmetry, I argue that upper and lower desert limits raise different moral issues.  Punishment in excess of an offender’s desert is fundamentally unfair and an abuse of government power.  Below-desert penalties (in my view, but also supported by widespread practice) raise less compelling issues — of arguable unfairness to victims, law-abiding persons, other offenders, and this offender (the “right to be punished” argument).

Recognition of Expressive Goals, Symbolic Censure, and Conditional Hard Treatment.  Current limiting retributive theory gives insufficient weight to norm-reinforcement and other expressive punishment goals.  And most writers who have endorsed such goals claim that they require actual “hard treatment.”  But other writers recognize that suspended prison or jail sentences and other forms of conditional hard treatment can convey useful expressive messages about relative crime seriousness. Such measures also provide an efficient tool for risk management, and they are widely used in most modern systems.

Recognition of the Ends-Benefits Proportionality Principle along with Parsimony.  A very important component of Morris’s theory is the utilitarian and humanitarian principle of parsimony — measures should be no more severe than necessary to achieve their purposes.  But utilitarian philosophy also recognizes another important limiting principle, which I have termed ends-benefits proportionality — the direct and collateral public and private costs and burdens of a penal measures (or of a more severe measure) should not exceed its likely crime control and other practical benefits (or added benefits).

Social Equality as a Further Limiting Principle. One significant collateral cost of punishment is its tendency to reinforce and perpetuate disadvantage, especially for nonwhite offenders, their families, and communities.  These impacts are unfair, and they cause more crime.  A sentencing system cannot eliminate entrenched racial disparities in our society, but it also should not worsen them.  (I discuss the problem of disproportionate racial impact in chapter 5 of my book, and in the 7th post in this series.)

Within-Desert Adjustments Based on Offender Risk and Needs but Not on Deterrence.  Norval Morris was highly skeptical of offender-specific risk assessments, but endorsed case-level enhancements based on needs for greater deterrence. My model takes the opposite approach, viewing such deterrence enhancements as unreliable but encouraging the use, within desert limits, of validated risk assessments.

Retention of Substantial Sentencing Discretion and Control by Judges. The case-level application of sentencing laws should not be tightly controlled by the legislature, sentencing commission, prosecutors, or correctional authorities by means of mandatory-sentence laws, inflexible guidelines, or parole release discretion. Sentencing by judges is, and will be seen as, more impartial, more transparent, and more readily subject to the rule of law and appellate review.

Front-End, System-Wide Management of Correctional Resources.  Correctional resources must be carefully managed so as to set appropriate priorities, avoid overcrowding, and maintain program effectiveness.  These fundamental punishment-policy and resource-use decisions should be made on a comprehensive basis by the legislature or sentencing commission, and should not be left to correctional authorities on a case-by-case basis for the subset of offenders received by each corrections agency.

That's all for now.  In my next post (#4) I'll discuss the widely varying sentencing structures that currently exist in American state and federal jurisdictions, and the need for a new procedural model inspired by the best state sentencing guidelines systems.

Just Sentencing # 2: Existing sentencing theories are inadequate

Richard Frase, guest-blogging on Just Sentencing: Principles and Procedures for a Workable System (Oxford 2012):

In my first guest-blog post I provided a summary of my book’s main features and arguments.  Although most articles and books on sentencing are primarily either theoretical or practical, my book presents a fully-developed sentencing model that integrates theory and practice.  In this post and the next one I’ll discuss issues of principle (procedural issues will be addressed in posts four and five).  This post examines existing sentencing theories and the need for a new (or at least, restated) hybrid theory.  The next post will show how the model in my book meets that need.

Punishments require clear and convincing normative justification — by definition, they impose unpleasant and usually harmful consequences on offenders and their families, and they also consume scarce public resources.  But the most widely-accepted purposes of punishment often conflict in particular cases.  To further complicate matters, there are other important normative questions embedded in the choice of sentencing procedures, in particular: what official(s) or agency(ies) should make case-specific sentencing decisions?  And who should decide how to make the best use of limited available punishment resources?

The most widely-accepted purposes of punishment are crime control and retribution.  Punishments have the potential to achieve crime control through several mechanisms: rehabilitation of offenders, to address the causes of their offending; incapacitation of higher-risk offenders, usually by means of secure custody; specific and general deterrence of this and other would-be offenders, by instilling fear of punishment; and moral education to define and reinforce societal norms that restrain criminal behavior even when (as is often true) the chances of detection and punishment are slight — the sentence sends a message to the offender and the public that the punished behavior was wrong, and the severity of the sentence shows how wrong it was.

The other major traditional punishment goal, retribution (or “just deserts”), views punishment as justified because of its inherent value as a good thing in itself, regardless of whether it yields effective crime control or any other desired consequences.  Under the stronger versions of retributive theory offenders should be punished simply because they deserve it, the severity of their punishment should closely match their degree of blameworthiness (retributive proportionality), and equally blameworthy offenders should receive equally severe punishments (retributive uniformity).  A competing retributive theory — “limiting” retributivism — merely sets outer limits on punishment imposed to achieve other goals (especially crime control), thus producing a range of retributively permissible severity for any given case.  Finally, some desert-based theories justify punishment in terms of its expressive or communicative value; on this view, conveying deserved censure to offenders, and inviting an appropriate response from them, are viewed as good things for society to do whether or not any such response is obtained.

There are a number of other widely-recognized punishment theories, purposes, and limitations. Utilitarian philosophers have long argued that sentencing in proportion to crime seriousness deters offenders from committing a more serious crime, and helps to match punishment costs with crime-control benefits.  Sentencing uniformity also has practical benefits: it permits more accurate forecasts of future prison populations and other correctional resource needs.  And to the extent that the public subscribes to these values, making sentences more uniform and proportional improves the moral-education effects of penalties and maintains critically needed public respect and support for the criminal law and law enforcement.  Other practical goals of punishment include promoting satisfaction, closure, and compensation for crime victims and communities (restorative justice); reassuring the public that something is being done about crime; and facilitating the offender’s successful reintegration into society following his or her release from incarceration.  Finally, various administrative purposes and limitations must be taken into account in any theory of punishment, in particular: the need to encourage guilty pleas and other forms of offender cooperation; and the necessity to avoid prison and jail overcrowding and prioritize the use of these and other correctional resources.

Other sentencing goals are (like retribution) deemed to be good things in themselves.  These include the avoidance of disparities based on race or other clearly illegitimate criteria; the requirement that punishment respect norms of humane treatment and human dignity; and procedural fairness.  The lengthy list of punishment principles above virtually guarantees that, in many cases, the principles will yield conflicting results.  For example, if two equally blameworthy offenders commit the same crime but one poses a much higher risk of reoffending, putting the low-risk offender on probation and sending the high-risk offender to prison saves scarce correctional resources while effectively promoting public safety.  But doing so produces sentencing disparity from the perspectives of retributive proportionality and uniformity, risks depreciating the seriousness of the low-risk offender’s crime, and undercuts other practical values served by uniform and proportionate punishments.  Restorative justice measures provide another example: they may produce sanctions that, from retributive, crime-control, and/or efficiency standpoints, are either too severe (e.g., because of vengeful victim or community views) or not severe enough.

How can these various conflicts be resolved? Some retributive theorists argue that different forms of punishment (prison for high risk offenders; probation with onerous conditions for lower risk offenders) can still have equal punitive “bite,” thus maintaining proportionality.  But as I show in this book, such an approach poses many practical problems (greater cost; little reserved punishment power to handle violations of release conditions), and no jurisdiction has ever implemented such a system.

Limiting retributivists have generally resolved crime-control-versus-desert conflicts by positing a broad range of “not-undeserved” penalties based on the supposed inherent imprecision of desert judgments. But imprecise limits may result in no real limits at all.   And states with desert-based sentencing guidelines have not had difficulty reaching consensus about relative degrees of typical-offense desert, and about the factors that make a case atypical.  But these states, and indeed virtually all modern punishment systems, have in effect adopted an asymmetric model, giving greater emphasis to upper than to lower offense-proportionality limits.

In my next post (#3) I’ll discuss principled arguments in favor of an asymmetric model.  I’ll also discuss my hybrid model’s other sentencing principles, purposes, and limitations.

Just Sentencing #1: Overview

Hello everybody.  I’m Richard Frase and I’d like to thank Doug Berman for inviting me to blog about my new book, Just Sentencing: Principles and Procedures for a Workable System, published last week by Oxford University Press and available here.

Of course, I can’t fully convey the whole book in a few blog posts, especially a book with the broad scope describe below. All I can do is provide a summary of what I think are the book’s main features and arguments. Whether you agree or disagree with what I say in these summaries, I hope you’ll consider the fuller statements contained in the book.

My book presents a hybrid sentencing model that combines clearly-stated normative principles with procedures that have proven successful in practice in the most fully-developed state sentencing guidelines systems.

The model’s theoretical structure is an expanded version of limiting retributivism that sets desert-based limits on sentence severity, within which crime control and other non-retributive purposes and limitations of punishment are applied.  The latter include: expressive sentencing goals; parsimony (least restrictive alternative); utilitarian proportionality; social (and especially, racial) equality; retention of substantial judicial sentencing discretion; and front-end, system-wide management of correctional resources.  The model provides a role for all traditional and emerging purposes and limitations of punishment, and resolves the conflicts that often arise when those principles are applied to specific cases.  I argue that a hybrid sentencing theory is normatively superior and practically necessary.  Any purely retributive or purely crime-control model would fail to recognize widely-shared competing values, and would not succeed in practice.  Indeed, all modern sentencing systems with which I am familiar are hybrids of one kind or another, combining retributive principles with crime-control goals.

A sentencing theory, no matter how well it resolves important normative concerns, is of little utility without an accompanying set of workable procedures designed to implement the chosen sentencing principles.  Indeed, pure theory is incomplete even on normative grounds; concrete structures and decision rules help to illustrate and clarify theoretical concepts and the normative choices being made.

But sentencing procedures must likewise achieve an acceptable balance, especially between two competing procedural ideals — rule versus discretion — each of which has important advantages. Rules promote consistency and predictability; discretion promotes flexibility and efficiency. Procedures must also strike a workable balance in the use of custodial and non-custodial sentencing options, and in the powers of systemic policy makers (the legislature and sentencing commission) and case-level decision makers (judges, attorneys, and correctional officials).

Although sentencing guidelines are often seen as reflecting strong preferences for rules over discretion, and for system-wide over case-level policy-making, that is not how the best state guidelines systems actually work.  Like the proposed model, these systems structure sentencing discretion but leave judges and other officials with a substantial degree of discretion to tailor the form and severity of sanctions to the facts of particular cases.  My book shows how the core principles and procedures of the proposed model have been implemented in several states, and endorsed in model sentencing codes and standards.

The proposed model fills the need for a workable new sentencing rationale and structure to replace the once-monolithic indeterminate model which is now widely discredited because of its reliance on unfettered judicial and parole discretion.  My book also forthrightly addresses and resolves some of the most difficult challenges faced by American sentencing policy, including: overuse of prison sentences; racial disparities in prison and jail inmate populations; and the justification for and degree of sentence enhancements for repeat offenders.

The principles and procedures of any viable new model will, of necessity, be more complex than the indeterminate model that dominated American sentencing for most of the 20th Century.  The new model must harmonize and provide suitable roles for multiple, often-conflicting principles (indeterminate sentencing only needed to harmonize rehabilitation and public protection).  And the new model’s procedures must strike a better balance between rule and discretion (indeterminate sentencing’s procedures were simple but very unbalanced — all discretion, no rules).

But the model described above is not actually that hard to design. As I show in this book, my “new” model already exists to a great extent in several state systems, some of which have been in operation for decades.  My book gives particular emphasis to and draws its greatest inspiration from the guidelines systems in Minnesota, Washington, Oregon, Kansas, and North Carolina.  By the mid-1990s it seemed likely that the sentencing reform model adopted in these states would continue to spread, but then sentencing reform lost momentum and direction.  I argue that the virtues of this model remain, and that the perceived barriers to its broader adoption have disappeared or were always exaggerated.

Aside from the fragmented and often incoherent state of American sentencing, there are further reasons to return to the project of sentencing reform.  The massive recent increases in the size and cost of American prison populations, and their persistently high rates of racial disproportionality, provide reasons enough to seriously reexamine the sentencing principles and procedures that drove or at least allowed such increases and disparities.

I argue that states that have adopted my model should recognize its virtues and seek to maintain and improve their systems.  States that have not yet tried this model (or tried and then abandoned an incomplete version) should give this approach serious consideration.  The benefits of the model can be quite substantial, given the high individual and societal costs imposed by contemporary U.S. sentencing systems, and the modest expense of even a well-funded guidelines reform.  Beyond material costs and benefits, the proposed model can help states achieve what may be the most important product of a sentencing system — just punishments that are perceived as just by offenders, victims, and the public. The model can achieve just sentencing because it embodies clearly articulated and widely shared principles, combined with proven procedures well-suited to implementing those principles.

That's all for now.  In my next post I'll discuss existing sentencing theories, and the need for a new (or restated) hybrid model.

Professor Frase guest-blogging on "Just Sentencing"

Especially because I so greatly enjoyed Professor Stephanos Bibas earlier this year guest-blogging on "The Machinery of Criminal Justice" (series here), I am very pleased to be able to now welcome Professor Richard Frase as a guest-blogger to discuss sentencing issues raised by his terrific new book titled "Just Sentencing: Principles and Procedures for a Workable System," which was just published by Oxford University Press and is available here.

Richard's first guest post will appear later today, and everyone can get excited about the series based on this summary of the book's coverage via this page at the Robina Institute of Criminal Law and Criminal Justice:

What are the most important purposes of punishment, in general and in particular cases?  What makes just sentencing? These eternal questions are very difficult to answer because traditional as well as emerging sentencing purposes often conflict.  Retributive and non-retributive institutions and intuitions of justice are both deeply-rooted and each equally hard to ignore.  There is no generally accepted or well-elaborated theory to guide and evaluate recent or proposed sentencing changes, and most of the major books on sentencing theory are outdated.  There is a compelling need for a new sentencing model.

In Just Sentencing, Richard Frase describes and defends a hybrid sentencing model that integrates theory and practice–blending and balancing both the competing principles of retribution and rehabilitation and the procedural concern of weighing rules against discretion.  Frase lays out a sentencing reform model based on the theory of limiting retributivism.  The theory accommodates retributive values–especially the human-rights-based need to limit maximum sanction severity–along with crime-control goals such as deterrence, incapacitation, rehabilitation, and moral education.  It also promotes efficiency and provides sufficient flexibility to incorporate victim and community participation, local values and resource limitations, and restorative justice programs. Frase presents his significantly expanded version of the limiting retributive model and distinguishes it from versions proposed by others. Next, he demonstrates the practical feasibility and widespread support for this approach by showing how it has been successfully implemented in Minnesota, while also identifying the less developed limiting retributive elements found in almost all Western countries.  The final part of the book identifies and attempts to resolve the model’s most important theoretical and practical challenges, and suggest further improvements.

Just Sentencing is the first book in over forty years to present a fully developed punishment theory which incorporates both utilitarian and retributive sentencing purposes.