Reducing use of short-term imprisonment: the role of prior convictions and ”artbrotten”1

by professor ANDREWVON HIRSCH

1. Introduction
The present article examines, briefly, certain stategies for reducing reliance on short-term imprisonment (”korta fängelsestraff” or for short, ”KF”). I will assume that KF is more severe than its alternatives of conditional sentence or probation. The reason (which I need not elaborate here) is that imprisonment, even for shorter periods, ordinarily involves a greater restriction of the person’s liberty than, say, probation. Invoking KF thus involves a step-up in the onerousness of the sanction, even if not necessarily a large one. I will assume, also, that KF has significant adverse collateral consequences, most notably, the special social stigma of being imprisoned; and the disruptive effects on offenders’ everyday lives (e.g., most Swedes have only 5 weeks holidays, so that a two-month prison sentence would disrupt their employment).
    The emphasis in this article will be conceptual — particularly, on the rationale of using short prison sentences. Particular concern will given to questions of proportionality of sentence. In my analysis of such questions, I will be relying on the theory of proportionality that is sketched in my recent volume, Proportionalitet och straffbestämning (Uppsala: Iustus 2001).2 In speaking of ”short” imprisonment, I will be speaking of prison sentences of under one year — and particularly, of those under six months. Under present Swedish law, KF treatment is in principle permitted for cases in which the penal value is between two weeks and one year. Here, two major provisions govern the actual invocation of imprisonment: those relating to prior-convictions (återfall), and to ”artbrotten”. The former call for the use of conditional sentence or probation, instead of actual imprisonment, if the offender has no criminal record or has one that is not extensive. The second constitutes a partial exception to the first — and requires resort to short-term imprisonment for certain types of crimes, even upon the first conviction. Reducing the use of KF in the present Swedish sentencing structure would thus call for either of two steps: (1) altering the presumptions

1The author is grateful for comments provided by Karin Påle, Petter Asp, and Nils Jareborg.2Andrew von Hirsch, Proportionalitet och straffbestämning, Uppsala 2001.

regarding previous convictions, or (2) eliminating artbrott or restricting the scope of its application.

2. Alter the presumptions on prior convictions?
For intermediate-level crimes, the ”återfall” provisions ordinarily call for the use of conditional sentence or probation, but require imprisonment when the offender has accumulated a substantial record of previous convictions. This limited authority to invoke imprisonment appears, in my view, to be justified, for two reasons: (1) justice-based arguments about diminished ”tolerance” for repetition;3 and (2) practical arguments, to the effect that probation may already have been tried and failed in such cases. To what extent, then, would it be desirable to revise the ”återfall” presumptions, in order to restrict further the use of KF for repeat offenders, and how might this be done? Two possibilities exist here.
    One possibility would be to change the treatment of the less serious of those crimes which now receive ”sanctions” under Ch. 30 BrB. If these crimes are made eligible only for fines under Ch. 29, and no longer for ”sanctions” under Ch. 30, repeaters could not go to prison. But this approach would require the legislature’s altering the applicable penalty scales, which is not likely to occur; and it would also eliminate the use of conditional sentence or probation for such crimes. This does thus not seem an attractive option.
    An alternative, less ambitious change would be to alter the speed at which KF is invoked for repeat offenders. For some crimes such as burglary, conditional sentence or probation is ordinarily tried several times before a short prison term is invoked. For other crimes such as assaults, KF is invoked more quickly. Were these treated similarly to burglary — that is requiring several repetitions — KF’s use would be somewhat diminished. However, only a modest number of prison sentences appear likely to be involved. Changing the presumptions concerning previous offending therefore does not seem a very promising way of reducing reliance on KF.

3. Artbrott: eliminate, or reduce its scope?
A considerable number of first offenders or persons with only modest criminal records, who otherwise would have received a conditional sentence or probation, now go to prison because their crimes are of a type that is deemed to qualify for ”artbrott” (hereafter, ”AB”) treatment. Eliminating AB, or narrowing its scope of application, thus could substantially reduce the use of short prison sentences in Sweden. Should such a step be undertaken, and if so how?

3The ”tolerance” argument is developed in Andrew von Hirsch (note 2), ch. 7 and 9.8; and is discussed in Per Ole Träskman, ”Om återfall i brott”, SvJT 1999, 200 f.

3.1 Artbrott: can it be justified?
According to proportionalist sentencing theory, a just system’s primary determinant for deciding the degree of onerousness of a sanction should be the seriousness of the offence of conviction.4 It has been debated in the literature whether the presence or absence of previous convictions should be considered at all, but a number of proportionality theorists (including myself) have argued, as noted above, that limited reliance on the presence or absence of a record would be consistent with a proportionality rationale.5 AB imprisonment, however, is different and more problematic. It involves increasing the onerousness of the sanction for reasons not related at all to the defendant’s criminal choices — that is, neither related to the blameworthiness of his current crime nor to the extent of his earlier criminal involvement.
    A further objection to AB has been its arbitrary character, at present. AB’s use is now essentially uncontrolled, as no explicit standards govern its application. Imprisonment might thus eventually be imposed for a wide variety of offences. It has been suggested that criteria be developed, indicating when the invocation of AB status would be appropriate. Since such an effort is likely also limit the scope of AB status, it would have the further advantage of clarifying limits on the use of short prison terms. A number of suggested criteria have been put forward, including one scheme proposed by Karin Påle and myself in a 1999 article.6 Trying to devise criteria for regulating AB’s use would presuppose, however, that imposing AB status is at least sometimes defensible in principle as fair and workable. Påle and I already indicated at the end of our 1999 article that we had doubts that it is.7 Further reflection confirms the basic objections, namely: — AB status involves increasing the onerousness of the sanction, as just noted, for reasons unrelated to the blameworthiness of the offender’s current or past criminal choices. The increase aims, instead, at general prevention: it is concerned with the penalty’s supposed impact on the behaviour of other persons. This involves infringing ordinal proportionality;8 and using the offender as a means, in order to influence the conduct of others. AB imprisonment thus constitutes a derogation from justice. — A further objection may be raised to the invidious character of the factors likely to be relied upon for invoking AB treatment. Under

4See Andrew von Hirsch (note 2), ch. 4.5See s. 2 above, and citations in note 3.6Andrew von Hirsch and Karin Påle, ”Artbrott”, SvJT 1999, 241 ff. My present views are variance in certain respects with those there suggested — see, e.g., discussion accompanying note 9 below.7Ibid. 256–258.8”Ordinal proportionality” is the requirement that penalties be ordered consistently with the comparative seriousness of crimes. For discussion of this concept, see Andrew von Hirsch (note 2), 56–59.

Karin Påle’s and my proposed 1999 scheme,9 for example, that status would depend in part on the person’s social position: namely, whether his crime is of the ”Svensson” type committed by ordinary citizens who would have ”something more to lose” from being imprisoned. (The supposition would be that such citizens are more likely to be deterred by a threat of imprisonment, even for brief periods.) Reliance on such social-status factors to help decide the severity of the sanction raises serious questions of equity.
    Could artbrott treatment be defended on a ”modified” proportionality scheme?10 Under such an approach, limited deviations from the proportionate sentence would be permissible, where needed to achieve other legitimate penal ends, such as crime prevention. The difficulty here is that the factual premise for such a strategy of argument is open to challenge: there is no valid empirical basis for supposing that the invocation of AB status and the resulting imposition of imprisonment — with its resultant deviation from ordinal proportionality requirements — is likely to have significant added preventative effects. Deterrence studies — most recently, the 1999 Cambridge deterrence report11 — have consistently questioned whether any measurable marginal preventive effect can be gained from manipulating sentence levels generally. Even greater skepticism would be warranted, where the sentence increases are relatively limited in magnitude, as they would be here.
    How, then, should AB be viewed? A critical perspective would see AB as an historical remnant of an earlier way of thinking about sentencing. Prior to 1989, when the current law took effect, the stated statutory aims of the sentence,12 especially that of general prevention, served not so much as principles capable of assisting the choice of the penalty, than as aspirations — that it would be a desirable and hopedfor result were prevention achieved. The crucial change in the 1989 legislation was to provide principles fashioned actually to assist making the choice of sentence. Artbrott, however, has been a surviving remnant of the earlier way of thinking. In order to carry forward the process of modernisation of the sentencing provisions begun in 1989, as well as in order to reduce reliance on KF imprisonment, the AB provisions thus should be repealed.

3.2 What if there were a general-preventive effect? — Reducing artbrott’s scope
Part of the case against AB, just sketched, involves the assertion that there exists insufficient evidence of a general preventive effect in in-

9See Andrew von Hirsch and Karin Påle (note 6), 248–249.10Such a model is sketched in Andrew von Hirsch (note 2), appendix 2.11See, the Cambridge deterrence report, Andrew von Hirsch, A.E. Bottoms, E. Burney and P.O. Wikström, Criminal Deterrence and Sentence Severity, Oxford 1999. See also, Daniel Nagin, ”Criminal deterrence research at the outset of the twenty-first century”, in M. Tonry (ed.), Crime and Justice: A Review of Research, vol. 23, Chicago 1998, 51–91.12Former 1:7 BrB.

voking KF. The question might well be asked: what if there were such evidence? On a straightforward proportionalist sentencing model, AB would continue to be unacceptable, for the reasons stated above (s. 3.1) — namely, that AB treatment constitutes a deviation from ordinal proportionality. Sustaining AB would require, instead, the adoption of a ”modified” proportionality model, just discussed,13 which allows limited deviations from the deserved sentence. The question would remain, however, whether even such a model would necessitate the extensiveness of AB treatment that exists today. A case could be made that the scope of AB treatment should be narrowed considerably.
    A modified proportionality model introduces, to a certain extent, a second way of thinking about the determination of sanctions. The scaling of penalties relative to each other would continue to be decided mainly on penal-value considerations — ones that consider the forseeable harmful consequences of a standard actor’s behaviour,14 and the degree of culpability involved in that behaviour. The ”modification” introduced in the model, however, would be that limited deviations from such proportionate sentences be allowed on the basis of a different conception: a utilitarian theory of general deterrence. This theory focuses on aggregate harmful consequences, and gives less weight, if any, to the offender’s degree of culpability. These differences of perspective should be borne in mind when utilising a modified proportionality model and applying it to AB determinations. Supposing we were considering the appropriateness of applying AB treatment to what has been considered its paradigm case: aggravated drink-driving (grovt rattfylleri). To determine where this crime fits in the general penalty structure, we should be concerned with proportionality, and thus consider the penal value. This type of crime, rightly, has been assigned an intermediate penal value, for two kinds of reasons. First, the risk of the typical individual actor’s having a serious accident is not great: we are thus not dealing with high risks of serious harm (and in the eventuality that the actor is actually inebriated and thus constitutes a more serious danger, or in the event that serious harm does actually occur, the actor may be charged with other, higher-ranking crimes). Second, the actor acts negligently at most, rather than intentionally, with respect to the feared personal injury.
    Once we we have decided on the general location of the offence in the penalty structure, however, the perspective under a modified proportionality model would shift to one of deterrence. Let us see

13See discussion at note 10 above.14Concerns about culpability dictate that the harm, for purposes of proportionality requirements, should be the forseeable injurious consequences of a single standard actor’s conduct — because the actor is not responsible for the injury done by other, independent actors, and should not be blamed for unforeseeable consequences of his own behaviour; see Andrew von Hirsch, Past or Future Crimes, New Brunswick, NJ 1985, 64–65.

how this would happen, under the (possibly counterfactual) supposition that imposing a short prison sentence for this type of offence could measurably enhance deterrent effects. The crucial difference is that, under a deterrence rationale, harm would be considered in aggregate. With this change of perspective, one could argue that the risks of drink-driving should be viewed differently and given greater weight.
    Even if there is a relatively small risk that any individual alchoholconsuming (but not inebriated) driver would hurt someone when he drives, the practice of drink-driving among a large population of drivers would be likely to lead to a significantly increased incidence of fatalities or serious accidents. And the culpability-related argument, that only the typical actor’s own behaviour should be considered, would no longer hold — because of deterrence theory’s reduced emphasis on culpability concerns. This might suggest, albeit far from conclusively, that there would be some utility in a modest sentence increase, in reducing the incidence of serious accidents in aggregate.
    Even if such an argument were accepted, however, it would not carry over readily to other applications of AB treatment. What is critical in the drink-driving situation is that the contingent aggregate risks involve extensive and serious harms. When those contingent aggregate risks are of less significant kind, the case for AB treatment weakens. Let us consider another kind of offence to which some courts have accorded AB treatment: burglary and ransacking of an apartment.15 Here, the magnitude of the collateral risks are significantly lower than with drink-driving: notwithstanding the invasion of privacy involved in such conduct, there is small risk of significant physical injury — or for that matter, of other types of serious harm on a substantial scale.16 The case for AB status becomes weaker still, where the conduct involves still smaller or more tenuous harms. An instance is grafitti crimes, which a district court in Västerås has considered to be an AB.17 What these various instances suggest is that the case for AB treatment weakens considerably, as the aggregate collateral risks of the conduct diminish. There would thus be a case for reducing AB’s scope, even on assumptions designed to be most favourable to it.

3.3. The optimal solution: eliminate artbrott
What would be the optimal solution to AB’s difficulties? It would, in my judgment, to eliminate the institution entirely. There are, as discussed above (s. 3.1) two major drawbacks of AB: (1) it constitutes a derogation from fairness, because of the departure from ordinal proportionality requirements;18 and (2) the supposed basis for AB treat-

15See Andrew von Hirsch and Karin Påle (note 6), 254–255. The Gothenburg Court of Appeal has treated this as an AB offence.16To this effect, see ibid.17See ibid., 254.18See discussion accompanying note 8 above.

ment — enhanced general deterrence — remains unsubstantiated, because of the absence of confirmatory evidence that increases in severity will reduce the incidence of the criminal behaviour.19 It is ethically problematic to impose sentences that are thus more onerous, on the basis of crime-preventive effects that merely might occur.
    To these basic reasons for abolition, two supplemental ones may be noted. First, as the essays in this volume indicate, there is growing need for taking significant steps toward reducing reliance on short terms of imprisonment. An important way of accomplishing this end — while preserving today’s basic sentencing structure — is to end the practice of invoking short prison terms on AB grounds. Second, before the present sentencing system was introduced in 1989, invoking ostensible preventive aims was the principal rationale for sentencing generally. The new legislation (adopted 1988) largely replaced that with a different way of thinking, based on notions of penal value. But because of the relative novelty of such ideas, it was thought wise to preserve the older approach within specified limits, to decide cases (such as drink-driving) where preventive thinking had been particularly prominent. Now, after more than a decade of use, the proportionalist approach has shown itself as a workable method of deciding sentences — and so there is much less remaining need to preserve remnants of the former manner of reasoning which characterises AB.
    Eliminating AB requires legislation: namely, the repeal of the relevant statutory provisions (30:4 BrB). It is far from certain, however, whether Sweden’s Parliament will be willing to consider such a potentially controversial step — involving, as it could, the expenditure of considerable legislative time on a proposal lying well outside the Government’s main legislative programmes. It is thus necessary to consider the possibility that repeal will not prove a realistic option. What alternative remedial steps might then be taken? We shall consider this, next.

3.4 An alternative approach: narrowing artbrott’s scope
If repeal of the AB provisions is not feasible, an alternative, less radical solution would be to narrow the scope of AB’s application. Such a reform could be devised within the AB provisions’ operating assumption, that deviations from the deserved sentence for generalpreventive ends might, at least in some instances, be appropriate to help prevent harm. Even on that assumption, however, there would be good reason for narrowing AB’s scope, given the derogation from proportionality that AB treatment necessarily involves, and given the continuing uncertainty about when and to what extent any supposed preventive effects could materialise. AB’s application should thus be restricted to situations when its use seems the most plausible.

19See discussion accompanying notes 10 and 11, above.

When might this be? A possibility has been suggested already, in s. 3.2 above. There, it was examined how AB treatment might fare if (1) a ”modified” proportionality model were adopted, (2) it were assumed that general-preventive effects might sometimes occur, and (3) AB’s use were assessed by the extent and gravity of the collateral risks of the conduct. Judged by that standard, it was suggested that a case might be made for continued AB treatment for drink-driving (grovt rattfylleri), as there the collateral aggregate risks involve substantial losses of life or serious injuries.20 The case for preserving AB treatment for another previously-cited example, burglary with ransacking would already be weaker, since those collateral risks would be less grave. The argument would be the weakest where the collateral risks were still less significant, as true in the previously-mentioned example of grafitti offences; and also for AB treatment for sales of small quantities of cannabis or other lesser drugs. This kind of analysis thus does provide some priorities for when AB might be retained or eliminated.

3.5 A ”reduction strategy” for artbrotten
If it is desired to restrict AB’s scope, for the reasons just outlined, how might this be accomplished? We might then develop a ”reduction strategy” involving a prioritised scheme for scaling back AB’s use.
    One important element in a proposed reduction strategy would be a moratorium on further expansion of AB categories. If limiting AB’s use is the end to be achieved, it plainly becomes counterproductive to continue to allow the courts to add significantly to the list of crimes that are treated in this fashion. Such a moratorium would have the further advantage of eliminating the ”what more?” question. Any proposed reform, that attempts to specify general criteria for when AB treatment should be permissible, carries with it the risk that the suggested criteria might (to a degree difficult to anticipate) legitimate extending AB status to additional categories of criminal behaviour. With this moratorium, it can safely be assumed that any stated priorities designed to narrow AB’s use will not backfire -— to legitimise expansion.
    The other major element of a reduction strategy would be a prioritised winnowing-out of the categories of criminal behaviour that now receive AB treatment. The basis of this winnowing-out has been suggested (ss. 3.2 and 3.4). AB might be retained, for the time being, for crimes (such as drink-driving) that involve extensive and serious collateral aggregate risks. But offences which involve significantly less grave collateral risks would come to be treated under the normally-applicable sentencing standards, and no longer considered AB.
    For retention of AB treatment, it would be desirable to require that the collateral risks of the conduct should be both extensive and seri-

20See s. 3.2 above.

ous.21 Extensiveness refers to how widespread the potential injurious consequences are likely to be; seriousness, to how substantially those risked consequences would impinge on the vital interests of the individuals affected. Drink-driving seems to satisfy both of those norms: a considerable number of persons may be badly hurt through the practice of drink-driving. With burglary-and-ransacking, the conduct is both less widespread and its potential consequences less devastating to victims. For crimes affecting state or collective interests, such as tax evasion, these norms would have to be couched somewhat differently, but to a similar purport.

3.6 Which institution should implement reduction of artbrott ? -- A role for the courts
If a narrowing of the scope of AB treatment is desired, which is the appropriate body to decide such matters? This is, in my view, something that the courts may legitimately undertake.
    Ordinarily, changes in sentencing policy are in the province of the legislature. If it were desired to alter the number or scope of aggravating circumstances — say, by narrowing the significance given to defendants’ planning of the criminal enterprise — such a change should be a matter of legislation. The reason is that the legislature has specified various categories of aggravation in 29:2 BrB, and has included planning as an aggravating circumstance. The legislature is thus the appropriate agency for deciding whether a change in those specifications is desirable.
    The legislature, however, has chosen to handle AB differently. The statute does not define ”artbrott”, nor does it specify — even in the most general of terms — when AB treatment may properly be invoked. Instead it merely states, in 30:4 BrB, that ”the nature of the crime” is a ground for invoking imprisonment.
    The preparatory works for the statute go little beyond this: the 1987/88 Proposition introduces the idea of AB, and suggests this might be utilised for certain crimes of intermediate seriousness (such as drink-driving) which, under prior practice, regularly recieved short prison sentences.22 But the Proposition does not list what other types of crimes should be included, nor does it require that all crimes that previously were given short prison sentences should be included as AB’s. By proceeding in this fashion, the legislature in my view has effectively delegated to the courts the discretion to decide which kinds of crimes should be given AB treatment.
    The courts, in exercising this delegated discretion, also have not chosen to tie their own hands. While the courts have accorded AB status to a number of crime categories, such as drink-driving (and more

21Similar norms of extensiveness and seriousness were suggested in Andrew von Hirsch and Karin Påle (note 6), 248–250.22Prop. 1987/88: 120, 100.

substantial cases of tax evasion), which regularly received prison sentences under prior practice, they have not considered themselves bound to carry over all such practices. The courts have also not considered prior practice to be exhaustive of AB categories, for they have from time to time added new categories: indeed, I have mentioned instances already, concerning grafitti crimes and burglary and ransacking of an apartment.23 If the courts thus do not consider themselves bound by pre-1989 practice, and if they deem themselves authorised to expand AB coverage, the authority delegated to them by the legislature should be deemed to include narrowing that coverage. Although the courts thus should take the primary responsibility for reducing AB’s scope, such efforts would be aided by some degree of encouragement at the legislative level. Should the Government decide to propose new legislation on other aspects of short-term imprisonment, a general statement in the Proposition — stating that reducing AB’s use is a desirable goal and one that should be undertaken by the courts — would be most helpful, and probably would make the courts more willing to undertake the task.
    In order to assure a reasonable degree of uniformity regarding changes in the norms for AB, such changes are best undertaken by the appellate courts; and their directives would also carry the requisite authority. Whether this task should fall to the Supreme Court alone, or to that Court in conjunction with the six Courts of Appeal, is a matter better addressed by those who are more familiar than I with the workings of Sweden’s judicial system.

23See s. 3.3 above.