Incapacitation: Penal Confinement and the Restraint of Crime.
Incapacitation lies at the core of the emotionally and ideologically charged debate about crime and how to prevent it. But the most vocal participants in this debate often lack factual support for their competing claims. Franklin Zimring and Gordon Hawkins call for scholarly attention and empirical research to unmask the tenuous theoretical and empirical foundations of incapacitation as the dominant justification for imprisonment (pp. 12-15),(6) and to debunk predominantly conservative claims that additional imprisonment is cost-effective (pp. 131-33).(7) Their counteroffensive aims to reveal critical deficiencies in the current methodologies used to estimate incapacitation effects, to propose an alternative research strategy, to present some preliminary findings, and to suggest policy implications of those findings.
Available research studies, based on either surveys of prisoners' self-reports of criminal activity or official arrest records, provide widely varying estimates of the marginal benefit of an additional incarcerated offender on crime reduction--from 3 crimes per prison-year, to 14, to 187 (p. 80). Zimring and Hawkins criticize the common model underlying such estimates for its oversimplified view of the criminal career (pp. 44-47). They note that it fails to account for variations in different offender groups' propensity to commit crimes and the degree of selectivity of the incarceration strategy. A selective strategy incarcerates "high rate" offenders first, and thus results in diminishing marginal returns to crime reduction as more and more lower-rate offenders are added to an expanding prison population (pp. 45-53).(8) Furthermore, both offender surveys and official records generate individual crime rate figures that may systematically overestimate the benefits of incarcerating each additional prisoner because they neglect community-level effects. For example, group crime may persist in the absence of one member,(9) and others in the community may "substitute" for the imprisoned offender, taking advantage of the new criminal opportunity (pp. 85-86, 88).(10)
In an effort to consider substitution and community-level effects neglected in the offender-survey and official-record studies, Zimring and Hawkins advocate a "natural experiment" methodology (pp. 92-95)(11) to study the effect of a large increase in imprisonment on California's crime rate in the 1980s (pp. 100-27). The authors' initial findings actually confirm a negative relationship between imprisonment levels and crime volume in California--i.e., additional years of imprisonment are associated with prevention of crime. They estimate hypothetical levels of crime that would have occurred in the absence of the large increase in the rate of imprisonment (pp. 108-14). Actual rates for seven index felonies(12)--both individually and in the aggregate-were below the projected rates, with burglary and larceny offenses constituting more than 90% of the difference between the actual and expected figures (pp. 114-22).
The authors juxtapose these crime-reduction findings, however, with evidence that burglary and larceny rates were reduced most substantially among juveniles--the population for whom incarceration increased the least (pp. 122-25). This "qualitative test" (p. 126) of the findings demonstrates that, for the crimes most affected by the rise in incapacitation, the decrease in arrests occurred among a population different than the group actually incapacitated. Zimring and Hawkins offer this as evidence refuting a simple negative correlation between imprisonment and crime rates, but they do not provide any guidance to resolve the apparent contradiction.
In addition to leaving this contradiction unexplained, the authors' "natural experiment" methodology contains a fundamental flaw. Zimring and Hawkins claim that the very addition of 120,000 prisoners in just over a decade reflects an abrupt "change in policy" assumed to affect the crime rate (p. 104). They contend that the size of the "policy shift" and its abruptness mitigate "the dangers of mistaken causal attribution" (P. 93).(13) They do not, however, raise the critical issue of "simultaneity bias"--the possibility of a simultaneous causal interaction between imprisonment levels and crime rates;(14) i.e., while an increase in imprisonment may reduce the crime rate, any increase in the rate of crime simultaneously may raise the level of imprisonment (assuming the prison sentence for any given offense remains constant).(15) Thus, Zimring and Hawkins may draw erroneous conclusions from an estimated unidirectional relationship and may underestimate the reduction in crime associated with an increase in imprisonment. This would weaken their challenge to the apparent negative correlation between imprisonment and the crime rate.(16)
To "break" this simultaneity bias, or to isolate the independent effect of changes in levels of imprisonment on the rate of crime, researchers need to find a variable that affects the level of imprisonment but does not independently affect the crime rate.(17) In an innovative article, Steve Levitt uses the status of prison-overcrowding litigation as such a variable.(18) When prison-overcrowding litigation forces states abruptly to curtail the number of new prisoners, which in turn reduces the incarceration rate,(19) Levitt finds that crime rates do indeed increase. Levitt's estimates suggest that, absent the quadrupling of the prisoner population of the past two decades, violent crime would be twice as high today and property crime would be over 80% more frequent.(20) His estimates are larger than those of previous studies that, like Zimring and Hawkins's, fail to account for simultaneity bias and thus may systematically underestimate the dampening effect of imprisonment on crime rates.(21)
Incapacitation is part of Zimring and Hawkins's ongoing effort to develop more rigorous analyses of the impact of imprisonment on crime:(22) They challenge the foundations of political slogans and rhetoric, synthesize empirical and policy concerns, and focus on the practical implications of their research efforts. Their harsh critique of the role of economic analysis in criminal justice policy, however, ironically contradicts their valiant call for data and measurement in a field not always responsive to empirical analysis.(23)
Zimring and Hawkins conclude that "[t]he criminal law is a moral system, not an economic one" (p. 153). Yet, if imprisonment is a moral question, then people should be punished for their actions, regardless of incapacitation's cost or its ability to prevent any crime.(24) Implicit in Zimring and Hawkins's concern for social welfare, however, is a markedly more utilitarian perspective on incapacitation. Taking a broader view of the role of economic analysis would strengthen this perspective and lead to improvements in the authors' methodology.(25) Certainly, cost-benefit analysis can at least inform a utilitarian discussion of the best "mix" of crime control and prevention techniques and, thus, the determination of an empirically informed penal policy.(26) In their zeal to denounce contemporary research's emphasis on monetized cost-benefit analysis that compares the dollar costs of prison construction with the dollar savings of crimes averted (pp. 131-54),(27) Zimring and Hawkins downplay the increasingly important role of such analysis in evaluating alternatives to incarceration.(28) Similarly, their dismissal of certain types of empiricism blinds them to the possibility of incorporating economic methodologies into their natural-experiment framework.(29)
Incapacitation is a compelling reminder of the danger of an ideology of crime control rooted in little more than political rhetoric. However, as long as public opinion connects increased levels of imprisonment with reduced levels of crime, policymakers will stifle any impulse to implement the kind of research Zimring and Hawkins suggest. "[F]aith rather than measurement is now, as throughout penal history, the engine for current reliance on general incapacitation in penal policy" (p. 74). Zimring and Hawkins's call for empiricism exposes society's blind faith in the alleged social benefits of incapacitation. Such empiricism, however, must now incorporate economic methodologies and findings in order to identify the true relationship between imprisonment and crime.
-Catherine M. Sharkey
(1.) Elizabeth Alexander The New Turn of the Screw: Why Good News about Controlling Incarceration Rates Safely May Not Be Welcome, 66 S. Cal. L. Rev. 209, 210 (1992) (citing Marc Mauer, Americans Behind Bars: One Year Later 1 (1992)). (2.) There were 1,012,851 men and women locked behind bars in 1994, as compared to 240,593 in 1975. Michael Tonry, Intermediate Sanctions in Sentencing Reform, 2 U. Chi. L. Sch. Roundtable 391, 395 nn.22, 23 (1995). (3.) Studies of the effect of incarceration on crime rates usually ignore crime within prisons. See, e.g., Steve Levitt, the effect of Prison Population size on Crime Rates: Evidence from Prison Overcrowding Litigation National Bureau of Economic Research Working Paper No. 5119, 1995); William Spelman, Criminal Incapacitation (1994); Franklin E. Zimring & Gordon Hawkins, the Scale of Imprisonment (1991) [hereinafter Imprisonment]; Thomas B. Marvell & Carlisle E. Moody, Jr., Prison Population Growth and Crime Reduction, 10 J. Quantitative Criminology 109 (1994). Crimes committed within the prison walls, if explicitly acknowledged in a utilitarian analysis of incapacitation, would reduce the calculated social benefit of crimes averted in society. (4.) Eugene H. Methvin, Doubling the Prison Population Will Break America's Crime Wave, Corrections Today, Feb. 1992, at 28. (5.) Joan Petersilia, California's Prison Policy: Causes, Costs, and Consequences, 72 Prison J. 8, 9 (1992). In a similar vein, Phil Gramm claims that confining one prisoner for an additional year in prison would save society $430,000 in crimes averted. See Phil Gramm, Drugs, Crime and Punishment: Don't Let Judges Set Crooks Free, N.Y. TIMES, July 8, 1993, at A19. Zimring and Hawkins point out that were Gramm's claim true, the 700,000 additional prisoners confined since 1980 would have created some $300 billion in savings--a figure larger than the federal deficit (p. v). (6.) Zimring and Hawkins assert that incapacitation has assumed its dominant position by default due to decreasing reliance on rehabilitation (pp. 3, 4, 20, 21-22, 24), and deterrence (pp. 20, 26). However, they relegate a limited discussion of the arguably central justification of moral desert to a later chapter on jurisprudence (pp. 63, 64-67, 69, 74). Proponents of a moral framework for punishment based on "just deserts" reject the incapacitative rationale. See, e.g., John Kleinig, Punishment and Desert 82-87 (1973); Richard G. Singer, Just Deserts 11-18 (1979); Andrew von Hirsch, Censure and Sanctions 94-97 (1993); Andrew von Hirscii, doing Justice: the Choice of Punishments 19-26 (1976). While empirical analyses can shed light on the relative effectiveness of utilitarian rationales-incapacitation, rehabilitation, and deterrence-such analyses are irrelevant to the nonutilitarian, moral-desert theorist. (7.) See, e.g., William Barr, U.S. Dep't of Justice, the case for more Incarceration (1992); Edwin W. Zedlewski, National Inst. of Justice, Making Confinement Decisions (1987). (8.) But see Marvell & Moody, supra note 3, at 115-18, for an attempt to adjust the individual crime rate of prisoner-s or arrestees to reflect these variations. (9.) Sociologists have at least begun to study co-offender effects on crime rates. See, e.g., Albert J. Reiss, Jr., Co-offending and Criminal Careers, in 10 Crime & Justice: A Review of Research 117 (Michael Tonry & Norval Morris eds., 1988); Albert J. Reiss, Jr. & David P. Fanington, Advancing Knowledge About Co-offending: Results from a Prospective Longitudinal Survey of London Males, 82 J. Crim. L. & Criminology 360 (1991). (10.) John Donohue and Peter Siegelman offer an illustration of this "substitution effect": If ... the average prisoner stole 51 cars in the year immediately prior to his incarceration, then his incapacitation will not reduce the number of auto thefts by 5 1. Many crimes are the product of criminal rings or gangs, and the loss of one criminal member will merely lead to the recruitment of another. While the proposed imprisonment might reduce auto thefts somewhat if the replacement thief that fills the void is less talented or energetic than his predecessor, one simply cannot assume that crime will fall by the full 51. John J. Donohue III & Peter Siegelman, is the United States at the Optimal Rate of Crime? 31 (American Bar Found. Working Paper No. 9404, 1995). Substitution effects are likely to be particularly high for gambling and drug offenses because there are likely to be many others available to take the place of those imprisoned. See, e.g., Mark A.R. Kleiman, Against excess 122-23, 143-45 (1992). (11.) In the social sciences, researchers conduct "natural experiments" by seeking out and comparing situations in which the factors under study vary while other conditions remain constant. For a more extensive discussion of Zimring and Hawkins's natural-experiment methodology, see Franklin E. Zimring Gordon J. Hawkins, Deterrence: the Legal Threat in Crime Control 263-68 (1973). The natural-experiment methodology has been used to study the effect of incarceration policies on the crime rate. See, e.g., Sheldon Ekland-Olson et al., Crime and Incarceration: Some Comparative Findings From the 1980's, 38 Crime & Delinq. 392 (1992). While natural-experiment methodology is a growing trend in economics, some scholars remain skeptical of this technique, and some are even critical of controlled experiments. See generally James J. Heckman, Randomization and Social Policy Evaluation, in Evaluating Welfare and Training Programs 201 (Charles F. Manski & Irwin Garfinkel eds., 1992) (discussing limitations of the experimental method in economics in estimating impacts of social programs). (12.) The seven index felonies are larceny, burglary, robbery, vehicle theft, assault, rape, and homicide (p. 114, Table 6.3). (13.) Zimring and Hawkins suggest that "the more abrupt a change in policy, the more plausible the assumption that the policy change rather than some other factor is responsible for any large contemporaneous fluctuation in the rate of a presumed dependant variable" (p.94). (14.) None of Zimfing and Hawkins's suggested methods for minimizing the risk of mistaken causal attribution magnitude, multiple assessments, and repetitive trials" (p. 93)-addresses simultaneity bias, which is a key element. See Franklin M. Fisher & Daniel Nagin, On the Feasibiliiy of Identifying the Crime Function in a Simultaneous Model of Crime Rates and Sanction Levels, in Deterrance and Incapacitation: Estimating the Effects of Criminal Sanctions on Crime Rates 361, 363-65 (Alfred Blumstein et al. eds., 1978); Marvell & Moody, supra note 3, at 110, 119-20, Daniel Nagin, Crime Rates, Sanction Levels, and Constraints on Prison Population, 12 Law & Soc'y REV. 341, 346-50 (1978); Robert J. Sampson, Crime in Cities: The Effects of Formal and Informal Social Control, in Communities and Crime 271, 286-88 (Albert J. Reiss, Jr. & Michael Tonry eds., 1986) (vol. 8, Crime & Justice: A Review of Research, supra note 9). (15.) See Marvell & Moody, supra note 3, at 119. Alternatively, some researchers believe that crime rates negatively affect sanctions; for example, the "limits of punishment" model posits that as crime rates increase, assuming a constant level of imprisonment is maintained, the rate of imprisonment will decrease. See Fisher & Nagin, supra note 14, at 365; Marvell & Moody, supra note 3, at 119-20. (16.) If, on the other hand, the "limits of punishment" model is true, see supra note 15, then taking into account simultaneity bias would reduce the magnitude of the negative correlation between imprisonment and crime and, thus, could in fact bolster Zimring and Hawkins's argument. (17.) For a straightforward discussion of the technique for estimating the effects of simultaneously related variables, see Nagin, supra note 14, at 346-50. (18.) See Levitt, supra note 3. Levitt classifies six stages of prison litigation-from (1) no overcrowding litigation filed, to (6) release of the prison system from court supervision-in twelve states in which the entire state prison systems have come under court order concerning overcrowding. Id. at 7. Other researchers have used alternative measures of prison capacity to conduct a similar instrumental variable analysis. See, e.g., Nagin, supra note 14, at 353-55 (using average incarceration rate of each state); Sampson, supra note 14, at 284-99 (using actual rated design capacity of jaits). (19.) Levitt estimates the incapacitative effect when the incarceration rate is decreased. See Levitt, supra note 3, at 4. Zimring and Hawkins propose this technique for a hypothetical community-level study as a complement to their own study but do not highlight its potential significance in addressing simultaneity bias (p. 95). While testing changes in one variable in both directions may provide a certain amount of verification, it is still testing a unidirectional relationship and thus neglects any potential bias if the variables simultaneously affect one another. (20.) See Levitt, supra note 3, at 17-18. Levitt's conclusions are derived from "elasticity" estimates of .424 and .379 for violent crime, and .321 and .261 for property crime. id. at 17. The "elasticity" of crime with respect to imprisonment is the percentage reduction in crime that would result from a 1% increase in the prison and jail population. High elasticity values imply that the increases in imprisonment have a substantial crime-reduction effect. For a good discussion of elasticity of crime, see SPELMAN, supra note 3, at 220-21. (21.) See, e.g., Spelman, supra note 3, at 220 (reporting "best, single guess" elasticity estimate of .16 from Rand Corporation.1978 surveys in California, Michigan, and Texas); Marvell & Moody, supra note 3, at 133 (calculating elasticity estimate of .16 for reported index crimes from state-level panel data). (22.) The authors' present work builds on their previous studies. see IMPRISONMENT, supra note 3; Franklin E. Zimring & Gordon Hawkins, Prison Population and Criminal Justice Policy in California (1992). (23.) See, e.g., Nils Christie, Crime Control as Industry 185 (1993) "[J]ustice does not consist of ready-made principles to be excavated using the methods applied in law or in the social sciences, but of common knowledge which each generation has to formulate into legal principles." . (24.) See id. at 185 ("A suitable amount of pain is not a question of utility, of crime control, of what works. It is a question of standards based on values. It is a cultural question." ; supra note 6. (25.) In their critique of existing research methodologies, the authors claim that "[n]o attention has been previously devoted to the systematic explanation of factors tending toward substitution" (p. 55), neglecting the central role that substitution effects play in economic analysis, see, e.g., Donohue & Seigelman, supra note 10, at 31-32. Zimring and Hawkins do not cite regression studies of aggregate crime rates, which, by measuring the net change in crime, do capture group offenses and substitution offenses, see, e.g., Levitt, supra note 3; Sampson, supra note 14, at 271. (26.) Cost-benefit analysis need not be the final word on policy determinations. "While economics is a powerful and valuable tool in the analysis of legal and public policy issues, those who use it need an appreciation of its limitations." John J. Donohue III & Ian Ayres, Posner's Symphony No. 3: Thinking About the Unthinkable, 39 Stan. L. Rev. 791, 793 (1987) (book review). In the present context, "[t]he finding that prisons appear cost-beneficial does nothing to reduce the importance of identifying and correcting those factors that lie at the source of criminal behavior." Levitt, supra note 3, at 26. (27.) The authors present a conservative economist's flawed monetary cost-benefit methodology as representative of the economic approach: Edwin J. Zedlewski, When Have We Punished Enough?, 45 Pub. Admin. Rev. 771, 778 (1985) (arguing "that greater social benefits are derived from prison incarceration than are usually assumed" (p. 142)). Zediewski claims that one year's imprisonment costs $25,000, compared to some $430,000 in social costs avoided by incapacitation (p. 143) (citing Zedlewski, supra note 7, at 3, 4). The authors contend that Zedlewski's monetized estimates merely reflect an inflated underlying estimate of 187 nondrug crimes per year saved with each additional prisoner incarcerated an overestimate by a range of between 1,000 and 2,000 percent' (p. 145). (28.) The authors do admit a certain utility in using cost-benefit analysis to weigh policy alternatives valuating incapacitation within the broader context of crime control (pp. 74-75). Cf. Michael C. Musheno et al., Community Corrections as an Organizational Innovation: What Works and Why, 26 J. Res. Crime & Delinq. 136, 137 (1989) ("[A) total commitment to the incarceration of all adult felons represents a costly enterprise that cannot be sustained in practice. Ironically, the institutionalization of incapacitation is providing a strategic opening for community corrections and other alternatives to incarceration (e.g., electronic monitoring)." . (29.) The authors criticize "the value of low-budget multiple regression studies of the variety frequently found in econometrics and other social science areas" (p. 93). Regression analysis, a statistical technique that isolates the independent influence of each of several variables on a dependent variable of interest, is used in many sound studies. See. e.g., Spelman, supra note 3, at 2-10 (describing economic model of crime control through incapacitation, specifying social tradeoff between crime and imprisonment-both measured in dollar amounts); Joel Waldfogel, Are Fines and Prison Terms Used Efficiently? Evidence on Federal Fraud Offenders, 38 J.L. & Econ. 107 (1995) (constructing theory of efficient punishment based on economic model of punishment production). There is no reason why a natural-experiment methodology cannot incorporate econometric modeling. See, e.g., John B. Knight & Richard H. Sabot, Education, Productivity, and Enequality: the East African Natural Experiment 7-14 (1990).
Professor of Law, Boalt Hall School of Law; Director, Earl Warren Legal Institute, University of California at Berkeley. Senior Fellow, Earl Warren Legal Institute. Incapacitation: Penal Confinement and the Restraint of Crime. By Franklin E. Zimring(*) and Gordon Hawkins.(**) New York and London: Oxford University Press, 1995. Pp. v, 188. $27.95.
|Printer friendly Cite/link Email Feedback|
|Author:||Sharkey, Catherine M.|
|Publication:||Yale Law Journal|
|Article Type:||Book Review|
|Date:||Mar 1, 1996|
|Previous Article:||To Secure These Rights: The Declaration of Independence and Constitutional Interpretation.|
|Next Article:||Constitutional Law as Fiction: Narrative in the Rhetoric of Authority.|