Costs and Benefits of Preventing Crime
edited by Brandon C. Welsh, David P. Farrington, and Lawrence W. Sherman
In March 1999, the Department of Criminology and Criminal Justice of the University of Maryland convened a two-day seminar that focused on the current state of knowledge on the monetary costs and benefits of crime prevention program effects. The conference brought together experimental criminologists (Lawrence Sherman, David Farrington, and Brendan Welsh), public policy analysts (e.g., Peter Greenwood from Rand Corporation), Commonwealth senior civil servants (e.g., Peter Golblatt from the Home Office), as well as researchers who have incorporated cost-benefit analysis into clinical research (Brian Yates and Faye Taxman) or developed strategies for evaluating the intangible costs of criminal victimizations (Mark Cohen). Philip Cook, who signs the book's Foreword, neatly captures both the purpose and the achievement of the volume. He points out that most evaluations of crime reduction programs have worked hard to "assess their effects on criminal behaviour but stopped short of estimating the associated costs and benefits. As a result we know more about what works than about what's worthwhile. The business of assessing costs and benefits, however, requires a considerable investment in the conceptual framework as well as in estimation technique. This book provides an important beginning on both".
Boulder, Colorado: Westview Press. 2001
The central piece (Chapter 3) of the book is Welsh and Farrington's review of the economic efficiency or monetary value of 6 developmental crime prevention programs (e.g., Elmira Prenatal-Early Intervention Project), 13 situational crime prevention projects (e.g., Stoke-on-Trent Improved-Street-lighting Project) and 7 correctional offender treatment efforts (e.g., New York City's Supported Work Social Experiment for ex- inmates and drug addicts). About half of these studies qualified as high-quality evaluation designs. Cost- benefit ratios were mostly estimated from the point of view of the tax-payer (How much do tax-payers have to pay to operate the program ? And how much do they receive from it ?). Whereas costs were often taken into full account, Welsh and Farrington conservatively assessed their benefits but nonetheless include direct tangible crime victim expenses, savings from reduced criminal justice, social and health expenditures as well as increased tax revenues from higher legal earnings. Findings suggest that all three crime prevention strategies can "demonstrate value for money". Moreover there is much to learn from "inefficient" programs as well.
The Stoke-on-Trent experiment (Painter and Farrington, 1999) examined the impact on crime levels brought about by upgrading the street lighting in a number of urban areas that failed to satisfy minimum standards and by installing additional lighting in previously unlit detached footpaths. After one year of operation of the improved-street-lighting scheme, findings showed a 25% drop in household victimizations in experimental areas, a 21% drop in adjacent areas ("diffusion of benefits") and a 12% increase in matched control areas (this increase cannot be interpreted as a displacement effect because these control areas were geographically distant from experimental targets) . The project achieved an efficiency (or benefit/cost) ratio of 2.93 ($2.93 costs avoided for every $1.00 invested). Benefits derived mostly from avoided tangible crime victim expenses and criminal justice (police) expenditures. Costs assessment included the additional maintenance and electrical energy costs as well as the annual debt payment on the capital cost.
New York's Supported Work randomised experiment aimed at reintegrating former offenders and drug addicts into the community through the provision of paid employment and methadone maintenance for ex-addicts (Friedman, 1977). After 16 months, the arrest rate for experimental subjects was found to be only marginally lower than for control subjects; moreover, there was no difference either in the prevalence of subjects participating in drug treatment programs. However, the treatment group members achieved higher employment earnings and reduced their reliance on social services. As a result the experiment achieved a desirable 1.13 efficiency ratio because much of the program benefits derived from savings in public goods and services and reduction in social service or welfare use. The interesting lesson from this experiment is that a failure in reducing recidivism rates can nonetheless be economically efficient - a very significant finding for recidivism research.
A half-rural community in upstate New York was the setting of the Prenatal-Early Infancy randomised experiment designed to improve the quality of care that mothers provide to their children (Olds et al., 1998). Women in the treatment condition received continued nurse home visits until children reached the age of 2 years. After two years was a three-fourth reduction in state-verified cases of child abuse and neglect, 32% fewer emergency room visits, 80% greater participation in the work-force and 43% fewer subsequent child- births for higher-risk mothers. Thirteen years after the completion of the program, fewer treatment mothers compared to control mothers were identified as perpetrators of child abuse and neglect and for the higher-risk sample fewer treatment mothers as well as their children had been arrested and fewer mothers exhibited substance abuse problems. Although the overall efficiency ratio for this experiment was 0.51 after two years and 0.62 after 13 years, the benefit-cost ratio for the higher-risk sample was 1.1 after one year and 4.1 after 13 years. Cost-avoided derived mainly from reductions in Food Stamps and Medicaid expenditures. Savings derived from increased tax revenues and governmental savings related to fewer cases of child abuse and neglect represents respectively 5 and 3% of total estimated governmental savings.
Most of the significant contributions of the book (including Daniel Nagin's concluding comments in Chapter 9) may be viewed as focused discussions on the various conceptual and estimation issues raised by this broad and comprehensive assessment of the economic efficiency of current crime prevention projects. This in turn gives a strong analytical unity to the book. Greenwood and associates (Chapter 4) reanalysed two early childhood experiments: the Early Infancy project in Elmira, New York and the Perry Pre-school project in Ypsilanti, Michigan (Schweinhart et al., 1993). Unsurprisingly, one theme of Greenwood's contribution is the need for "selective targeting" in crime prevention experiments. Another theme is that, for both projects, savings or benefits accumulated over time. The time-lag before cumulative savings to government exceeded cumulative costs occurred relatively quickly for Elmira's higher-risk families (4 years only) and relatively late for the experimental children in the Perry Pre-school project (25 years). Reduction in welfare cost represented the main benefit of the Elmira project (57% of all benefits), whereas reduction in criminal justice costs was the principal benefit achieved by the Perry Pre-school experiment (40% of all savings). Why? The Elmira experiment targeted mothers and only secondarily their off-spring and this accounts for the restricted duration of the follow-up (15 years). The Ypsilanti experiment, on the other hand, targeted 3 and 4 year olds and the follow-up was pursued 25 years after the completion of the intervention: as a result it captured the crucial transition from adolescence to adulthood where savings in reducing crime involvement are the highest. Another interesting finding of the Greenwood paper is that, although the estimated net present value of future benefits is very sensitive to the choice of the discount rate, both experiments "generated positive savings to government and to society as a whole for a reasonable range of rates".
An appealing feature of the book is that it openly acknowledges the fact that monetizing the intangible costs of criminal victimizations represents a crucial intellectual challenge for cost-benefit analyses of crime prevention programs. Mark Cohen's remarkable contribution (Chapter 1) persuasively underlines the importance of monetizing both the tangible and intangible crime costs and of incorporating the "crime victim's perspective in cost-benefit analysis". Non-marketable goods and services are generally not included in cost- benefit analyses, but, as Cohen notes, the most significant cost of crimes are the non-marketable "bads" they produce, namely pain, suffering and reduced quality of life. This explains why reductions in child abuse and child neglect were estimated to represent only 3% of the overall "benefits" achieved by the Elmira experiment. Cohen reviews the direct and indirect strategies for estimating the intangible or non-monetary costs of crime. One strategy is to infer victims' willingness to pay for increased safety by analysing actual market transactions and isolating a specific effect of crime levels on current prices (Hoen et al., 1987; Bartley, 1999). Another method is to analyse wage rate differentials across risky and less risky gobs and to isolate and estimate the coefficient of the "premium rate for accepting an increased risk of death" (Viscusi, 1993). This approach has also been used to assess the current market "wage" for offenders involved in drug and other criminal activities. Another strategy (used by Cohen, 1988) was to use jury award data to estimate the monetary value of pain, suffering, and lost quality of life for non-fatal accidental injuries. Miller, Cohen and Wiersema (1996) pursued this line of investigation and obtained data on jury awards to victims of physical and sexual assault and estimating non-monetary costs of crime on the basis of these cases. An alternative approach is to survey the public directly as potential victims and ask participants to place dollar values on non-market goods. This "contingent valuation" approach has been extensively used in environmental economics (Mitchell and Carson, 1989). Cohen presents a useful review of the most recent estimates of the cost of crime to victims that combine out-of-pocket losses, risk of death probabilities weighted by value of life estimates, and jury awards for non- fatal injuries. Cost of crime victimisations include losses in productivity, medical and mental expenditures, property or damage losses as well as quality of life losses: the average "cost" of a no-injury robbery is estimated to be $2,000 whereas the average cost of a sexual child abuse (including rape) is estimated to be $99,000. In an otherwise uninspiring cost-benefit inventory of crime prevention programs (Chapter 5) - sub- titled "A review of research findings with implications for Washington State" - Steve Aos and associates showed that a cost-benefit analysis that adjusts for intangible victimisation costs significantly increases the efficiency ratio of most crime prevention endeavours: for example, the benefits per dollar of cost for short-term financial assistance projects for inmates leaving prison is estimated increased from an unsatisfactory $0.77 to a more desirable $1.08.
Another interesting contribution - although a bit too eager for this reader's taste- is that of Faye Taxman and Brian Yates ("Quantitative exploration of the Pandora's Box of Treatment and Supervision: What Goes on Between Costs In and Outcomes Out"). Conventional cost-benefit analyses, including Cohen's assessment of monetary and non-monetary costs of crime, typically rely on aggregate data from national surveys and cross-sectional studies. Taxman and Yates suggest the relevance of incorporating cost-benefit analyses in the actual case management of offenders under correctional jurisdictions. They are currently directing a research project whose goal is to assess the efficiency and effectiveness of two types of case management of "offenders who are involved in property and drug crimes as a means of funding their substance abuse habits". The project involves a block randomised experiment in which subjects drawn from two settings (Alexandria, Virginia and Montgomery County, Maryland) are placed into stratified groups (high versus moderate risk offenders) and then assigned into "coercive" treatment condition (cognitive behaviour therapy, urine testing and graduated sanctions for non-compliance) and a standard probation control condition ("supervision with testing). Despite their promising title ("specifying what goes on between costs and outcomes") the hypothesised intervening "process variables" outlined in Figure 2.1 (p.59) are undefined - indeed the paper never actually justifies their relevance and merit. An attractive feature of their paper is that they have designed a strategy for developing "per-person" cost measures (supervision, drug testing, collateral contacts, time in cognitive therapy and other treatment services and so on) as well as "per-person" benefit measures (costs saving in welfare, health and criminal justice expenditures, etc.). Their goal is to analyse how "net benefit" and "time-to-return-on-investment" is related to treatment procedures, client characteristics, and "needs" (or "risk levels").
The book is divided into three parts. Contributions by Cohen and Taxman/Fayes make up the first section ("Methods and perspectives of economic analysis"). Papers by Welsh, Farrington and Greenwood are regrouped in the second section ("Economic Analysis Findings"). The third section ("International Policy Perspectives") is somewhat disappointing. The title of John Chisholm's contribution - "Economic Analysis of Crime Prevention: An Australian perspective" - is somewhat misleading since what it offers, basically, is a sketchy inventory of a number of American and British cost-benefit studies. (It is quite unfortunate that Chisholm did not choose, instead, to concentrate on his own research on the economic efficiency of incarcerating burglary offenders since their preliminary findings suggested an efficiency ratio of 3.5). Daniel Sansfaçon and Irving Waller's contribution provide a well-rehearsed globe-trotter-senior-civil-servant-pep-talk on crime prevention policies in a number of leading industrialised nations (Belgium, Canada, France, Netherlands, Sweden, U.K., U.S.A., New Zealand, Australia). Their main finding (not totally unexpected) is that very few governments, besides Great-Britain, United States, and The Netherlands, actually promote an evidence- and efficiency-based approach to crime reduction efforts. " In countries such as France, Sweden, or Denmark, where the national strategy is overarching and is not based on a project approach, there is much less to say about evaluation". Funded evaluations in France of local community prevention councils, for instance, are not so much designed to demonstrate crime reduction effects but to qualitatively document successful community action: "Has the local coalition been implemented? Has it facilitated citizen participation? Has it been active and visible? Is there a qualitatively different perception of community safety locally?" It is very highly unlikely that such "policy driven crime prevention" will ever emerge as an appropriate way to "make the best use of accumulated knowledge and practices". On the other hand, the contribution by Sanjay Dhiri and associates, all from the British Home Office, do a very good job of presenting the "Crime Reduction Programme" launched in 1998 by the U.K. government. This ambitious investment in an evidence- based approach to crime reduction implies that CRP funded initiatives must not only be suitable for evaluation but must also build on existing knowledge about a particular policy or technique by: a) demonstrating the contexts, mechanisms, and processes essential to the program's success; b) indicating the time scales and sustainability of benefits; c) quantifying its cost-effectiveness; and d) specifying the likely requirements for larger-scale implementation. This evidence-led approach to crime prevention also applies to related programs designed to combat drug use and related crime and to provide constructive prison regimes. Five key domains of research were identified by the CRP: a) working with families, children, and schools to prevent young people from becoming the offenders of the future; b) tackling crime in communities, particularly high-volume crime such as domestic burglary; c) developing products and systems that are more resistant to crime; d) more effective sentencing practices (e.g., improving fine enforcement); and e) working with offenders to ensure that they do not re-offend. And because the "one of the first tasks for the Home Office team was to begin to develop a common language among the many managers, practitioners, and evaluators involved in CRP initiatives", this chapter provides a very useful set of definitions and detailed guidelines on how to go about measuring inputs and costs, identifying and valuing outcomes and benefits (including those not easily valued), and comparing costs with outcomes. All in all, then, a very useful, insightful and innovative book on a neglected and important topic.
- Bartley W.A. (1997) A valuation of specific crime rates. Ph.D. diss., Vanderbilt University.
- Friedman L.S. (1977) An interim evaluation of the Supported Work experiment, Policy Analysis 3: 147- 170.
- Hoen J.P., M.C., Berger, and G.C. Blomquist (1987) A hedonic model of interregional wages, rents and amenity values, Journal of Regional Science, 27, 605-620.
- Mitchell R.C. and R.T. Carson (1989) Using surveys to value public goods. Washington D.C.: Resources for the future.
- Olds D.L., C.R. Henderson, C. R. Cole, J. Eckenrode, H. Kitzman, D. Luckey , L.M. Pettitt, K. Sidora, P. Morris, and J. Powers (1998) Long-term effects of home visitation on children's criminal and antisocial behavior: 15-year follow-up of a randomized controlled trial. Journal of the American Medical Association, 280: 1238-1244.
- Miller T.R., M.A. Cohen, and B. Wiersema (1996) Victim costs and consequences: A new look. Washington D.C.: U.S. Department of Justice, National Institute of Justice.
- Painter K.A. and D.P. Farrington (1999) Street lighting and crime: Diffusion of benefits in the Stoke- on-Trent project. In K.A. Painter and N. Tilley, (eds.), Crime prevention studies, 10:77-122. Monsey, N.Y. : Criminal Justice Press.
- Schweinhart L.J., H. Barnes, and D. Weikart (1993) Significant benefits: The High/Scope Perry Preschool study through age 27. Ypsilanti, Mich.: High/Scope Press.
- Viscusi W.K. (1993) The value of risks to life and health, Journal of Economic Literature 31: 1912- 1946.
Université de Montréal (Criminologie)