Professional Documents
Culture Documents
1. Introduction
Few policy issues have the potential to arouse as much intense public
debate as whether governments are doing enough to combat criminal
activity (or whether the actions being taken are the most effective). For
example, statistics indicating the pervasiveness of different types of crime
and trends in crime rates are widely disseminated and debated. Similarly,
public surveys of the publics fear of crimeparticularly violent crime
have demonstrated growing concern about perceived crime risks amongst
the general public as well as among specific demographic groups. As a
result, governments in many countries are in the process of investing significant political capital in numerous initiatives that aim to reduce crime in
general or that target specific types of offending, such as gun crime. Of
course, these initiatives also involve the commitment of substantial economic resources. Hence, one interesting question is the extent to which
these policy interventions are informed by economic reasoning; that is,
with reference to some form of cost-effectiveness or costbenefit thinking.
The economics of crime covers a wide range of issues, such as the effective deterrence and punishment of criminal activity (see, Di Lulio, 1996,
for a survey). A number of studies have also devoted attention to the economic costs that crime (and the fear of crime) imposes on society.
Interestingly, many of these studies construe the true costs of crime
broadly to include both tangible costs (e.g. financial outlays on personal
security and the criminal justice system) and intangible costs (e.g. the anxiety suffered by potential victims or the pain and suffering imposed on
actual victims). Indeed, for certain types of crimein particular those
Giles Atkinson is lecturer in environmental policy at the London School of Economics and Political
Science (LSE). Susana Mourato is senior lecturer in environmental economics, Department of
Environmental Science and Technology, Imperial College London. Andrew Healey is research
fellow at LSE Health and Social Care.
79
entailing violence or the threat of violenceit seems plausible to speculate that this intangible cost will be highly significant.
In this paper we review the findings of recent studies that have sought
to measure the costs of crime.
Firstly, we discuss those approaches that describe the big picture by
calculating the aggregate burden of all crime. While such studies have
yielded useful overall summaries about the magnitude of the crime problem, they have also revealed that surprisingly little is known, for example,
about the value of the intangible effects of crimes that entail actual violence or the credible threat of violence. However, a growing number of
contributions to this literature have begun the process of filling this gap
through novel applications of established valuation methodsused extensively, for example, in environmental economicsto a crime context.
Secondly, we discuss those new (micro) approaches that have sought to
value the preferences of individuals or households for reductions in crime
risks using stated preference methods such as contingent valuation.
Contingent valuation (CV) is a survey-based technique that typically
involves asking respondents to state their willingness to pay (WTP) for
specified policy changes: the elicited amounts are monetary measures of
respondents preferences (Mitchell and Carson, 1989; Bateman et al.,
2002). Although controversial, stated preference approaches are widely
used today to inform policy decisions. Indeed, this use has been codified
in policy-making by the recent publication by the UK Government of a
manual outlining best practice in using these methods in UK policy decisions (see, for example, Bateman et al. 2002). In the context of crime,
much of this work has focused on violent crime risks where the predominant costs are mainly intangible; that is, adverse impacts on well-being
associated with the physical pain and mental distress of victimisation. Of
course, placing a money value on the physical and psychological harm
resulting from violent crime is far from straightforward. Yet neither is it an
impossible task. Moreover, such findings have direct policy relevance
through the provision of valuable and new information about the benefits
of policy initiatives that aim to reduce crime.
80
of crime to society. Prominent amongst these contributions are those studies that assess the aggregate burden of criminal offending in a country.
These studies provide useful information primarily by indicating the magnitude of the crime problem, broadly construed, as a percentage of, say,
Gross Domestic Product (GDP) or in per capita terms. Moreover, in using
money values to convey these magnitudes, data, for example, about disparate tangible and intangible consequences of crime are described in a
readily comparable manner. Thus, estimates of aggregate burdens also
reveal useful information about the relative magnitude of the different
components of the costs of crime.
Such information might be a powerful argument for the strengthening
of policy efforts to reduce the incidence of crime. Even so, the specific policy implications of knowing the overall magnitude of the burden of crime
(and its components) are more limited. For example, this information does
not tell us which public policy interventions will be more cost-effective or,
more ambitiously, which will secure the greatest net benefits for society
(although some of the constituent data assembled might plausibly be redirected to this objective). Furthermore, any approach that seeks to paint
the overall picture will use a broad brush. As a result, it is almost inevitable
that some important details concerning gaps in our knowledge about the
intangible costs of crime are tackled in a relatively crude way (perhaps by
gap-filling missing data by borrowing values from other policy contexts).
Nevertheless, it is instructive to consider the question of what is the
money value of the aggregate burden of crime and then, bearing in mind
the above caveats, to consider in more detail how more policy-useful
measures of the benefit of reducing particular types of crime can be constructed. Recent illustrations of studies that have estimated the aggregate
burden of crime can be found for both the United States and the United
Kingdom. For example, in the case of the former, Anderson (1999) examines the problem within a rigorous costbenefit framework. A summary of
Andersons findings is provided in Table 1 and illustrates that the aggregate burden of US crime in 1997 has four components:
Firstly, crime-induced production is defined as expenditures that do not
contribute to welfare except for their association with crime. For example,
this category includes such items as financial outlays to protect ones person or household (so-called defensive expenditures) and the cost of
incarcerating criminals.
81
82
83
84
85
86
injuries and psychological distress likely to be associated with being a victim of a violent assault.
Perhaps unsurprisingly, it turns out that the process of eliciting these
values is every bit as interesting as the outcome in terms of the findings
themselves. For example, even if it can be assumed that these categories
of offence and their consequences for victims are well understood by
experts in the crime policy field, such characterisations are far less likely
to be meaningful to the wider public. Of course, it is a sample of the latter
that comprises the respondents in a CV study. This means that, in designing a CV scenario to elicit WTP, respondents need to be presented with a
comparable but meaningful description of the symptoms likely to be suffered after a criminal offence. Moreover, this description must provide as
clear a signal as possible as to the way in which these symptoms would
affect lives and well-being.
The approach taken by Atkinson et al. was to describe these crime outcomes in terms of the experience for a typical victim. In other words, what
was conveyed to respondents was the idea that if they were to be subjected to an assault of a given severity tomorrow, then it would be reasonable to expect them, on average, to experience a core set of adverse
health outcomes. In turn, the basis for selecting this core set of symptoms
was through an extensive review of the relevant medical literature, as well
as expert advice from the field. In particular, this available evidence indicates that the most important categories of outcomes relate to physical
injuries (perhaps requiring medical treatment) as well as a potential array
of symptoms of mental distress, such as those characterising an episode of
acute or long-term psychological reaction to a traumatic event. Table 3
summarises the physical injury and mental distress descriptions used by
Atkinson et al. Thus, common assault was described to respondents as
entailing no injury and short-term distress whereas other wounding was said to
result in moderate injury and medium-term distress. Lastly, serious wounding
was described as involving serious injury and long-term distress.
Turning now to the findings of this application of CV to violent crime
prevention in the United Kingdom, we limit our discussion to summarising the basic WTP results and their implications for calculating the costs
of violent crime in the UK. (See Atkinson et al. 2003 for a fuller discussion
of study description, results/analysis and caveats.) Each respondent was
asked to consider the benefits to themselves of reducing the risk of
87
88
Medium-term distress profile
Repeated recollections of assault
Difficulty falling asleep or staying asleep
(one or two nights each week)
Difficulty concentrating on daily tasks
Symptoms last for two weeks
No injury profile
None
Psychological
distress
Physical injury
Other wounding
Common assault
Serious wounding
Table 4: The cost of crime reduction: WTP and implied cost of statistical
crimes (/crime/year)
Variable
WTP
(Common assault)
WTP
(Other wounding)
WTP
(Serious wounding)
Risk of crime
4%
(2%)
1%
(0.5%)
1%
(0.5%)
106.31
(72.93139.69)
154.29
(108.21200.37)
178.33
(124.70231.96)
5,316
(3,8837,276)
30,858
(22,85841,548)
35,666
(26,40448,178)
Notes:
1. Adapted from Atkinson et al. (2003)
2. Calculated using parametric statistics on WTP.
3. CI: confidence interval.
4. Common assault (no injury + short-term psychological distress): risk reduction 2%
5. Other wounding (moderate injury + medium-term psychological distress): risk reduction 0.5%
6. Serious wounding (serious injury + long-term psychological distress): risk reduction 0.5%
suffering the health outcomes associated with one of the three types of
violent assault described in Table 3. Respondents were also informed of
the pre-policy risk of the incident (and its outcomes) occurring (column 2
in Table 4). These specified risks were an approximation of the actual
(average) risks of being a victim of a violent assault currently faced by a
member of the public in the UK. Each respondent was then asked to
express his or her WTP to reduce his or her chance of being a victim of
this offence by 50% over the next 12 months. The payment vehicle for
this change was a one-off increase in local charges for law enforcement.
This payment was justified to respondents as a contribution to the costs of
reducing the incidence of violent crime in the area where they live.
Table 4 (column 3) reports the mean WTP to reduce the risk of various
crime offences. The results show that mean WTP increases with the
severity of injury described to respondents. This is consistent with the
theory underpinning the use of WTP as a measure of value in that we
would naturally expect people to pay more where the severity of an
offence is increased. Note that although the severity of the incident is considerably higher for both other wounding and serious wounding when compared to common assault, the (baseline) risk (column 2) of being a victim of
these offences (1%) is four times smaller than the risk of suffering a
89
common assault (4%). However, while it can be shown that meaningful statistical differences are found between WTP for reducing the risk of common assault and WTP to reduce the risk of serious wounding, this is not the
case between the other mean amounts. This is probably due to a combination of the large variation in individual responses and the relatively
small sample sizes available for each sub-sample. In all, there were 522
usable responses; however, each respondent was only asked to value one
category of criminal offence. In particular, it cannot be assumed that valuing violent crime is an easy or palatable cognitive task for people, and studies such as this represent only first forays into this important policy
domain.
From the point of view of estimating the costs of crime or the benefits
of policy interventions that seek to reduce crime by a given (predicted)
number of offences, a useful indicator is the cost of a statistical crime. Put
another way, this transformation gives the implied amount that the average individual would be willing to pay to avoid a violent crime happening
with certainty. It is this price which can be multiplied by the number of
incidents per year of a specific type of crime to obtain an estimate of the
aggregate costs of these three categories of violent crime. As an illustration, for the no injury (common assault) case, mean WTP for a 2% risk
reduction is 106.31, giving a value of 53.16 per percentage reduction in
risk. Multiplied by 100 this yields 5,316 as the estimate of the value of a
statistical common assault. If we multiply this value (i.e. 5,316) by the
number of incidents of common assault in a given year in the UK, then we
can provide an estimate of aggregate (annual) cost of this category of
offence.
Using data drawn from the 2001/2 British Crime Survey (Simmons et al.
2002), Table 5 summarises the number of incidents of common assault and
wounding (other- and serious-) and the aggregate costs of these categories of
offence. With regards to the number of incidents, column 2 indicates that,
particularly in the case of common assault, the risks of being a victim are far
from being trivially small. Column 3 makes it plain that this volume of
incidents involving violent crime therefore translates into a substantial
annual burden for society. Note that these coststhe product of the number of incidents of a category of offence and its cost per (statistical)
crimerefer only to the intangible costs of these crimes, i.e. the cost of the
pain and suffering inflicted on victims. It is also worth noting that Table 5
90
Common assault
Other wounding
Serious wounding
Total ( million)
Percentage of GDP (%)
1,791
1,578
1, 81
9,521
17,822
2,905
30,248
3.0
Notes:
1. Underlying data on number of incidents adapted from Simmons and colleagues (2002). These are based on BCS interviews carried out in
a 12-month period over 2001/2. BCS respondents are asked to recall their experience of crime over the previous 12 months. Hence, these
data over the period during which the CV study in Atkinson et al. (2003) was implemented.
2. Simmons and Colleagues (2002) summarise data for total wounding only (659,000). Number of other- and serious-wounding incidents
estimated using information in Brand and Price (2000).
3. Numbers may not sum due to rounding.
4. GDP refers to Gross Domestic Product for United Kingdom (average of GDP for the years 2000 and 2001).
91
5. Concluding remarks
Our discussion has, in effect, come full circle. We began by describing two
recent studies that tried to estimate the aggregate cost or burden of crime,
which we described in terms of an effort to understand the big picture.
In doing so, we drew attention to the paucity of data in such studies, and
more generally, about the intangible costs of crime. As a result, studies of
aggregate costs of crime typically are forced to borrow value data elicited
in non-crime contexts. However, we also noted emerging interest in applying, to the crime context, stated preference methods such as CV used
extensively in other fields such as environmental policy. In particular, we
illustrated a recent example that used a CV study to elicit WTP to reduce
the risk of three different categories of violent crime in England and
Wales. Finally, we showed how the findings of that particular study can be
used to illustrate the aggregate costs of these crimes. The major distinction between this last example and the studies discussed in the first part
of this paper is that the former uses WTP data derived in an explicit crimecontext. Such studies are an important development.
Of course, understanding the big picture is just one application of
these data on the intangible costs of crime. Just as importantly, this information could enable better costbenefit comparisons by allowing a more
thorough appraisal of the benefits of policies that seek to reduce the incidence of violent crime. The studies we have described in the latter part of
this paper can be characterised as the first steps in facilitating this policy
application. Moving beyond these exploratory efforts would suggest the
need for further work in the application of stated preference methods to
crime reduction policies. Hence, the answers that policy-makers require
will not be available immediately. In the interim, this also suggests that
WTP data borrowed from existing similar but non-crime valuation studies remains a worthwhile first approximation. However, to reiterate, if context matters, monetary values derived from studies designed in an explicitly
crime context are ultimately a necessity.
Whilst there can be little argument against the view that valuing the
intangible costs of crime is an inherently challenging task, it is arguably
then less easy to justify avoiding doing so. It is an inescapable fact that in
a resource-constrained public policy environment, decision-makers
already place implicit valuations on crime and its impact on victims via the
92
relative public spending priorities that are given to different policies that
target criminal offending. A more explicit assessment is needed of the
monetary value of crime reduction, not just to improve the transparency of
public decision making but also to ensure that policy benefits can be compared directly with the costs of implementation. Whilst economic considerations are not the sole criterion for judging the desirability of specific
crime prevention programmes, this might lead to more consistent priority
setting where decision making is (at least partly) concerned with allocating available resources to maximum social advantage.
References
Anderson, D.A. (1999) The Aggregate Burden of Crime, Journal of Law and
Economics.
Atkinson, G., Healey, A. and Mourato, S. (2003) Valuing the Costs of Violent Crime:
A Stated Preference Approach, London School of Economics and Political Science,
mimeo.
Bateman, I., Carson, R.T., Day, B., Hanemann, M., Hanley, N., Hett, T., Jones-Lee,
M., Loomes, G., Mourato, S., Ozdemiroglu, E., Pearce, D.W., Sugden, R. and
Swanson, J. (2002) Economic Valuation with Stated Preference Techniques: A Manual,
Cheltenham: Edward Elgar.
Brand, S. and Price, P. (2000) The Economic and Social Costs of Crime, London: Home
Office Research Study 217.
Cohen, M.A. (1988) Pain, Suffering, and Jury Awards: A Study of the Costs of Crime
to Victims, Law and Society Review, 22: 538555.
Cohen, M.A. (2001) The Crime Victims Perspective in CostBenefit Analysis: The
Importance of Monetising Tangible and Intangible Crime Costs, in Welsh, B.,
Farrington, D.P. and Sherman, L.W. (eds.) The Costs and Benefits of Preventing Crime,
Colorado: Westview Press.
Cohen, M.A., Rust, R.T., Steen, S. and Tidd, S. (2001) Willingness to Pay for Crime
Control Programs, Vanderbilt University, mimeo.
Cook, P.J. and Ludwig, J. (2000) Gun Violence: The Real Costs, Oxford: Oxford
University Press.
93
Corso, P.S., Hammitt, J.K. and Graham, J.D. (2001) Valuing Mortality-Risk
Reduction: Using Visual Aids to Improve the Validity of Contingent Valuation,
Journal of Risk and Uncertainty, 23: 16584.
Di Lulio, J.J. (1996) Help Wanted: Economists, Crime and Public Policy, Journal of
Economic Perspectives, 10(1): 324.
Gibbons, S. (2002) The Costs of Urban Property Crime, London School of
Economics and Political Science, mimeo.
Hammitt, J.K. and Graham, J.D. (1999) Willingness to Pay for Health Protection:
Inadequate Sensitivity to Probability?, Journal of Risk and Uncertainty, 18: 3362.
Jones-Lee, M.W., Hammerton, M. and Phillips, P.R. (1985) The Value of Safety:
Results from a National Sample Survey, Economic Journal, 95: 4972.
Jones-Lee, M.W., Loomes, G. and Phillips, P.R. (1995) Valuing the Prevention of
Non-Fatal Injuries: Contingent Valuation vs. Standard Gambles, Oxford Economic
Papers, 47: 676695.
Kershaw, C., Budd, T., Kinshott, G., Mattinson, J., Mayhew, P. and Myhill, A. (2000)
The 2000 British Crime Survey, England and Wales, London: Home Office.
Krupnick, A., Alberini, A., Cropper, M., Simon, N., OBrien, B., Goeree, R. and
Heintzelman, M. (2002) Age, Health and the Willingness to Pay for Mortality Risk
Reductions: A Contingent Valuation Survey of Ontario Residents, Journal of Risk and
Uncertainty, 24(2): 161186.
Ludwig, J. and Cook, P.J. (2001) The Benefits of Reducing Gun Violence: Evidence
from Contingent Valuation Survey Data, Journal of Risk and Uncertainty, 22(3):
207226.
Miller, T.R., Cohen, M.A. and Wiersma, B. (1996) Victim Costs and Consequences: A New
Look, US Department of Justice, National Institute of Justice, Washington D.C.
Mitchell, R. and Carson, R. (1989) Using Surveys to Value Public Goods: The Contingent
Valuation Method, Resources for the Future, Washington.
Simmons, J. and Colleagues (2002) Crime in England and Wales, 2001/2, Home Office,
London.
Thaler, R. (1978) A Note on the Value of Crime Control: Evidence from the
Property Market, Journal of Urban Economics, 5: 137145.
94