307
© 2010 American Society of Criminology
Criminology & Public Policy • Volume 9 • Issue 2
Policy Essay
CRIME COSTS ACROSS
OFFENDER TRAJECTORIES
The costs of crime
Jens Ludwig
U n i v e r s i t y o f C h i c a g o
National Bureau of Economic Research
W
hat economist would not be delighted to see more work in criminology devoted
to benet–cost analysis (BCA)? In my policy essay, I would like to make three
points.
First, the application of BCA to crime policy raises several difcult (or at least subtle)
conceptual and practical issues, which include the question of whether to use “bottom-up”
versus “top-down” estimates for the costs of crime—an important decision because the two
procedures yield gures that differ by a factor of two. Philip Cook and I argued that the “top-
down” approach is the conceptually correct framework (Cook and Ludwig, 2000), although
this approach raises several measurement challenges that I will discuss here that are in desperate
need of intensive study.
Second, it is worth making explicit a point that has been raised implicitly in Cohen, Piquero,
and Jenning’s article (2010, this issue); the costs (as well as the benets) of crime prevention
might vary across offending trajectories, and so decisions about how to target resources across
offending trajectories need to focus on the ratio of benets to costs and not just focus on the
benet side of the ledger.
Finally, the practical policy implications of combining BCA and trajectory analysis are lim-
ited, as Cohen et al. (2010) have noted, by the difculty of identifying the offending groups of
people prospectively. One suggestion I have is to consider using information about the criminal
involvement of parents, because of previous evidence about strong intergenerational correlations
in offending behavior. But even if we cannot target interventions as well as we would like, the
social costs of crime are so large that American society seems likely to be underinvesting right
now in most forms of crime prevention, with the possible exception of mass incarceration.
Criminology & Public Policy
308
Estimating the Benets of Crime Prevention
The conceptually appropriate way to think about the costs of crime is what Cohen et al. (2010)
have called the “top-down” approach, but which Philip Cook and I preferred to term the “ex
ante” perspective (to be contrasted with the “bottom-up” or “ex post” perspective). The ex ante
perspective corresponds to the resource allocation problem facing policy makers; the mayor of
some large, cold Midwestern city must decide how much of the budget for next year should go
to crime prevention versus other pressing uses, such as schools, roads, public transportation,
snow removal, garbage collection, and homeless shelters. The public good that citizens receive
in exchange for devoting extra resources to crime prevention instead of alternative uses is a re-
duction in the risk that they, or that people they care about, will be victimized in the future. To
compare the value of this benet to the costs, we need to convert these benets to dollar terms,
and the appropriate way to do that is to measure the sum of what people in the community
are willing to pay (WTP) for changes in crime victimization risk.
The problem with the “bottom-up” or “ex post” perspective is that it either does not make
any sense, is not useful for policy purposes, or both. This alternative perspective focuses on try-
ing to value the “cost” of crime that has already occurred to identiable victims. The valuations
of some costs are easy to imagine (the stolen wallet, television, or broken window). But how
does one assign dollar values to nonmarket (intangible) costs such as the pain and suffering
associated with trauma, injury, or death? The ex post method often turns to jury awards, but
that just pushes the conceptual problem back a step; how do juries derive cost gures? One
possibility would be to try to identify the dollar amounts required to make victims whole, or
what economists call the “willingness to accept.” But anyone who has lost a parent, child, or
spouse to crime would say that no amount of money would ever compensate for their loss,
which for BCA purposes, in turn, would imply that we should be devoting every dollar of the
gross domestic product (GDP) to crime prevention (because the benets measured in this way
would be innite). When I teach BCA in my crime policy class at the University of Chicago
Law School and ask how juries come up with victim payments to compensate for intangible
crime costs, most law students respond with “the juries just make it up,” which I suspect comes
close to the truth.
But even after we have settled on the ex ante perspective as the conceptually appropriate way
to dene what we mean by the costs of crime, several difcult measurement challenges remain.
Many studies have tried to estimate WTP for changes in crime risks by looking at data from
housing markets and, specically, looking at what people are willing to pay for houses in safer
neighborhoods. But isolating the effects on house prices of safety versus other hard-to-measure
home and neighborhood attributes is extremely difcult in practice. Moreover, what I am willing
to pay to live in a 10% safer location understates what I would be willing to pay for a new police
program that reduced crime citywide by 10% because I put some value on the improved safety
of other city residents as well. So estimates for the safety/price gradient in the housing market
likely will understate societal WTP for crime control even if we were not concerned about the
possibility of omitted variable bias in our hedonic home-price regressions.
309
Volume 9 • Issue 2
The most common alternative to looking at actual housing market data is to use survey
methods to ask people to respond to hypothetical market scenarios, which is known as the
contingent valuation (CV) approach. But this approach assumes that people have well-formed
preferences for safety and are capable of thinking about marginal changes in crime victimization
risks. It is possible that these assumptions are met because most people do have some rst-hand
experience thinking, in at least a general way, about crime probabilities in deciding where to buy
or rent a place to live, but at the end of the day, who really knows? Environmental economists
have developed a large literature trying to learn more about whether CV “works” in that ap-
plication by, for example, seeing how WTP responses vary by how the questions are phrased,
sequenced, or preceded by the provision of different amounts or types of background informa-
tion and by trying to construct scenarios in which WTP survey responses can be benchmarked
against actual behavior. As far as I know, no similar research program is underway in the area
of crime, even though, in my view, it would have tremendous social value.
Counting Costs as well as Benets
Cohen et al. (2010) have sought to disaggregate the costs of crime across offending trajectories
with the idea of helping policy makers better target crime prevention resources. The authors
briey have alluded to the fact that for targeting resources, we also need to know something
about how the effectiveness of candidate interventions varies across offending groups. Put dif-
ferently, we need to know how the benets and the costs (and the ratio of benets to costs)
of interventions vary across offending trajectories. This seems to me to be a fundamentally
important point worthy of elaboration.
For example, Table 2 in Cohen et al. (2010) shows that the average lifetime costs of crime
by people in the lowest offending group (G2) is $144,996 (or put differently, the benets of
preventing criminal behavior by people in this trajectory), compared with a gure of $1,081,559
for those in the most socially costly group (G4). At one point in the article, Cohen et al. argue
that we should be trying to concentrate resources on the most socially costly offending groups,
but this outcome need not be the case. Suppose, for example, that we have a policy intervention
that is 20 times as effective in changing the behavior for teenagers in the G2 group compared with
those in the G4 group. In that case, it would be more cost effective to devote some incremental
increase in crime-prevention funding to people in the lower offending (G2) group.
Just to be clear, I am not arguing that Cohen et al. (2010) are necessarily wrong in arguing
for the targeting of additional resources to the highest offending trajectories. My only point
is that it is not self-evident. It is true that in the area of education research, many studies have
shown that more disadvantaged children seem to be more responsive to educational interven-
tions (see Currie and Thomas, 1995; Krueger, 1999). At least in principle, this trend need not
be true for crime prevention, or at least it need not be true for all types of crime prevention,
if one considers, for example, selective incapacitation as a possible policy lever or the fact that
criminal behavior by some people might be caused by underlying factors, such as organic brain
pathologies or mental health problems that are difcult to remediate. My main point is that
Criminology & Public Policy
310
we need to be attentive to the empirical possibility that some offending groups might be more
responsive than others to policy interventions, and so, we should be guiding resource allocation
decisions based on the ratio of benets to costs for different uses of crime-prevention resources
rather than focusing just on the benets.
Policy Implications
As Cohen et al. (2010) have noted, one practical difculty in translating trajectory thinking
into concrete policy recommendations is the difculty of identifying prospectively who falls
into which offending trajectories. Although I do not know the trajectory literature well myself,
I wonder if one potentially useful marker would be parental involvement in crime because of
the substantial intergenerational transmission of criminal behavior (for example, Hjalmarsson
and Lindquist, 2007).
For me, the main implication for crime prevention of the cost of crime literature is that
we should be doing a lot more of it. Previous studies have suggested that the costs of crime
in developed countries might be 10% of the GDP or more (Entorf and Spengler, 2002: 91),
which is consistent with estimates that the costs of crime in the United States might be around
$1 to $2 trillion per year (Anderson, 1999; Ludwig, 2006). These costs are so substantial that
even “low-tech” crime-prevention strategies, such as putting more police on the street, seem to
have benet–cost ratios from 4:1 up to 8.5:1 (Donohue and Ludwig, 2007). The benet–cost
ratio for the intensive Perry Preschool early childhood intervention might be as high as 12.5:1
(Beleld, Nores, Barnett, and Schweinhart, 2006), with up to 70% of the dollar value of the
Perry benets coming from reductions in criminal behavior. Even the large-scale Head Start
program seems like it passes a benet–cost test (Ludwig and Phillips, 2007).
Mass incarceration seems to me to be the one exception. As is well known to readers of this
journal, the United States has increased its incarceration rate seven-fold since 1970. Although
I believe that expanding the size of the prison population reduces crime, I also think it is likely
that we must experience diminishing returns to most things, including mass incarceration.
Whether keeping the marginal person imprisoned passes a benet–cost test at the present levels
of incarceration seems to be a close call (Donohue, 2009). But with that said, our current scale
of incarceration seems like an unambiguously bad idea when we recognize that the opportunity
cost of mass imprisonment is foregone spending on more productive uses, such as more policing
or early childhood interventions.
Being able to use trajectory methods to target crime-prevention resources more efciently
would be of potentially great value to public policy makers, assuming that the eld one day be-
comes better able to identify prospectively the offending trajectories of people. In the meantime,
I think Cohen et al. (2010) have added a stimulating discussion of the value of benet–cost
analysis to develop crime policy, and in particular, to improve on our current status quo.
311
Volume 9 • Issue 2
References
Anderson, David A. 1999. The aggregate burden of crime. Journal of Law and Economics, 42:
611–642.
Beleld, Clive R., Milagros Nores, Steve Barnett, and Lawrence Schweinhart. 2006. The
High/Scope Perry Preschool Program: Cost–benet analysis using data from the age 40
follow up. Journal of Human Resources, 41: 162–190.
Cohen, Mark A., Alex R. Piquero, and Wesley G. Jennings. 2010. Studying the costs of
crime across offender trajectories. Criminology & Public Policy. This issue.
Cook, Philip and Jens Ludwig. 2000. Gun Violence: The Real Costs. New York: Oxford
University Press.
Currie, Janet and Duncan Thomas. 1995. Does Head Start make a difference? American
Economic Review, 85: 341–364.
Donohue, John J. 2009. Assessing the relative benets of incarceration: The overall change
over the previous decades and the benets on the margin. Do Prisons Make Us Safer? The
Benefits and Costs of the Prison Boom. Retrieved September 27, 2009 from works.bepress.
com/john_donohue/65.
Donohue, John J. and Jens Ludwig. 2007. More COPS. Brookings Institution Policy Brief
158. Washington, DC: Brookings Institution.
Entorf, Horst and Hannes Spengler. 2002. Crime in Europe: Causes and Consequences. New
York: Springer.
Hjalmarrson, Randi and Matthew Lindquist. 2007. Like Godfather, Like Son. Working Paper.
College Park: University of Maryland, School of Public Policy.
Krueger, Alan B. 1999. Experimental estimates of education production functions. Quarterly
Journal of Economics, 114: 497–532.
Ludwig, Jens. 2006. Testimony to the U.S. Senate Judiciary Committee, September 19,
2006. Retrieved September 27, 2009 from judiciary.senate.gov/hearings/testimony.
cfm?id=2068&wit_id=5749.
Ludwig, Jens and Deborah A. Phillips. 2007. The Benefits and Costs of Head Start. NBER
Working Paper 12973. Cambridge, MA: National Bureau of Economic Research.
Jens Ludwig is McCormick Foundation Professor of Social Service Administration, Law, and
Public Policy at the University of Chicago; the Director of the University of Chicago Crime
Lab; the Co-Director of the National Bureau of Economic Research Working Group on the
Economics of Crime; and the Nonresident Senior Fellow in Economic Studies at the Brook-
ings Institution. His research interests involve urban problems, including crime, education,
housing, and concentrated poverty. In 2006, he was awarded the Association for Public Policy
Analysis and Management’s David Kershaw Prize for distinguished contributions to public
policy by age 40.
