Does the “Community Prosecution”
Strategy Reduce Crime? A Test of
Chicago’s Experience
Thomas J. Miles, University of Chicago Law School and University of
Chicago Crime Lab
Send correspondence to: Thomas J. Miles, University of Chicago Law School, Chicago,
IL, USA. Tel: 773-834-4163; Fax: 773-702-0730; E-mail: [email protected]
A new strategy of criminal prosecution, called “community prosecution,” emerged in
the past two decades. The strategy breaks with the traditional approach to prosecution
in which a prosecutor works in an office adjacent to a criminal court, processes a large
volume of cases, and measures success with conviction rates and sentence lengths. In
community prosecution, a prosecutor works directly in a neighborhood, develops relationships with local groups, aligns enforcement priorities with residents’ public safety
concerns, and seeks solutions to prevent crime. This article presents the first estimates
of community prosecution’s impact on crime. Over a fifteen-year period, Chicago’s
top prosecutor twice applied the community prosecution strategy in some (but not
all) neighborhoods, and this sequence of two “off/on” policy episodes permits plausible identification of the strategy’s impact. Differences-in-differences estimates show
that community prosecution reduced certain categories of crime, such as aggravated
assault, but had no effect on other categories, such as larceny. The diversity of practices under the rubric of community prosecution makes generalization difficult, but the
The author thanks Roseanna Ander, Seth Bour, Adam B. Cox, Steven Jansen, Julius
Lang, Jens O. Ludwig, Lisa Miller, Joy Repella, Max Schanzenbach, Robert Weisberg, an
anonymous referee, and participants at the Conference on Empirical Legal Studies, Law
& Society Annual Meetings, and the National Community Prosecution Conference for
helpful comments, Timothy A. Lavery of the Chicago Police Department for providing
data, and Albert Chang and Christopher Heasley for excellent research assistance.
The author gratefully acknowledges financial support from the SNR Denton Fund at
the University of Chicago Law School.
American Law and Economics Review
doi:10.1093/aler/aht012
Advance Access publication August 14, 2013
c The Author 2013. Published by Oxford University Press on behalf of the American Law and Economics
Association. All rights reserved. For permissions, please e-mail: [email protected].
117
118
American Law and Economics Review V16 N1 2014 (117–143)
estimates from Chicago show that the strategy has the potential to produce costjustified reductions in crime. (JEL: K14, K42)
1. Introduction
The traditional view of a criminal prosecutor is a legal advocate who seeks convictions and the imposition of penalties. This prosecutor operates in close proximity to the
courtroom, usually arguing in court or negotiating a plea. Faced with a large volume
of potential cases from across the jurisdiction, the prosecutor carefully chooses which
are worthy of her effort and screens out cases with inadequate evidence or less serious
violations. The usual metrics of her success are the rates of declination and conviction
and the length of sentences.
In the past two decades, an alternative strategy of prosecution, called community
prosecution, has emerged, and some commentators see it as “a paradigm shift in the
area of criminal justice” (Gray, 2008) and “a revolution in crime fighting” (Doolan,
2002). The alternative approach requires individual prosecutors to take responsibility
for specific geographic target areas. It typically involves deploying prosecutors directly
in the community, identifying the community’s public safety concerns by developing
close relationships with neighborhood groups, and employing solutions that prevent
crime ex ante rather than relying solely on prosecutions to punish it ex post.
Some prosecutors have been quick to apply the community prosecution approach.
Attorney General Eric Holder was an early adopter of the strategy when he was U.S.
Attorney for the District of Columbia in the 1990s (Coles, 2000, p. 31). In recent
years, the Department of Justice has given dozens of grants to local prosecutors’
offices to plan, implement, and improve community prosecution programs (Goldkamp
et al., 2002). According to one commentator, “community prosecution is going to be
part of the future trend in prosecution in the US” (Gramckow, 1997, p. 22). These
developments parallel a growing interest among legal academics in the possibilities
for devolving authority in the criminal justice system to local officials, including but
not limited to prosecutors, and in improving the responsiveness of decision-makers
to local constituencies. Among the aspirations of greater local control is a reduction
in the frequency and racial disproportion of incarceration (Lanni, 2005; Stuntz, 2011;
Bibas, 2012).
Despite its growing popularity, little is known about the effects of community
prosecution. Existing studies have focused on the manner of its implementation with
particular attention to organizational changes within the prosecutors’ office.1 There
1. Goldkamp et al. (2002, 2003) enumerate community prosecution efforts across
the country and catalog their leading features, such as program content, case-processing
Community Prosecution
119
is no study of its impact on crime. Indeed, the keen need for empirical evidence of
its impact on public safety is a common refrain among both advocates and critics of
community prosecution.2 The absence of empirical evaluation creates a risk that the
enthusiasm for community prosecution will lead to costly changes in prosecutorial
practice without providing any social benefits.
This paper fills that gap by presenting the first empirical evidence of the impact of
community prosecution on crime. It exploits a unique source of variation in the application of community prosecution in Chicago, Illinois. The State’s Attorney in Cook
County, Illinois (the county in which Chicago is located), opened several community
prosecution offices in the late 1990s and early 2000s. Four of these offices had territory falling within Chicago, while other parts of the city received no exposure to
the community prosecution approach. A budget crisis in early 2007 forced the abrupt
closure of these offices. For more than two years, the county’s community prosecution
effort ceased. In 2009, the State’s Attorney re-launched the program, and over the next
three years, the four community prosecution offices were re-opened.
This sequence of events lends itself to a differences-in-differences analysis of the
impact of community prosecution on crime. Unlike most differences-in-differences
analyses that rely on a pre/post-comparison from a single variation in policy, Chicago
experienced two “off/on” episodes in succession. This sequence of events provides
the opportunity to observe two applications of the policy for (nearly) every targeted
neighborhood. Moreover, the first and third policy transitions during the study period
(from “off ” to “on” and from “off again” to “on again,” respectively) featured different
offices opening and re-opening at different dates. The second policy transition (from
“on” to “off ”) was a response to a plausibly exogenous shock to the prosecutorial
budget.
adaptations, and collaborating partners. Boland (1998) and Coles and Kelling (1998) provide in-depth case studies of how particular prosecutors’ offices developed and implemented community prosecution strategies.
2. Gramckow (1997, p. 20) states, “There exists, however, no systemic assessment of a community oriented approach by a prosecutor’s office.” Similarly, Gray (2008,
pp. 209–10, internal citation omitted) observes, “While prosecutors that are involved in
community lawyering believe that it is working, little hard data has been collected that
clearly proves community prosecution’s effectiveness in preventing and reducing crime.”
Doolan (2002, p. 572, internal citation omitted) describes it as “[T]here is no formal
research that conclusively proves that community prosecution enhances public safety and
improves quality of life.” Finally, Forst (2000, p. 140) recommends, “If prosecutors really
wish to serve the community, they will subject their new programs to objective assessment and more systematic reporting.”
120
American Law and Economics Review V16 N1 2014 (117–143)
These policy changes are studied using a panel comprising annual observations
of police beats over the period 1995 to 2011. To preview the main results, community
prosecution appears to reduce certain categories of crime: murder, rape, and aggravated
assault. Estimated declines for burglary and motor vehicle theft are socially meaningful but at the cusp of statistical significance. Other observed reductions, such as for
robbery and disorderly conduct, are too imprecisely estimated to support firm statistical inferences. The impacts also vary widely by office, suggesting that community
prosecution’s effectiveness may depend on the manner of implementation.
The results imply that, on average, community prosecution as practiced in Cook
County is cost-justified. Given the range of practices under the rubric of community prosecution, the paper’s analysis does not speak to whether the results observed
here generalize or could be replicated elsewhere. But the paper’s estimates are the
first evidence that community prosecution can, under circumstances which are not
yet fully understood, lead to cost-justified reductions in crime. The paper’s analysis
also contributes to the larger literature on prosecutorial behavior. It represents the first
examination of whether prosecutorial decision-making affects crime rates and the first
cost–benefit analysis of prosecutorial behavior.
The paper proceeds as follows. Section 2.1 describes in more detail the elements of
the community prosecution approach generally, and Section 2.2 turns to the details of
Chicago’s experience with community prosecution. Section 3 describes the data analyzed and the empirical strategy employed. Section 4 presents the results, and Section 5
concludes with a discussion of the implications and directions for future work.
2. Community Prosecution
2.1. Defined
Community prosecution encompasses a wide variety of approaches, making an
authoritative definition elusive. A touchstone of community prosecution is its contrast with the traditional model of prosecution under which mostly reactive prosecutors
devote their energies to the use of adversarial legal process (Jacoby, 1980). In the traditional approach, which is sometimes called the “case processor” approach, a prosecutor
picks which cases to pursue from a portfolio of arrests and investigations assembled by
police. The prosecutor is resource-constrained in that she cannot prosecute all violations of the law that police have detected, and she must choose a small subset for prosecution. Prosecutorial discretion in this setting consists of weighing factors, such as the
seriousness of the offense, the sufficiency of the evidence, and the equal treatment of
similar defendants, in deciding whether to prosecute, whether to offer a plea bargain,
Community Prosecution
121
and how generous that bargain should be. The success of traditional prosecutors is usually assessed with metrics such as the rates of cases pursued, percentage of convictions
obtained, and average sentence lengths. Indeed, most of the empirical literature on
prosecutors has focused on these outcomes (Coffee et al., 1984; Boylan, 2005). The
economic model of prosecutors, which imagines expected sentence lengths as the
objective of their constrained optimization, draws heavily upon the traditional account
(Landes, 1971; Reignanum, 1998).
Community prosecution contrasts with the traditional approach in that it seeks
to involve neighborhood residents in identifying local public safety priorities and in
developing and implementing strategies to address them. It also emphasizes efforts to
prevent crime ex ante rather than relying exclusively on ex post punishments. The
community prosecution strategy does not offer a fixed set of programmatic initiatives or organizational and managerial changes for securing community engagement.
Instead, it is “an evolving strategy” (Coles, 2000) that permits tailoring to specific
contexts. “The different models of community prosecution efforts ... make it difficult to describe what community prosecution actually means and what it looks like”
(Gramckow, 1997, p. 9). Some commentators see this flexibility as a basis for criticism; that community prosecution is without “a coherent vision” and is “lacking in
structure, even chaotic” (Thompson, 2002, pp. 324, 360).3 Also, this flexibility makes
the evaluation of “community prosecution” as a strategy difficult, because the combination of policies it represents likely varies from application to application. Any
estimates of community prosecution’s impact, however well identified, may not have
external validity.
Despite these limitations, several functionally and geographically specific changes
to prosecutorial work are paradigmatic of community prosecution. The first element
is the definition of a target area, the set of neighborhoods or a geographic territory
that are the subject of the change in prosecutorial strategy. One or more prosecutors
are often designated as the “community prosecutors” for the targeted area, and their
workload is shifted to consist largely or exclusively of matters arising from the targeted
3. Attempts to enumerate the ways in which community prosecution breaks with
the traditional prosecution model have produced different tallies of its key elements
(Gramckow, 1997 (three); Coles, 2000 (four); Goldkamp et al., 2002 (seven elements);
Gray, 2008 (six)). The difficulty in defining community prosecution also makes measuring the extent of its use challenging. Nugent-Borakove (2004a, 2004b) surveyed prosecutors’ offices to learn the frequency with which they engaged in six “core elements”
of community prosecution. She found that nearly all offices adopted more than one but
fewer than five elements. Cunningham et al. (2006) analyzed the incidence of community
prosecution using the 2001 National Survey of Prosecutors. They found that the number
of forms of community engagement a prosecutor’s office undertook increased with the
size of the prosecutor’s office and with other measures of specialization in the office.
122
American Law and Economics Review V16 N1 2014 (117–143)
neighborhood. Often, the community prosecutor is located in an office within the target
neighborhood rather than in the central courthouse. The purpose of a local office is to
facilitate communication with the community and even to permit residents to walk in
and voice concerns directly to prosecutors.
The community prosecutor learns the neighborhood’s priorities and enlists its support primarily by developing relationships with various neighborhood associations,
such as business, school, and resident groups. Representatives of these organizations
are sometimes assembled into a formal advisory board, and the board may also include
representatives from other governmental units, such as the police and social welfare
agencies. While the community prosecutor receives information and advice from the
board, she retains final decision-making authority. When a community prosecutor
selects a case for prosecution, she may take responsibility for shepherding it through
all steps of the criminal process, from indictment to trial and sentencing, rather than
specializing in handling many cases at a single stage of the criminal process.
Community prosecution’s emphasis on preventing rather than punishing crime typically implies that a community prosecutor often attempts to address broader causes
of offending rather than relying exclusively on criminal prosecutions to reduce crime.
Examples are nuisance suits and civil enforcement of local health and safety ordinances in order to close drug houses. A community prosecutor may attend or even
help design youth intervention programs. Occasionally, she may mediate disputes in
the neighborhood. Frequently, she employs her knowledge of and access to other agencies to marshal resources and coordinate governmental responses to problems identified by the community.
Advocates of community prosecution generally leave undefined the goals that the
restructuring of prosecutorial work is intended to accomplish.4 The conventional metrics for evaluating prosecutors are qualitative, such as fidelity to law, fairness and good
judgment in selecting cases, zeal and acumen in pressing cases, and decency in extending leniency. Yet, community prosecution’s emphasis on public safety makes crime
rates plausible metrics of its performance.
Community prosecution, like community policing, originated in the “broken windows” theory of Wilson and Kelling (1982), a theory that sought to explain the
increases in crime rates during the 1960s and 1970s (Harp et al., 2004; Wolf, 2010;
Klitzing, 2012). According to the “broken windows” theory, visible signs of public
disorder, such as unrepaired windows and graffiti, and unchecked incidents of petty
4. Community prosecution may also provide publicity benefits for the lead prosecutor who is usually an elected official. But this is unlikely to be the sole or main objective. Community engagement may increase the risk of unfavorable publicity if it raises
expectations that cannot be met.
Community Prosecution
123
crime, such as loitering and vandalism, communicate the message that violations of the
law occur with impunity. The disordered environment leads to more serious offending.
The theory translated into at least two specific policy proposals: community policing
and community prosecution. To date, cities have been quicker to adopt some form
of community policing.5 Most prominently, New York adopted “order maintenance”
policing strategies that emphasized “zero tolerance” of certain misdemeanor offenses
such as subway fare-jumping. The “broken windows” theory and the policing policies it inspired are now contested in a vigorous academic debate.6 “Broken windows”
thus offers one potential mechanism through which community prosecution may affect
crime rates: tamping down on quality-of-life offenses may prevent petty crimes from
escalating into serious ones.
Yet, there are other potential mechanisms through which community prosecution
could influence crime. From an economic perspective, community prosecution may
facilitate the acquisition of more accurate information and thus allow a more effective
allocation of prosecutorial resources. It may do this in several ways. The placement of
a prosecutor, who was previously located in a remote courthouse, in the neighborhood
may lower a resident’s cost of providing information. Residents may perceive prosecutors as more trustworthy than police and thus may be more forthcoming. Residents and
even local beat officers may have greater confidence that a prosecutor focused on the
neighborhood rather than mass-processing a large volume of cases is more likely to act
upon any information provided. Several types of information may be provided, such as
residents’ preferences for enforcement or problem-solving efforts, the types of crimes
5. By 1999, 64% of local police departments had full-time officers engaged in
some form of community policing (Hickman and Reaves, 2001). Numerous scholars
have noted the connections between community policing and community prosecution.
For example, Stuntz (2011, p. 305) wrote, “Gradually, something akin to the community
policing movement has begun in the realm of prosecution. ‘Community prosecution’
has not transformed the work of prosecutors’ offices, as community policing has done
for many urban police forces. But the core idea – more neighborhood democracy, and a
more politically responsive style of law enforcement – is the same.” Also, Wright (2012,
p. 361) observed, “Some efforts to strengthen the connection between prosecutors and
the public use the rubric of ‘community prosecution.’ These initiatives draw on general
concepts developed in the now-mature ‘community policy’ movement.”
6. Several studies have contested the connection between order maintenance policing and lower crime rates. Levitt (2004) pointed out that crime in New York began its
decline before the shift in policies. Harcourt and Ludwig (2006) contested that Kelling
and Sousa’s (2001) claim that New York’s aggressive misdemeanor arrest policies. They
argue that it’s mid-1990s crime drop is more properly interpreted as mean reversion following the crack cocaine epidemic a decade earlier. Harcourt and Ludwig (2007) found
no relationship between misdemeanor marijuana arrests and serious crime in New York.
Finally, Fagan and Davies (2000) demonstrated that the “stop and frisk” practice, which
was part of New York’s order maintenance strategy, had a disparate racial impact.
124
American Law and Economics Review V16 N1 2014 (117–143)
that are particularly disruptive to the neighborhood, and the identities of individual
offenders who pose the greatest danger.
Community prosecution is also a form of decentralization that gives individual
prosecutors leeway to experiment with different solutions in an entrepreneurial manner. Having acquired new information, the prosecutor may use it in several ways. She
might use it to marshal governmental and community resources to change the neighborhood conditions in ways that reduce the risk of crime. Or, she may target resources
against particular offenders to ensure their incarceration. Each of these strategies might
reduce serious forms of crime immediately without affecting the incidence of qualityof-life crimes. These strategies require the prosecutor to draw upon interpersonal and
leadership skills that might go unused in more traditional prosecutorial work. Community prosecution may, therefore, make better use of the types of human capital that
a jurisdiction’s prosecutors possess. Despite its origins in “broken windows” theory,
community prosecution may affect crime through numerous other mechanisms.
2.2. Community Prosecution in Chicago
The primary criminal prosecutor in Chicago is the State’s Attorney for Cook
County, Illinois, who employs about 860 lawyers, constituting the second largest prosecutor’s office in the nation. In the late 1990s, the State’s Attorney embraced the community prosecution approach and over the next six years, opened offices (calling them
“community justice centers”) in five target areas. Each office had responsibility for
handling cases within a specific geographic territory covering several police districts.7
Chicago has twenty-five police districts, and each police district, in turn, comprises
an average of eleven police beats. Figure 1 presents a map with the boundaries of
Chicago’s police districts, and districts within the target area of a community prosecution effort at any time during 1995–2011 are shaded.
Each community prosecution office was assigned two experienced prosecutors,
who exclusively handled matters concerning the target community rather than taking cases from across the county. The more junior of the two prosecutors served an
eighteen-month rotation in the community prosecution office and focused mainly on
misdemeanor cases and lesser felonies. The more senior prosecutor handled serious
felony cases and was permanently assigned to the community prosecution office.
Each office had a steering committee consisting of business, religious, advocacy,
and citizen groups that met monthly and advised prosecutors of the community’s
public safety concerns, but the prosecutors retained final decision-making authority. The work of the community prosecutors was substantially different from that of
7.
The material in this paragraph is drawn from Repella (2010).
Community Prosecution
125
Figure 1. Police Districts and Community Prosecution Coverage in Chicago, Illinois. Note: Police districts that received community prosecution coverage at any
time during 1995–2011 are shaded.
traditional, case-processing prosecutors. They devoted part of their time to forms of
crime prevention and problem-solving that did not involve prosecutions, such as mediating disputes between neighbors and participating in school intervention programs.
When bringing prosecutions, they selected very specific cases. Each week, they
combed police lists of arrests occurring in the neighborhood and targeted repeated
offenders with the aim of incapacitating the small number of offenders responsible for
many crimes. In choosing cases, they placed greater weight on a suspect’s history of
arrests than a traditional prosecutor might. When not looking to arrestee lists, they
took input from the steering committee and tried to identify cases that would have a
large impact on quality-of-life in the community. Sometimes, this objective led them to
126
American Law and Economics Review V16 N1 2014 (117–143)
target serious offenses, such a repeat armed robbers, and at other times, minor offenses.
In bringing a case, the community prosecutor coordinated with police and informed
the judge during an initial hearing that the case was a specifically selected community
prosecution. In so doing, the prosecutor may have elicited closer attention from the
police and court for the case. Each community prosecution office produces annually
an internal report for the State’s Attorney detailing the specific activities and community engagement efforts undertaken and describing the major prosecutions that were
brought and how they were resolved. Cook County’s approach thus included all of
the paradigmatic elements of community prosecution: targeting specific geographic
areas, engaging the community, responding to the problems identified, collaborating
with other agencies, altering prosecutorial priorities, and restructuring the work of
prosecutors.
The community prosecution effort continued in Cook County until early 2007 when
a budget crisis caused termination of the program, including closure of all neighborhood offices. For more than two years, no community prosecution efforts occurred.
In 2009, a new State’s Attorney who had campaigned partly on a promise to reinvigorate the community prosecution strategy took office (Abernathy, 2008). During
the next three years, she re-opened the four community prosecution offices having
target areas within the city of Chicago.8 Table 1 shows the locations, target police districts, and dates of each community prosecution office. By December 2011, 38.7% of
Chicago police beats fell within a community prosecution target area, and 37.2% of
Chicago’s population resided in a police beat that was part of a community prosecution
target area.
3. Data
The data analyzed in the paper are a panel of Chicago’s police beats covering the
years 1995–2011. Information on crime rates was gathered from two sources. The city
of Chicago reports information on all criminal incidents since 2001 (City of Chicago,
2011), and these criminal incidents were aggregated into annual, beat-level tallies. For
the years 1995–2000, the author obtained beat-level data on crimes from the Chicago
Police Department.
In the FBI’s Uniform Crime Reports, there are three major groupings of offenses:
violent crimes consisting of murder, rape, aggravated assault, and robbery, and property crimes consisting of burglary, larceny, and motor vehicle theft; the third category, known as part II offenses, consists of less serious offenses, and they include
8. A fifth office in Maywood, Illinois, was not re-opened, but its target area did
not fall within the city of Chicago.
Table 1. Openings and Closings of Community Prosecution Offices in Cook County, Illinois
Date of change in community prosecution coverage
Chicago police districts covered
North side (5333 N. Western Ave., Chicago)
South side (9059 S. Cottage Grove Ave., Chicago)
West side (4 W. Chicago Ave., Oak Park)
Expansion in coverage of west-side office
Central (943 S. Maxwell Ave., Chicago)
Maywood (S. 5th Ave., Maywood)
19, 20, 23
4, 6, 22
15
25
11, 12
–
Source: Repella (2010) and author’s review of media reports.
Opened
Closed
September 1998
May 1999
September 1999
–
August 2000
July 2003
March 2007
March 2007
March 2007
–
March 2007
March 2007
Re-opened
May 2009
August 2009
July 2010
July 2011
August 2011
–
Community Prosecution
Location of community prosecution office
127
128
American Law and Economics Review V16 N1 2014 (117–143)
Table 2. Differences between Chicago Police Beats Receiving Community
Prosecution Coverage and Not: Means and Standard Deviations (in parentheses)
Was police beat ever covered by
community prosecution office?
Yes
No
(1)
Log Violent crime rate
Log Property crime rate
Log Part II crime rate
Population
Percent black
Percent Hispanic
Percent over age 18
Median income
N
7.037 (0.853)
8.405 (0.485)
8.731 (0.754)
12,574.30 (5,314.69)
0.442 (0.418)
0.248 (0.308)
0.876 (0.136)
43,807.69 (16,748.88)
1,836
(2)
6.823 (0.952)
8.349 (0.607)
8.556 (0.742)
14,590.16 (6,806.16)
0.316 (0.381)
0.306 (0.306)
0.874 (0.137)
43,560.58 (17,270.51)
2,907
Note: Summary statistics calculations are weighted by the population of each police beat.
non-aggravated assault, arson, vandalism, weapons offenses, drug offenses, prostitution and sex offenses (other than rape), crimes against family members and children,
and disorderly conduct. These offenses are categorized as less serious than violent and
property crime, and they are less often studied. But, they are examined here because an
important justification of community prosecution is the prediction from the “broken
windows” theory that enforcement of minor offenses prevents the spread of more serious crime.
Information on the population and demographic characteristics of the beats were
taken from the decennial censuses of 2000 and 2010 and were linearly interpolated to
obtain annual values. A small number of police beats had very low populations, such
as those encompassing O’Hare Airport, making their crime rates exceedingly noisy.
These beats were removed, leaving 279 beats in the sample. Consequently, the main
data set was an annual panel of 279 police beats from 1995 through 2011, which had
4,743 (= 279 beats × 17 years) observations.
Table 2 shows means and standard deviations of the main variables stratified by
whether the police beat ever fell within a community prosecution target area. The two
groups of police beats are similar along some dimensions but differ across others.
Their median incomes and proportion of their populations that were adults are nearly
identical. They contrast in their population sizes and racial and ethnic compositions.
The populations of police beats that eventually fell within a community prosecution
target area were smaller by about 14%. The share of their populations that were black
was higher by 13 percentage points and the share that was Hispanic was lower by
5 percentage points.
Community Prosecution
129
The State’s Attorney did not articulate the criteria for selecting target neighborhoods, and it seems unlikely that randomization was the selection device. A cynical
possibility is that the State’s Attorney attempted to assure an appearance of the initiative’s success by choosing neighborhoods where crime was already lower. The means
in Table 2 cast doubt on this possibility. The police beats receiving community prosecution coverage had slightly higher crime rates than other police beats. These beats had
3% more (log) violent crime, 1% more (log) property crime, and 2% more (log) part
II offenses per capita than other beats. The means in Table 2 show that the community
prosecution was not systematically applied to police beats where crime was already
lower. Instead, the average beat chosen for the strategy had slightly more severe crime
problems.
4. Estimation Strategy
To control for the multiple influences on crime rates, regressions were estimated
using ordinary least squares. Each dependent variable took the form ln Cit , the (logged)
rate of a given offense category in beat i and year t. The main equation for the impact
of community prosecution was
ln Cit = (Community Prosecution Openit )δ D D + αi + αm + αt
+ θi (trendi ) + βXit + εit ,
(1)
where the terms αi and αt are fixed effects for beat i and year t, respectively. The term
θi captures the influence of a beat-specific time trend, although specifications without
this term are also reported. The vector Xit contains a set of beat- and year-specific
control variables that are commonly included in crime analyses: the percent of the
population that is black, Hispanic, and aged over 18; the percent of female-headed
households; and (the log of) median household income. All of the variables in Xit are
calculated from linearly interpolated measures from the decennial censuses. For beats
that had no crimes in a given year, the log crime rate is set to zero, and the observation
is permitted to have a separate intercept (Pakes and Griliches, 1980).
The indicator variable Community Prosecution Openit takes the value of 1 when
beat i falls within the target area of community prosecution office that is active in year
t. In years when the community prosecution effort was commenced or terminated, its
value is the fraction of months during which the neighborhood office was open, and
otherwise, its value is 0. The coefficient δ D D is the differences-in-differences estimate.
In later specifications, more elaborate functions are used to measure the community
prosecution policy, such as separate indicators for each community prosecution office.
130
American Law and Economics Review V16 N1 2014 (117–143)
The error term is εit . Bertrand et al. (2004) showed that when outcomes are serially
correlated, standard errors from differences-in-differences estimators may be inconsistent. In addition, the community prosecution strategy affects multiple police beats,
raising the possibility that common group errors may bias estimated standard errors
downward (Moulton, 1990). Accordingly, two sets of standard errors, one clustered
at the level of police beats and the other clustered at the level of police districts, are
reported for each regression coefficient.
5. Results
5.1. Baseline Estimates
Table 3 reports differences-in-differences estimates for the major categories of
offenses: violent, property, and part II. Each cell of the table contains an estimated
δ D D coefficient from a separate regression. Each row reports estimates for a different
dependent variable, and the columns vary the right-hand side specification, progressively including a richer set of controls. The estimates in column (1) include only fixed
effects for years and for police beats. The fixed effects for years control for citywide
trends in crime rates, and the fixed effects for police beats control for unobserved beat
characteristics that are fixed over time. To control better for observable differences
across police beats, the regressions in columns (2) include the set of demographic and
income characteristics of beats that are shown in Table 4 and that prior literature has
shown to correlate with crime.9
The estimates in columns (1) and (2) may suffer from omitted variable bias if crimefighting policies or criminological conditions changed contemporaneously with the
implementation of community prosecution.10 To address this, the regression in column (3) introduces linear time trends for each police beat, which eliminates variation
in crime rates caused by factors that vary linearly over time and that are specific to
individual police beats. Identification of the impact of community prosecution in these
9. Specifically, the control variables included are those reported in Table 4: the
median income in each police beat, the percentage of female-headed households in the
beat, and the percentage of the population in the police beat that is black, Hispanic, and
under age 18.
10. For example, if police deployments correlated with community prosecution,
the estimates would reflect the combined effect of police and prosecutor behavior.
Unfortunately, the city refuses to release information on police deployments at the district level (let alone the beat level), citing risks to police operations and officer safety
(Miahalpoulos, 2011).
Community Prosecution
131
Table 3. Impact of Community Prosecution on Rates of Major Offenses:
Differences-in-Differences Estimates
Dependent variable:
(log) rate of offense type
(1)
(2)
Violent
−0.0388
(0.0196)∗∗
[0.0404]
−0.0321
(0.0194)∗
[0.0327]
−0.0676
(0.0163)∗∗
[0.0281]∗∗
Property
0.0167
(0.0166)
[0.0311]
0.0038
(0.0164)
[0.0288]
−0.0330
(0.0134)∗∗
[0.0179]
Part II
0.0454
(0.0222)∗∗
[0.0520]
0.0450
(0.0208)∗∗
[0.0445]
0.0179
(0.0153)
[0.0197]
Y
Y
Y
Demographic and income Controls?
Beat × linear time trend?
(3)
Note: Ordinary least squares coefficients with standard errors in parentheses and brackets. Standard errors
in parentheses are clustered by police beat, and those in brackets are clustered by police districts. All regressions are weighted by the population of each police beat.
equations comes from within-beat variation in community prosecution after netting out
beat-specific linear trends.
Table 3 shows that the crime categories vary in their sensitivity to the regression specification. For all of the categories, clustering standard errors at the district
level implies substantially less precision in the estimates, but as more control variables are added to the regressions, the standard errors tighten. Also, the inclusion of
beat-specific trends moves all of the point estimates in a negative direction. For violent crime, all of the estimated coefficients have negative signs, and their magnitudes
imply that community prosecution produced socially meaningful drops in this category
of crime. When the regression includes beat-specific time trends, the point estimate is
nearly −0.07 and statistically significant even when the standard errors are clustered
by police districts.
The estimates for property crimes move in a progressively negative direction as the
regression specification becomes more demanding. In the parsimonious specification
of column (1), the estimate for property crimes is 0.017, but the inclusion of controls
for demographics and income levels lowers it to almost 0. When beat-specific trends
are added to the regression in column (3), the estimate falls to −0.033. In contrast, the
estimated coefficients for part II offenses remain positive. Yet, even here, the addition
of beat-level trends to the regression causes the estimates to fall from 0.045 to 0.018.
The fact that the estimates for all three crime groupings are more negative (or closer
to 0) in the presence of beat-specific trends suggests that crime trends moved in the
132
American Law and Economics Review V16 N1 2014 (117–143)
Table 4. Impact of Community Prosecution on Rates of Major Offenses: Alternate Specifications of Community Prosecution
Dependent variable: log rate of offense type
Panel A
Any active community prosecution
office
Police beat on border of any active
community prosecution office
Panel B
North side community prosecution
office active
South side community prosecution
office active
West side community prosecution
office active
Central community prosecution
office active
Panel C
Active community prosecution
office during the first episode
Active community prosecution
office during the second episode
Violent
Property
Part II
(1)
(2)
(3)
−0.0752
(0.0171)∗∗
[0.0286]∗∗
−0.0397
(0.0143)∗∗
[0.0172]∗∗
0.0179
(0.0164)
[0.0224]
−0.0307
(0.0220)
[0.0219]
−0.0271
(0.0190)
[0.0221]
−0.0001
(0.0181)
[0.0225]
−0.0072
(0.0288)
[0.0513]
−0.0945
(0.0232)∗∗
[0.0261]∗∗
−0.1115
(0.0287)∗∗
[0.0558]∗∗
−0.0872
(0.0261)∗∗
[0.0364]∗∗
−0.0449
(0.0158)∗∗
[0.0182]∗∗
−0.0250
(0.0190)
[0.0239]
−0.0905
(0.0239)∗∗
[0.0307]∗∗
0.0058
(0.0291)
[0.0205]
−0.0116
(0.0237)
[0.0253]
0.0277
(0.0159)∗
[0.0266]
0.0801
(0.0313)∗∗
[0.0268]∗∗
0.0089
(0.0395)
[0.0182]
−0.0693
(0.0167)∗∗
[0.0282]∗∗
−0.0356
(0.0373)
[0.0788]
−0.0332
(0.0139)∗∗
[0.0188]∗
−0.0297
(0.0207)
[0.0298]
0.0187
(0.0160)
[0.0211]
0.0058
(0.0247)
[0.0312]
Note: Ordinary least squares coefficients with standard errors in parentheses and brackets. Standard errors
in parentheses are clustered by police beat, and those in brackets are clustered by police districts. All regressions are weighted by the population of each police beat. Coefficients are from regressions with the same
specification as those in column (3) in Table 3 and are weighted by the population of the police beat.
opposite direction to the activation of community prosecution offices. Taken at face
value, these estimates are consistent with the hypothesis that community prosecution
may modestly increase the reporting of quality-of-life crimes. In sum, Table 3 shows
that when the analysis accounts for trends within each police beat, community prosecution appears to reduce violent crime and perhaps property crime and that any increase
in part II offenses is small.
Community Prosecution
133
These estimated responses to community prosecution raise questions about the
mechanism through which it affects crime. The “broken windows” theory predicts
that if unchecked, visual clues will trigger an escalation in the frequency and seriousness of offending. If the community prosecution strategy operates through a “broken
windows” mechanism, quality-of-life crimes should decline prior to the drops in more
serious offending. The estimates in Table 3 are precisely opposite of this prediction.
Violent crimes, which are commonly thought to be the most serious, experience the
largest and statistically significant declines following the introduction of the community prosecution strategy. The estimated drop in property crimes is smaller in magnitude and statistically insignificant, and quality-of-life crimes, as measured by part II
offenses, experienced a statistically insignificant increase.
A second prediction from “broken windows” theory is that any effects on crime
may take time to manifest. Small disorders lead to larger ones and ultimately to serious crimes. The effect of community prosecution should, therefore, grow over time.
To test this prediction, the variable for active community prosecution coverage was
decomposed into a series of binary variables for the number of years, since the strategy
was implemented in a police beat. In addition, indicator variables for the years prior
to the application of community prosecution strategy were included to assess the time
path of crime rates in such beats immediately before the implementation of the strategy. Figure 2 shows these coefficients and their 95% confidence intervals for the three
major crime categories.11 When the community prosecution variable is decomposed
in this way, none of the estimates are statistically significant, and thus, these results
provide impressionistic evidence of the dynamics of community prosecution.
Again, the results are contrary to the “broken windows” prediction. Figure 2a and b
show that the declines in violent and property crime are immediate and stable. Within
the first year of introducing the strategy, the point estimates for violent and property crime are each about −0.04, where they remain for the next four or five years.
Even the modest rise in part II offenses, occurring during the third and fourth years
after the implementation of community prosecution, appears transitory. By the sixth
year, the point estimate for part II offenses has returned to its initial level. The patterns
in Figure 2 suggest that community prosecution as implemented in Cook County did
not influence crime through the mechanisms hypothesized by “broken windows.”
Figure 2 also shows whether the police beats where community prosecution was
introduced had sharply different crime rates in the three years prior to its implementation than other police beats. Consistent with Table 2, average rates of violent and part
11. The 95% confidence intervals were calculated with standard errors clustered
on police districts.
134
American Law and Economics Review V16 N1 2014 (117–143)
0.4
(a)
0.3
0.2
0.1
0
-3
-2
-1
-0.1
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
0
1
2
3
4
5
6
7
8
-0.2
-0.3
-0.4
(b)
0.4
0.3
0.2
0.1
0
-3
-2
-1
-0.1
-0.2
-0.3
-0.4
(c)
0.4
0.3
0.2
0.1
0
-3
-2
-1
-0.1
-0.2
-0.3
-0.4
Figure 2. Dynamic Response to Community Prosecution. (a) Violent Crime; (b)
Property Crime; (c) Part II Offenses. Notes: Figures show estimated regression coefficients and 95% confidence intervals (on vertical axis) of the impact of community
prosecution by the year since the implementation of community prosecution (on
horizontal axis) in a police beat. Coefficients are from regressions with the same
specification as those in column (3) in Table 3 and are weighted by the population
of the police beat. Standard errors are clustered by police district.
Community Prosecution
135
II crimes were slightly higher during these years in beats where community prosecution was implemented. But these differences are modest, and importantly, the estimates
alleviate some standard concerns about endogeneity, such as that the State’s Attorney
selected for the strategy beats where crime was already declining. No distinct trends
appear in community prosecution beats before the strategy was implemented.
5.2. Alternative Specifications
To probe more deeply the relationship between crime rates and community prosecution, Table 4 presents three alternative specifications. Panel A considers the possibility that community prosecution may merely displace crime rather than reduce it. If
displacement was the impact of the strategy, criminal activity might relocate beyond
the boundary of a community prosecution district. The regression in Panel A tests
this prediction by including an indicator variable for each police beat that was adjacent to an active community prosecution district. If displacement were the cause of the
observed declines in crime, the coefficient for a police beat adjacent to an active target
area should be positive. Panel A shows that for every crime grouping, it is negative
and statistically insignificant. For violent and property crimes, the estimated coefficient is about −0.03, and for part II crimes, it is very close to 0. The estimates for
an active community prosecution effort are robust to the inclusion of this border variable. In fact, the declines in violent and property crime appear slightly larger and more
precisely estimated in the presence of the border variable. These results suggest that
displacement was not primarily responsible for the drops in crime where community
prosecution was implemented.
Panel B of Table 4 reports coefficients from regressions in which the indicator for
an active community prosecution office was replaced with separate indicators for each
of the four offices with target areas within the city of Chicago. The consistency of the
estimates for violent and property crime is remarkable. Three of the four estimates
for violent crimes fall between −0.087 and −0.112, and three of the four for property
crimes lie within −0.025 and −0.091. A conclusion from these results is that the presence of an active community prosecution office most often produced sizable declines
in rates of violent and property crime.
Yet, the estimates also show that within each crime category, the impact of community prosecution varied by as much as 11 percentage points across the offices.
For example, each office’s estimated impact on violent crime was negative, but the
point estimates range from −0.007 (on the north side) to −0.112 (on the west side).
Similarly, the estimates for part II offenses ranged from 0.080 (on the west side) to
−0.012 (on the north side). There are numerous potential explanations for the variation in the estimated impacts across offices, and these data do not permit direct tests of
136
American Law and Economics Review V16 N1 2014 (117–143)
them. Particular offices may have prioritized different categories of offenses, perhaps
in response to specific community preferences. Individual prosecutors in the offices
may have exerted greater efforts, either generally or with respect to particular types of
offenses, or they may have possessed greater talent for eliciting community support.
The effect of community prosecution may be heterogenous, varying with unobserved
neighborhood characteristics such as the engagement of community groups.
The results in Panel B also alleviate concerns about the endogeneity of the community prosecution strategy. If the State’s Attorney prioritized neighborhoods where
community prosecution was most likely to succeed or where crime was already declining, then offices that were opened (and re-opened after 2007) earlier should have experienced the largest crime declines. Panel B shows that, if anything, the opposite was
true. Community prosecution appears to have exerted a modest effect on crime rates
on the north side, and yet, its office was the first to open and to re-open after the 2007
shutdown. The west-side office was the third in the sequence of both openings and
re-openings, and yet, its point estimates reveal that it had the largest declines in both
violent and property crimes. These findings discount the possibility that the State’s
Attorney prioritized districts where community prosecution would have the greatest
observable impact.
The implementation of community prosecution in Chicago consisted of two discrete episodes. The first episode began in the late 1990s and ended with the sudden shut-down in early 2007, and the second episode began in mid-2009 with the
re-opening of several offices and continued until the end of the observation period.
The closures in 2007 offer the advantage of abruptness and the likelihood that they
were exogenous to the performance of individual offices. The State’s Attorney closed
all the offices as a budget-cutting measure rather than a selective trimming of less successful offices. A potential disadvantage of this source of variation is that it occurred
simultaneously across all offices, making it difficult to distinguish from other contemporaneous changes that may affect crime. For example, the deterioration in general
economic conditions may have affected crime rates (Raphael and Winter-Ebmer, 2001;
Gould et al., 2002), and the State’s Attorney’s budget-cutting measures included terminations of dozens of prosecutors who were not assigned to community prosecution
(Abernathy, 2008). The openings and re-openings of community prosecution have the
advantage that, in both episodes, these transitions occurred at different dates over three
years. This staggered pattern of reactivation allows the impact of community prosecution to be more plausibly disentangled from citywide contemporaneous trends. But,
the post-activation period in the second episode of community prosecution is relatively
brief, providing fewer data points in which to observe its effect.
Panel C of Table 4 reports a specification in which the single indicator for community prosecution is split into two variables, one for the first episode and one for the
Community Prosecution
137
second episode. Within each crime category, the estimates for each episode bear the
same signs. None of the estimates for the second episode are statistically significant, in
part because their standard errors are larger. The sizes of the estimated coefficients for
property crimes are similar (−0.0332 and −0.0297). In contrast, both violent and part
II crimes feature estimates that are larger (in absolute value) during the first episode
of community prosecution. Despite this difference, the estimates from both episodes
point to similar conclusions about the impact of community prosecution on crime.
5.3. Particular Offense Categories
The community prosecution strategy does not offer ready predictions about the
crime categories it is likely to affect. As it seeks to incorporate the community’s public safety concerns, community prosecution allows flexibility to target a wide range
of offenses. To understand better which offenses community prosecution affected in
Chicago, Table 5 presents estimates for each specific crime category.
The estimated coefficients for each of the violent and property crime categories
imply declines. Some reductions are substantial, such as for murder, rape, and burglary. Perhaps, the most impressive is the statistically significant decline of nearly 8%
in aggravated assault, the offense that, in most years, constitutes more than half of
all violent crimes. The estimates for offenses in the part II category are less consistent. Some offenses, such as disorderly conduct and arson, show large but imprecisely
estimated declines. Other offenses, including the most common ones in this category—
drugs, weapons, and non-aggravated assault—appear unaffected by community prosecution. Still other offenses, including prostitution and the miscellaneous category,
show increases in response to community prosecution.
These results provide further evidence that “broken windows” was not the operative mechanism of community prosecution. There is no consistent decline in crimes
related to the maintenance of public order. While the point estimate for disorderly conduct implies a substantial drop, that for non-aggravated assault implies no meaningful
impact. The conflicting signs and magnitudes of the estimates for the part II offenses
contrast sharply with the persistently negative point estimates for the offenses in the
violent and property categories.
Another possible implication of “broken windows” is that community prosecution
engages residents, raises expectations of public order, and increases their willingness
to report offenses to police. If so, community prosecution should prompt an increase in
the reporting of “victimless” crimes that affect public order but that require third parties
to inform police of them. The results in Table 5 do not provide consistent support
for this prediction. The point estimate for prostitution is positive, while that for drug
crimes is very close to zero.
138
American Law and Economics Review V16 N1 2014 (117–143)
Table 5. Impact of Community Prosecution on Rates of Specific Violent
Crimes: Differences-in-differences Estimates
Dependent variable: log rate of offense type
Panel A: Violent offenses
Murder
(1)
−0.0984
(0.0423)∗∗
[0.0403]∗∗
Rape
(2)
−0.1066
(0.0420)∗∗
[0.0392]∗∗
Agg.
assault
(3)
−0.0784
(0.0224)∗∗
[0.0343]∗∗
Larceny
(2)
−0.0192
(0.0177)
[0.0246]
Motor vehicle
theft
(3)
−0.0474
(0.0174)∗∗
[0.0271]∗
Disorderly
conduct
(2)
−0.1424
(0.0620)∗∗
[0.1423]
Vandalism
(3)
−0.0229
(0.0143)∗
[0.0196]
Arson
(4)
−0.0776
(0.0422)∗
[0.0546]
Weapons
(2)
−0.0076
(0.0392)
[0.0461]
Prostitution
(3)
0.0531
(0.0637)
[0.0774]
Misc. part
II offenses
(4)
0.0275
(0.0191)
[0.0204]
Robbery
(4)
−0.0516
(0.0222)∗
[0.0428]
Panel B: Property offenses
Burglary
(1)
−0.0861
(0.0233)∗∗
[0.0473]∗
Panel C: Part II offenses
Non-agg.
assault
(1)
0.0039
(0.0133)
[0.0218]
Drugs
(1)
0.0040
(0.0402)
[0.0629]
Notes: Ordinary least squares coefficients with standard errors in parentheses and brackets. Standard errors
in parentheses are clustered by police beat, and those in brackets are clustered by police districts. All regressions are weighted by the population of each police beat. Coefficients are from regressions with the same
specification as those in column (3) in Table 3 and are weighted by the population of the police beat.
6. Discussion and conclusion
The regression estimates indicate that community prosecution in Chicago caused
sizable reductions in murder, rape, and aggravated assault. Estimated drops in robbery,
burglary, and motor vehicle theft were socially meaningful but statistically insignificant. To place the estimated effects of community prosecution in context and to assess
whether it is a cost-justified policy, monetary values can be attached to the estimated
crime reductions and the cost of implementing the policy. The costs of community
prosecution in Chicago consisted of opening of a local office, the assignment of two
Community Prosecution
139
Table 6. Estimated Benefits of Community Prosecution: Implied Value of
Reductions in Violent and Property Crime
Offense
Murder
Rape
Aggravated
assault
Burglary
Motor vehicle
theft
Reductions in
incidents per
target area
implied by point
estimates in
Table 5
(1)
Reductions in
incidents per
target area
implied by upper
bound of point
estimates
(2)
McCollister
et al. (2010)
estimated cost
per offense
(3)
5.94
21.68
173.40
0.92
4.84
75.64
$8,982,907
$240,776
$107,020
$8,900,000
$221,000
$63,000
242.66
122.98
113.30
34.51
$6,462
$10,772
$29,000
–
Cohen
et al. (2004)
estimated cost
per offense
(4)
Note: The estimates from McCollister et al. (2010) are for “household burglaries,” and those from Cohen
et al. (2004) are for “serious assault” and “armed robbery.”
experienced prosecutors to each office, and the hiring of support staff for the office.
The starting salary of an assistant State’s attorney in Cook County is under $56,000.
Making the generous assumption that an experienced prosecutor earns a salary of
$75,000 and that other employment benefits raise this cost by a third, the cost of two
experienced prosecutors is roughly $200,000. Rents are likely to vary widely in the
city, and rough estimate for office space and a support staffer may be an additional
$200,000.12 This brings a reasonable cost estimate for a community prosecution office
to $400,000 per year.
Table 6 translates the estimated crime reductions into monetary values. Column (1)
reports that the average annual incidents per target area that the estimates in Table 5
imply were prevented by community prosecution. Only averages for the offense categories having estimates statistically significant at the 10% level or better are shown.
To reflect the degree of uncertainty in the estimates, column (2) reports the estimated
reductions implied by the upper bound estimates of the 95% confidence interval. (As
the point estimate here is negative, the upper bound estimate is the bound closest to
zero.) Columns (3) and (4) contain two sets of recent estimates of the social cost of
individual offense categories: McCollister et al. (2010), which tallies direct and direct
12. In fact, the State’s Attorney paid no rent on any of the neighborhood community prosecution offices in Cook County. The use of each of the office spaces was donated
by local landlords.
140
American Law and Economics Review V16 N1 2014 (117–143)
costs, and Cohen et al. (2004), which uses a willingness-to-pay measure. It is immediately apparent that under either set of estimated crime reductions and under either cost
estimates, community prosecution was cost-justified. Although this is a ballpark calculation, the benefits appear to exceed the costs by several multiples. This analysis is,
of course, incomplete. It leaves out important potential benefits, such as improvements
in residents’ perceptions of public safety, and potential costs, such as longer periods
of incarceration for specific neighborhood offenders.
Another key limitation is that these results may not generalize to other applications of community prosecution. A wide range of practices fall under the banner of
community prosecution, and if other jurisdictions employ a different combination of
prosecutorial activities, similar declines in crime may not be achieved. Despite this
weakness, the paper’s findings are an important advance. They demonstrate for the
first time that community prosecution can be implemented in a way that produces
cost-justified reductions in crime.
The results also provide insight into the mechanisms through which community
prosecution can reduce crime. The theoretical foundation for community prosecution, like its more prominent sibling movement, community policing, was “broken
windows” theory. That theory predicts community engagement affects quality-of-life
offenses initially and later influences more serious crimes. The concentration of the
estimated crime declines in the most serious offense categories and the immediacy
and stability of those declines are not consistent with that prediction. The findings
thus contribute to the literature that casts doubt on the mechanisms hypothesized by
“broken windows.”13
The results also point to directions for future research. The paper’s findings are
likely the first evidence showing that the organizational structure of prosecutors’
offices and the tasks to which they allocate their energies can affect the incidence
of crime. Prosecutors are often described as the most powerful decision-makers in the
criminal justice system, and it is surprising that more is not known about how their
choices affect public safety. With respect to community prosecution, the differencesin-differences estimation approach employed here does not isolate the precise mechanism through which the strategy lowered crime in Chicago. Future research should
attempt to measure the specific tactics deployed by community prosecutors—whether
13. Harcourt (1998, 2001) criticized the primary statistical evidence for the
disorder-to-crime link of “broken windows,” Skogan (1990) cross-sectional study of New
York neighborhoods. Harcourt showed that after correcting for missing values or controlling for neighborhood characteristics, many types of offenses do not correlate with
measures of public disorder. Also, Sampson and Raudenbush (1999) used highly detailed
data on the visual appearance of public spaces in neighborhoods to show that there was
no correlation between disorder and crime.
Community Prosecution
141
it is targeted prosecutions, dialogue with community groups, mediation or other
problem-solving efforts that do not involve prosecution, or something else—and link
these practices to the responses in crime.
References
Abernathy, Samantha. 2008. “Incoming Prosecutor Looking at Options to Reopen
Community Branches,” Medill Reports – Chicago. November 12. Available at:
http://news.medill.northwestern.edu/chicago/news.apsx?id=105217 (accessed
December 15, 2011).
Bertrand, Marianne, Esther Duflo, and Sendhil Mullainathan. 2004. “How Much
Should We Trust Differences-in-Differences Estimates?” 119 Quarterly Journal
of Economics 249–75.
Bibas, Stephanos. 2012. The Machinery of Criminal Justice. New York: Oxford
University Press.
Boland, Barbara. 1998. “Community Prosecution: Portland’s Experience,” in David
R. Karp, ed., Community Justice: An Emerging Field, 253–77. Lanham, MD:
Rowman & Littlefield Publishers, Inc.
Boylan, Richard T. 2005. “What Do Prosecutors Maximize? Evidence from the
Careers of U.S. Attorneys,” 7 American Law and Economics Review 379–402.
City of Chicago. 2011. Data Portal. Available at http://data.cityofchicago.org/
Public-Safety/Crimes-2001-to-present/ijzp-q8t2 (accessed September 13).
Coffee, John C., Jr., Richard Gruner, and John J. Hansen. 1984. What do prosecutors prosecute? An initial statistical profile of federal prosecutor behavior at the
investigative and charging stages. Columbia Law School Working Paper, No. 7,
June 18.
Cohen, Mark A., Roland Rust, Sara Steen, and Simon Tidd. 2004. “Willingness-toPay for Crime Control Programs,” 42 Criminology 86–106.
Coles, Catherine M. 2000. Community prosecution, problem-solving, and public accountability: the evolving strategy of the American prosecutor. Kennedy
School of Government Working Paper, October.
Coles, Catherine M., George L. Kelling, with the assistance of Michael H. Moore.
1998. Prosecution in the community: a study of emergent strategies, a cross site
analysis. Kennedy School of Government Working Paper, September.
Cunningham, Wm. Scott, Brian C. Reneauer, and Christy Khalifa. 2006. “Sharing
the Keys to the Courthouse: Adoption of Community Prosecution by State Court
Prosecutors,” 22 Journal of Contemporary Criminal Justice 202–19.
Doolan, Devin J. 2002. “Community Prosecution: A Revolution in Crime Fighting,”
51 Catholic University Law Review 547–82.
Fagan, Jeffrey, and Garth Davies. 2000. “Street Stops and Broken Windows:
Terry, Race, and Disorder in New York City,” 28 Fordham Urban Law Journal
457–504.
142
American Law and Economics Review V16 N1 2014 (117–143)
Forst, Brian. 2000. “Prosecutors Discover the Community,” 84 Judicature 135–39.
Goldkamp, John S., Cheryl Irons-Guynn, and Doris Weiland. 2002. “Community Prosecution Strategies: Measuring Impact,” Bureau of Justice Assistance
Bulletin. November.
————. 2003. Community Prosecution Strategies. Washington, DC: Department
of Justice, August.
Gould, Eric D., Bruce A. Weinberg, and David B. Mustard. 2002. “Crime Rates and
Local Labor Market Opportunities in the United States: 1979–1997,” 84 Review
of Economics and Statistics 45–61.
Gramckow, Heike. 1997. “Community Prosecution in the United States,” 5
European Journal on Criminal Policy and Research 9–26.
Gray, Kelley Bowden. 2008. “Community Prosecution: After Two Decades, Still
New Frontiers,” 32 Journal of the Legal Profession 199–214.
Harcourt, Bernard E. 1998. “Reflecting on the Subject: A Critique of the Social
Influence Conception of Deterrence, the Broken Windows Theory, and OrderMaintenance Policing New York Style,” 97 Michigan Law Review 291–389.
———. 2001. The Illusion of Order. Cambridge, MA: Harvard University Press.
Harcourt, Bernard E., and Jens O. Ludwig. 2006. “Broken Windows: New Evidence from New York City and a Five-City Social Experiment,” 73 University
of Chicago Law Review 271–320.
———. 2007. “Reefer Madness: Broken Windows Policing and Misdemeanor
Marijuana Arrests in New York City, 1989-2000,” 2007 Criminology and Public
Policy 165–81.
Harp, Caren, Michael Kuykendall, Marcia Cunningham, and Teresa Wise. 2004.
Juvenile Delinquency and Community Prosecution: New Strategies for Old Problems. Alexandria, VA: National District Attorneys Association.
Hickman, Matthew J. and Brian A. Reaves. 2001. Community policing in local
police departments, 1997 and 1999. Bureau of Justice Statistics Special Report,
NCJ 184794. Washington, DC: Government Printing Office.
Jacoby, Joan E. 1980. The American Prosecutor: A Search for Identity. Lexington,
MA: D.C. Heath & Co.
Kelling, George L., and William H. Sousa, Jr. 2001. Do police matter? An analysis
of the impact of New York City’s police reforms. New York: Manhattan Institute
Civic Report No. 22, December.
Klitzing, Nicholas W. 2012. “Fixing the Unfixable: Community Prosecution as a
Problem-Solving Strategy to Reduce Crime and Restore Order in East St. Louis,”
32 St. Louis University Public Law Review 157–99.
Landes, William M. (1971), “An Economic Analysis of the Courts,” 14 Journal of
Law and Economics 61–107.
Lanni, Adriaan. 2005. “The Future of Community Justice,” 40 Harvard Civil RightsCivil Liberties Law Review 359–405.
Community Prosecution
143
Levitt, Steven D. 2004. “Understanding Why Crime Fell in the 1990s: Four Factors
that Explain the Decline and Six that Do Not,” 18 Journal of Economic Perspectives 163–90.
McCollister, Kathryn E., Michael T. French, and Hai Fang. 2010. “Cost of Crime
to Society: New Crime-Specific Estimates for Policy and Program Evaluation,”
108 Drug and Alcohol Dependency 98–109.
Mihalopoulos, Dan. 2001. “Knowing Where the Police Are Posted,” New York Times,
November 4. A19A.
Moulton, Brent R. 1990. “An Illustration of a Pitfall in Estimating the Effects of
Aggregate Variables on Micro Units,” 32 Journal of Econometrics 385–97.
Nugent-Borakove, M. Eliane. 2004a. The Changing Nature of Prosecution: Community Prosecution v. Traditional Approaches. Alexandria, VA: American Prosecutors Research Institute.
———. 2004b. What Does It Mean to Practice Community Prosecution? Organizational, Functional, and Philosophical Changes. Alexandria, VA: American
Prosecutors Research Institute.
Pakes, Ariel, and Zvi Griliches. 1980. “Patents and R&D at the Firm Level: A First
Report,” 5 Economic Letters 377–81.
Raphael, Steven, and Rudolf Winter-Ebmer. 2001. “Identifying the Effect of Unemployment on Crime,” 44 Journal of Law & Economics 259–83.
Reiganum, Jennifer F. 1988. “Plea Bargaining and Prosecutorial Discretion,” 78
American Economic Review 713–28.
Repella, Joy. 2010. Interview with Assistant State’s Attorney for Community Prosecution, April 23.
Sampson, Robert J., and Steven W. Raudenbush. 1999. “Systemic Social Observation of Public Spaces: A New Look at Disorder in Urban Neighborhoods.” 105
American Journal of Sociology 603–51.
Skogan, Wesley G. 1990. Disorder and Decline: Crime and the Spiral of Decay in
American Neighborhoods. Berkeley, CA: University of California Press.
Stuntz, William J. 2011. The Collapse of Criminal Justice. Cambridge, MA: Harvard
University Press.
Thompson, Anthony C. 2002. “It Takes a Community to Prosecute,” 77 Notre Dame
Law Review 321–72.
Wilson, James Q., and George L. Kelling. 1982. “Broken Windows: The Police and
Neighborhood Safety,” 249 Atlantic Monthly. 3 March. 29–38.
Wolf, Robert V. 2010. Community Prosecution and Serious Crime: A Guide for Prosecutors. Alexandria, VA: National District Attorneys Association.
Wright, Ronald F. 2012. “Community Prosecution, Comparative Prosecution.” 47
Wake Forest Law Review 361–72.