Content uploaded by Matthew Freedman
Author content
All content in this area was uploaded by Matthew Freedman on May 07, 2015
Content may be subject to copyright.
1
Your Friends and Neighbors:
Localized Economic Development, Inequality, and Criminal
Activity
Matthew Freedmana, Emily G. Owensb
September 2012
Abstract
We exploit a sudden shock to demand for a subset of low-wage
workers generated by the 2005 Base Realignment and Closure
(BRAC) program in San Antonio, Texas to identify the effects of
local economic development programs on crime. We use a
difference-in-difference methodology that takes advantage of
variation in BRAC’s impact over time and across neighborhoods.
We find that appropriative criminal behavior increases in
neighborhoods where a fraction of residents experienced increases
in earnings. This effect is driven by residents who were unlikely to
be BRAC beneficiaries, implying that inequality can increase
crime. We find less evidence of an impact on serious violence.
Keywords: Inequality, Crime, Local Economic Development, Criminal Opportunities
JEL Codes: K4, R5, H5, J4
a Cornell University, Department of Economics, 262 Ives Faculty Building, Ithaca, NY 14853 (e-mail:
freedman@cornell.edu).
b Cornell University, Department of Policy Analysis and Management, 137 MVR, Ithaca, NY 14853 (e-mail:
ego5@cornell.edu).
2
1. Introduction
A large literature dating back to Becker (1968) and Ehrlich (1973) links economic incentives
to criminal behavior, where individuals divide their time between legal and illegal “work” in
order to maximize their expected utility. A direct implication of this theory is that policy
interventions that improve labor market opportunities, especially opportunities for low-wage
workers, should reduce crime. At the same time, to the extent that policy interventions provide
financial benefits only to a few individuals, they can inadvertently increase the return to criminal
behavior for non-beneficiaries. Compared to the vast empirical literature examining how an
individual’s own legal labor market opportunities affect his or her propensity to engage in
crime,1 research on the impact of changes in other people’s well being on criminal behavior is
scant and contradictory.2
In this paper, we provide evidence on the impact of differential economic opportunity on
criminal activity following a large increase in demand for a specific subset of workers in San
Antonio, Texas. Between 2007 and 2010, the Department of Defense (DoD) spent roughly $2
billion on the renovation and construction of four military bases in the city as part of the 2005
Military Base Realignment and Closure (BRAC). This increase in federal expenditure, which
was roughly equal to 3% of the 2007 metropolitan area GDP and represented a seven-fold
increase in typical military construction spending in the area, created a surge in demand for
certain construction workers, specifically construction workers eligible to work on federal
contracts, which required identity verification and criminal background checks. At the same
time, due to the broader recession, construction workers who could not meet these requirements
faced a substantial reduction in employment opportunities.
While the 2005 BRAC shuttered military installations and withdrew an important source of
employment and income in some cities, it delivered substantial economic benefits to others. In
fact, the consolidation of military operations in San Antonio was described as the “largest
economic development event in the city’s history.”3 Total federal expenditure on construction in
San Antonio associated with the 2005 BRAC was roughly equivalent to annual federal
1 This literature is reviewed in Piehl (1998), Fagan and Freeman (1999), Bushway and Reuter (2001), and Mustard
(2010).
2 This is likely due to the fact that most research does not use quasi-experimental or experimental variation in
inequality. Notable exceptions to this include Bjerk (2010) and Kling et al. (2005). Both identify a positive impact of
inequality on property crime, using, respectively, instrumental variables and experimental variation in local
inequality.
3 http://www.embracebrac.org/
3
expenditure on place-based programs with a national scope, such as the New Markets Tax Credit
and Low-Income Housing Tax Credit.4 However, not only were the billions spent on
consolidating military operations in San Antonio confined to a relatively small geographic area,
but the direct economic benefits of BRAC were, at least initially, concentrated in the hands of
construction workers on federal contracts.5
Using data on employment and neighborhood conditions from the Census Bureau, we show
that poverty rates increased in San Antonio as a whole over this period, but were more stable in
neighborhoods where, historically, more construction workers lived. Also, households in these
construction-intensive neighborhoods were more likely to purchase second cars. We find much
smaller changes in median household income and housing values in affected communities.
Patterns of neighborhood change in San Antonio stand in stark contrast to those in nearby
Austin, where socioeconomic conditions showed little signs of improvement during the 2000s in
areas with more construction workers.
We then show that in San Antonio, BRAC was associated with an increase in car theft,
burglary, and robberies committed by residents of neighborhoods with relatively large
concentrations of construction workers, and that this increase in criminal behavior was driven by
people who had been accused or convicted of felonies in the past. Due to strict employment
guidelines for federal contractors put in place in 2004, these people were unlikely to benefit
directly from BRAC, but are instead better characterized as the neighbors of BRAC
beneficiaries. Our results for car theft and burglary in particular are highly robust to a number of
alternative specifications and controls for pre-treatment trends in criminal activity. We find much
weaker evidence that BRAC neighbors were more likely to engage in crimes that do not have a
clear economic motive. We argue that the most plausible explanation for the observed changes in
criminal behavior is a rational response to an increase in criminal opportunities generated by the
increased earning power of local, BRAC-eligible construction workers.
A large literature focusing on individuals who gain or lose jobs generally finds that the
beneficiaries of improvements in economic conditions commit fewer crimes. Our results suggest
4 The federal government allocated $26 billion to the New Markets Tax Credit program between 2003 and 2009, or
an average of $3.7 billion per year, to encourage commercial investment in low-income communities throughout the
country (Freedman 2012). Lost federal tax revenues associated with the Low-Income Housing Tax Credit, which
subsidizes affordable rental housing development, were just under $5 billion per year in the mid-2000s (Eriksen and
Rosenthal 2010).
5 The San Antonio Business Journal estimated that 80% of the total economic impact of BRAC was the direct result
of the increased labor market opportunities for construction workers (Thomas 2009).
4
that the associated decline in aggregate crime may be somewhat attenuated by increased criminal
activity by those who do not benefit. To the extent that business cycles or economic development
programs increase local disparities in income, our results indicate that increased acquisitive
crime may be an unanticipated and unfortunate consequence. However, our findings do not
imply BRAC was a net negative for San Antonio; taking the social cost of increased criminal
activity into account, we estimate that the local economic multiplier used by the DoD to evaluate
the impact of BRAC was at most 0.2% too large.
The paper proceeds as follows. In the next section, we briefly summarize the existing
literature on economic inequality and criminal behavior. In Section 3, we provide institutional
background on the 2005 BRAC, with particular emphasis on the selection process and pattern of
spending. We also discuss other public works projects that partially coincided with BRAC.
Section 4 provides a theoretical framework for thinking about the impact of BRAC on criminal
behavior. We then describe the data we use to measure the impact of BRAC on crime in Section
5, and outline our differences-in-differences identification strategy in Section 6. In Section 7, we
present evidence that BRAC was associated with specific socioeconomic improvements in
neighborhoods with more construction workers, but also with higher rates of acquisitive crime.
We conclude with a brief discussion of the results and their implications in Section 8.
2. Inequality, Criminal Opportunities, and Economic Development
In the now standard economic model of criminal behavior (Becker 1968, further developed
by Ehrlich 1973), rational agents will engage in crime if doing so increases their lifetime
expected utility. There are essentially three parts to an individual’s decision: the utility associated
with legal employment, the utility associated with engaging in crime, and the expected utility
loss from being punished for criminal acts. On the margin, people should equalize the expected
net return of spending an additional hour in legitimate and illegitimate activity.
Researchers using quasi-experimental variation in employment opportunities, particularly
opportunities available to low-wage earners, have generally found that higher wages and lower
unemployment rates are associated with lower aggregate crime rates (Raphael and Winter-Ebmer
2001, Machin and Meghir 2004, Mocan and Rees 2005, Machin and Marie 2006). An enormous
literature also indicates that increasing the expected cost of crime, either by increasing penalties
5
or increasing the probability of detection, will reduce the incidence of crime, although there is
still some debate about the magnitudes of these effects (Durlauf and Nagin 2011).
There is comparatively little evidence on how responsive people are to variation in the
private return to crime, or “criminal opportunity.” Criminal opportunity is typically broken into
three components: (1) propinquity, the cost of obtaining information about the return to a
criminal act, which in criminology is often measured as the physical distance between and
offender and victim (Canter and Youngs 2009), (2) payoff, the gross private return to committing
the crime, and (3) vulnerability, the expected level of resistance by the victim (Cook 1986).6
Experimental evidence suggests that variation in criminal opportunities are potentially of
great importance in explaining crime patterns, and specifically that income inequality can induce
worse-off people to offend (Harbaugh et al. 2011). However, empirical findings outside of the
lab are not entirely conclusive. Glaeser and Sacerdote (1999) estimate that steeper local income
gradients can explain at most one-fourth of the elevated crime rates in cities compared to rural or
suburban areas. A handful of cross sectional studies of income inequality and property crime find
a positive, although often statistically imprecise, relationship between the two (Fajnzylber et al.
2002, Kelly 2000, Hsieh and Pugh 1993). However, the time series analyses in Brush (2007) and
Saridakis (2004) yield negative relationships. Meanwhile, Kling et al. (2005) find that boys who
moved to slightly wealthier neighborhoods as part of the Moving to Opportunity experiment
were more likely to be arrested for property crimes than the control group, but girls were not
affected. Bjerk (2010) presents quasi-experimental evidence that increasing income segregation,
which implies that poor people are less likely to interact with wealthier people, lowers property
crime rates at the city level, but increases violence.
In this paper, we combine data on where criminals live with quasi-exogenous variation in
local economic conditions generated by military spending during the 2000s in the city of San
Antonio, Texas. While this spending may have reduced the propensity to engage in crime among
direct beneficiaries, we show that because of the targeted nature of the spending program, an
important effect of this program was to increase the criminal opportunities of the average San
Antonian. In the Cook (1986) framework, increasing the income of construction workers likely
increased their neighbors’ payoff from committing property crime, especially to the extent that
6 Technically, Cook (1986) describes four components of criminal opportunity, but he includes the expected loss
from punishment. We draw a distinction between expected punishment and other parts of the Cook (1986) definition
to highlight the relative lack of research by economists on this particular issue.
6
neighbors may have been negatively affected by the Great Recession. At the same time, if
construction workers responded to the increased demand for their labor by working more and
spending less time at home, this could cause a corresponding increase in the vulnerability of their
households.7
Our identification strategy is in part based on the fact that our local economic shock
benefited a specific subset of the working population, creating geographic heterogeneity in the
impact of the program based on where these workers lived. In that sense, our identification is
similar to that of Machin and Marie (2006), who exploit geographic heterogeneity in the impact
of a reduction in unemployment insurance benefits in the UK to identify the net relationship
between crime and economic strain. However, without information on who was committing
crime, they are unable to disentangle the impacts of reduced income and reduced criminal
opportunities. Further, as we will show, the benefits of BRAC were even more tightly
concentrated in the hands of a particular subset of the population than were the costs of the UK
benefit cut.
In addition to this geographic variation, we take advantage of highly detailed information on
the accused criminals. Due to federal employment rules, BRAC jobs were only open to people
without serious criminal records. This institutional detail, combined with our rich data, allows us
to separately identify the impacts of the legal and criminal opportunities created by the program.
Because our dataset contains information on all felony cases dating back to the 1970s, we are
able to differentiate between the criminal activity of people who could have potentially benefited
from BRAC and the activity of those who were not eligible to work on BRAC projects, but were
plausibly aware that others benefited from a BRAC windfall.
The implementation of the 2005 BRAC was in many ways comparable to place-based
programs, such as state enterprise zones, the Low-Income Housing Tax Credit (LIHTC), the
New Markets Tax Credit, Weed and Seed, and Business Improvement Districts (BIDs), in the
sense that its impacts were concentrated in certain geographic areas. A small literature has found
mixed evidence on the impacts of these programs on crime. For example, Bushway and Reuter
(2001) review evaluations of Weed and Seed, which generally find no impact on crime at all. In
contrast, Cook and MacDonald (2011) find that BIDs, where businesses pay extra taxes or fees
7 Cantor and Land (1985) emphasize this perverse impact of lowering unemployment on crime - workers are away
from their home more often. Research on criminals’ response to changes in the vulnerability of victims, particularly
through hand gun regulation, is divided; see Cook et al. (2011) for a review of this literature.
7
to finance improvements within a designated area (often including improvements to security),
reduce property crime as well as violent offenses.
Meanwhile, Freedman and Owens (2011) find that rental housing development in low-
income areas subsidized by the LIHTC is associated with reduced rates of assault and robbery,
but higher rates of car theft. By improving the quality of the housing stock in poor
neighborhoods, the LIHTC program may have attracted slightly wealthier people to low-income
communities, which could increase criminal opportunities for existing residents. Similarly, by
improving the welfare of some residents and not others within neighborhoods, the 2005 BRAC
in San Antonio might be expected to foster more acquisitive crime in affected communities.
However, not only was BRAC spending in San Antonio large and highly concentrated, but as we
will show, its impacts on crime are likely to have operated more through changes in the
purchasing power of certain residents than through changes in neighborhood composition.8
3. The 2005 Military Base Realignment and Closure
In 2005, Congress established a new Defense Base Closure and Realignment Commission,
which was tasked with orchestrating the first military base realignment and closure (BRAC) in
ten years.9 The goal of the 2005 BRAC was to increase the efficiency of the DoD by
concentrating domestic military operations in a smaller number of areas. In May of 2005, the
commission announced that San Antonio would become the new “home of Military Medicine
and Installation Command” for the U.S. Military.10
In previous BRACs, the DoD officially ranked “local economic impact” as the third most
important criteria in their reshuffling decision. When the selection criteria for the 2005 BRAC
were announced, local economic impacts had fallen to the 12th most important criteria. Instead,
locations with more available space and little history of residents complaining about base
activities received preference in the BRAC selection decision (Sorenson 2007).
BRAC affected four bases in San Antonio. Air Force units from other parts of the country
were reassigned to three existing bases in the San Antonio area: Randolph Air Force Base,
8 Existing research on the impacts of place-based initiatives often suffers from lack of power; partly as a result, there
is only limited evidence that geographically targeted state and federal economic development programs improve
economic conditions at all, let alone enough to generate a measurable impact on criminal behavior (Glaeser and
Gottlieb 2008). The sheer magnitude of spending associated with BRAC in San Antonio makes it an attractive
candidate to study the social impacts of local economic development programs.
9 Previous BRAC rounds occurred in 1988, 1991, 1993, and 1995.
10 DoD Document AFD-101004-006.ppt.
8
Lackland Air Force Base, and Brooks City-Base. Most new military activity would occur at Fort
Sam Houston, a base roughly two miles northeast of the Alamo in downtown San Antonio,
which was designated as a new major medical research and education center for the DoD.
Overall, the DoD predicted that the 2005 BRAC would bring roughly $8.3 billion to San Antonio
by 2011 (Nirvin 2009).
This economic boon was not equally shared by all residents. In fact, approximately 80% of
the federal money would be spent on construction and renovation (Nirvin 2009). The federal
government awarded $92 million in BRAC construction contracts in September of 2007,
followed by an additional $1.2 billion in 2008 and $700 million in both 2009 and 2010.11 To put
these expenditures in perspective, the military spent between $65 and $100 million on
construction in San Antonio per year prior to 2005 (AFD-071217-009).12
In the same way that the effects of place-based economic development programs are spatially
concentrated, BRAC’s effects were felt more in some neighborhoods than others. In particular,
its initial impacts were felt most acutely in neighborhoods in which a large fraction of workers
were employed in the construction industry. Figure 1 highlights the plausible spatial
heterogeneity in the impact of BRAC spending on the purchasing power of households in
different parts of Bexar County, which contains the city of San Antonio. The figure shows the
fraction of employment in the construction industry across census block groups based on 2000
Decennial Census data and 2005-2009 American Community Survey data. The construction
industry is a relatively important employer in several of the larger block groups in the southern
and central parts of the county as well as in many of the smaller block groups that constitute
downtown San Antonio. Notably, the spatial distribution of construction workers changed little
during the 2000s.
While BRAC was a federal project, construction jobs were contracted out to private
companies. The companies that won BRAC contracts were primarily headquartered in San
Antonio or had large branches in the area. In September of 2009, a representative of the
11 Depending on the credit constraints of construction workers in San Antonio, it is plausible that their purchasing
power increased when BRAC was announced rather than when BRAC contracts were awarded. For this reason, we
do not emphasize this variation in military expenditure between 2007 and 2010, and in our analysis we explicitly
allow for an increase in the purchasing power of construction workers, and thus the criminal opportunities of their
neighbors, after the announcement of BRAC.
12 Obviously, the timing of a grant or construction contract will not line up with when workers are actually paid. For
this reason, we will use variation in the total number of dollars awarded in each BRAC year as a proxy for variation
in the potential consumption of construction workers.
9
Association of General Contractors of San Antonio estimated that two-thirds of all commercial
construction in San Antonio was taking place on one of the bases, and that without BRAC,
unemployment in the construction industry would be “at 15-17 percent.” (Thomas 2009).
There is some evidence to suggest that BRAC spending buffered the San Antonio
construction industry against the economic downturn and collapse of the housing market in the
late 2000s. In Figure 2, we compare employment and wages of construction workers in Bexar
County to those in nearby Travis County, where the state capital Austin is located, as reported in
the Quarterly Census of Employment and Wages (QCEW) from 2001 to 2010.13 From 2005 to
2007, Bexar was losing construction jobs relative to Travis, and those jobs paid roughly 86% of
Travis wages. After BRAC spending began, this trend reversed; the gap in construction jobs
between Bexar and Travis grew by over 3,500 between 2007 and 2009, an amount roughly
equivalent to some DoD estimates of the number of construction jobs created by BRAC (Thomas
2009). Over the same period, the wage premium in Travis shrank to less than 10%. Clearly, San
Antonio construction workers fared better than workers in neighboring markets during the
second half of the decade.
While BRAC was “keeping a lot of people busy … who would otherwise be struggling to
find work” (Thomas 2009), these jobs were not necessarily open to all construction workers.
Civilians working on federal contracts enjoy relatively lucrative wages and benefits,14 but face
some additional barriers to employment. Specifically, on August 27, 2004, President George W.
Bush issued Homeland Security Presidential Directive 12, which required that all employees of
federal contractors have a “verified identity” if they were to be allowed access to a federal
government facility. The language of this Directive was subsequently interpreted by the
Government Accountability Office (GAO) as requiring all employees of companies with federal
contracts to undergo a criminal background check (U.S. GAO 2007).15 Because of Directive 12,
construction workers with criminal histories, or without proper documentation of their eligibility
13 QCEW data are not available prior to 2001. We define construction using the two digit NAICS code 23.
14 The Davis-Bacon Act requires construction contractors to pay prevailing wages and benefits to workers on federal
projects.
15 The requirement that federal contractor workers undergo background checks has been the subject of some
controversy, but was upheld by the Supreme Court in 09-530 National Aeronautics and Space Administration et al.
v. Nelson et al. (2011). That case established that the U.S. Federal Government had the specific right to know about
“‘violations of the law,’ ‘financial integrity,’ ‘abuse of alcohol and/or drugs,’ ‘mental or emotional stability,’
‘general behavior or conduct,’ or ‘other matters’” related to the character of non-federal employees working on
federal contracts.
10
to work in the U.S., were in principle ineligible to work on contracts awarded by the federal
government.
The announcement of BRAC coincided with another shock to lower income, working San
Antonio residents, particularly those working in construction and tourism. In June of 2005, Phil
Hardberger was elected city mayor. During his four-year term, Mayor Hardberger oversaw three
major projects. First, after Hurricane Katrina, Mayor Hardberger successfully lobbied to have the
displaced New Orleans Saints NFL franchise temporarily move to San Antonio and play half of
their home games in the city’s major sports arena, the Alamodome. Second, between 2006 and
2007, Mayor Hardberger oversaw a significant extension of the San Antonio Riverwalk, one of
the city’s major tourist attractions, which is lined with restaurants, bars, and hotels. Finally,
Mayor Hardberger spearheaded the renovation of the historic downtown Main Plaza in 2008.
Because these were locally-initiated projects that directly affected multiple industries and were
not subject to federal contractor rules, it is less obvious that the Hardberger projects can be
interpreted as shocks to local economic inequality, but we will be careful to take these projects
into account in our empirical analysis of BRAC.
In Figures 3 and 4, we provide graphical evidence on the impact of BRAC as well as the
Hardberger projects on employment and wages in San Antonio, again using data from the
QCEW. In Figure 3, we plot the number of jobs in Bexar County in construction (NAICS 23),
tourism and food services (NAICS 71 and 72), and health care and social services (NAICS 62).16
We chose these two additional industries because of their importance to the San Antonio
economy; roughly one quarter of San Antonio jobs were in these two sectors in 2011 (Greater
San Antonio Chamber of Commerce 2011). Tourism in particular should have directly benefited
from the Hardberger projects, and people working in the food service industry may have
received some early spillover benefit from BRAC.17
Employment in these three sectors follows two noticeably different paths. Both health care
and tourism jobs grew steadily through the decade, with some downturn in tourism jobs in 2009.
16 The number of construction jobs in the QCEW actually created by BRAC is difficult to determine. However, the
Army Corp of Engineers estimated that, during the construction period, BRAC brought on average more than 2,200
construction workers to bases every day (Thomas 2009). Note that this is roughly two-thirds of the size of the jump
in QCEW construction jobs in 2007.
17 There is a weak, but positive correlation (ρ=0.09) between the residential choices of tourism and construction
workers in Bexar county. Tourism workers tend to live in the southwestern part of Bexar County. There is a heavy
concentration of both construction and tourism workers along Route 281 in central San Antonio, just east of Fort
Sam Houston.
11
The stability of these industries in the late 2000s is in line with other work suggesting that these
San Antonio did not suffer as much as other parts of the country during the Great Recession
(Puentes and McFerrin 2012). Meanwhile, construction jobs declined between 2001 and 2003,
stabilized in 2004, then after the beginning of the Hardberger projects, jumped by almost 6%.
After the first BRAC contract was awarded in 2007, construction employment jumped again,
from 40,000 jobs to 43,000 jobs. However, even as BRAC spending was ramping up, total
construction jobs declined in 2009 and 2010 after Hardberger left office and his projects wound
down. One implication of this is that construction workers who could not meet federal
employment standards faced increasing economic hardship in these years.
Trends in wages further confirm that high-paying BRAC jobs were an increasingly large
share of all construction work. In Figure 4, we see a corresponding increase in the wages of
construction workers, relative to the health care industry, that coincides with the Hardberger
projects and implementation of BRAC. Notably, workers in tourism also saw a jump in their
salary, especially in 2006, which we attribute to the shared benefit of the Hardberger projects
among these two industries. While not directly apparent in the figure, in terms of purchasing
power, growth in construction wages was substantially larger than in tourism wages. Average
wages in construction increased from $635 a week in 2001 to $781 a week by 2006, and were
$862 a week by 2010. Workers in the tourism industry earned, on average, $279 a week prior to
2006. Their average weekly wages increased to $327 in 2006, fell by $14 a week in 2007, and
finally rose to about $330 per week between 2008 and 2010.
Based on the graphical evidence, as well as DoD and Chamber of Commerce publicity, we
conclude that BRAC had a substantial effect on the number and nature of employment
opportunities for construction workers in San Antonio. At the same time, there were other
important changes in the market for construction workers, in particular the Hardberger projects
and the Great Recession, which will be important to take into account in our analysis.
4. Local Economic Shocks and Crime
We will use a simple model of appropriative conflict in the spirit of Ehrlich (1973),
Grossman and Kim (1995), and Bjerk (2010) to think about the possible impact of the BRAC on
the criminal behavior of two types of neighbors: construction workers who directly benefit from
BRAC and people who, because of their human capital, immigration status, or criminal history,
12
do not receive any direct benefit from BRAC. While simple and stylized, this model captures
some key features of the relationship between criminals and victims, and in particular the role of
income shocks and geographic space.
Suppose that a given individual i can earn wi in the legitimate labor market. They can
supplement their income by stealing, which gives them an expected return of si additional dollars,
but with the potential loss of utility u(f) if they are caught. Apprehension happens with
probability p. Each neighbor will engage in crime in a given period if and only if u(wi + si) –
pu(f) ≥ u(wi). Assuming that people are risk averse, with u’ > 0 and u” < 0, a higher legal wage
will reduce person i’s likelihood of engaging in crime, as it will reduce the extra utility gained
from an additional si, but not the disutility associated with punishment. It is also the case that as
si increases, the incentive to commit crime increases. This begs the question: what determines si?
We follow theoretical research in criminology, specifically routine activity and the distance-
decay hypothesis (Cohen and Felson 1979), and define si as
iijj
j
dhf
w
),(
, where dij is the linear
distance between neighbors i and j and hj represents neighbor j’s investment in protection from
crime.18 We include a flexible function in distance in order to capture several possible
mechanisms relating distance to criminal behavior; not only is travel costly, but, following the
idea of propinquity, the amount of income that other people have, wj, may not be known with
certainty when someone decides to commit a crime. We assume that individuals have better
information about the wages of people who live closer to them, making one’s neighbors more
attractive targets for theft than someone whose actual resources are less clear.19 This is
particularly important when we think about the plausible impacts on different types of criminal
behavior. Any person passing through a neighborhood might notice a fancy car parked in the
driveway, but neighbors are much more likely to notice delivery vans, electronics boxes, or
backyard furniture that would suggest opportunities for burglary or theft.20
18 To the extent that employment opportunities may reduce the amount of time that construction workers spent at
home, BRAC may have reduced investment in protection. Since we are unable to observe changes in hj, we focus on
observable variation in dij.
19 At the same time, people are more likely to know (or at least recognize) their neighbors, so these mechanisms are
counterbalanced by the fact that anonymity itself may also contribute to criminal behavior (Wilson and Herrnstein
1985).
20 There is some evidence that the income elasticity of consumption of “visible” goods, specifically cars, clothing,
and furniture, is higher than the income elasticity of less conspicuous items like underwear or life insurance (Heffetz
2011).
13
A direct implication of this formulation of criminal opportunity is that the change in criminal
opportunities generated by BRAC will be geographically concentrated around BRAC
beneficiaries. There is a fair amount of empirical support for this hypothesis. For example,
Bernasco et al. (2012) estimate that robbers in Chicago are over five times more likely to operate
in a census block for each log-kilometer closer it is to their home. Rhodes and Conley (1981)
estimate that the average burglary occurred 1.2 miles from a burglar’s home, the average rape
about 0.73 miles from a rapist’s home, and the average robbery less than 1.6 miles from a
robber’s home. They also point out that people living in “target rich” areas travel shorter
distances to engage in crime. Consistent with Rhodes and Conley, Phillips (1980) finds that
assaults and rapes tend to be committed within a mile of the offender’s home. Wiles and Costello
(2000), meanwhile, find that most offenders travel less than two miles from home to commit
burglary and robbery, but travel as much as two and a half miles on average to commit larceny.
Recall that BRAC created a positive, temporary shock in demand for construction workers
who were eligible to work as federal contractors. This should have lowered the incentives for
these construction workers to engage in crime, but simultaneously made these workers more
attractive criminal targets for the remainder of the population. The net effect of BRAC on crime
is therefore unclear, as it depends on both the prevalence and behavioral response of BRAC
beneficiaries and those who did not directly benefit. Among all San Antonians not employed on
BRAC projects, we expect that those who live closest to BRAC construction workers, ceteris
paribus, would be most likely to increase their criminal behavior because of BRAC. Note also
that, to the extent that criminals who commit crimes close to home are more likely to be caught,
we might expect more of these offenders to be arrested.21
5. Estimating the Impact of BRAC on Crime
We estimate the impact of BRAC on criminal behavior using data on all felony charges filed
in Bexar County District Court between 1976 and 2010, focusing on the post-2000 data. There
are two features of this dataset that merit discussion. First, individuals only appear in these data
if they had felony charges filed against them, and multiple people could be accused of the same
criminal act. The fact that only a fraction of crimes result in a felony charge is a limitation of our
data, but it is important to point out that all research on the characteristics of offenders using
21 This could be reflected in our simple model by making p a function of dij.
14
official reports suffers this limitation. For example, researchers regularly interpret the age of
arrestees in the UCR as representative of changes in the age of offenders (see, for example,
Donohue and Levitt (2001) and Lochner and Moretti (2004)).
In Table 1, we present estimates of the number of criminals per index crime (Panel A) and
number of adults arrested (Panel B) in Bexar County using county-level crime estimates from the
Uniform Crime Reports (UCR) for 2000-2009.22 Most of the difference between felony charges
and crime rates appears to be driven by police activity, as the number felony charges is much
closer to the number of adults arrested each year. Not surprisingly, there are more felony charges
for murder, rape, and robbery per known crime than for less serious offenses, and no more than
2% of the larcenies occurring in Bexar County appear to result in felony charges being filed,
corresponding to 10-30% of adults arrested.23 While roughly 40% of adults arrested for sexual
offenses appear to be charged with felonies, definitional differences between UCR and state
statutes make this difficult to interpret.24 There is a general increase in the number of felony
charges filed per crime in Bexar County over time, but this time trend is substantially weaker
conditional on adults arrested.
If the fraction of people arrested and eventually charged with felonies were changing over
time in a way that was correlated with the fraction of people who worked in construction, then
our estimates will reflect a combination of both increased criminal behavior and increased
probability of punishment for criminal behavior when some residents earn more money. Instead
of being problematic bias, we argue that this would be consistent with a behavioral change in our
model of criminal opportunity; specifically, people deciding to commit crimes closer to where
they live, where they may be more likely to be apprehended.25
22 These estimates are derived from the UCR County-Level Detailed Arrest and Offense Data, which are not yet
available for 2010.
23 Police clearance rates for larceny are generally quite low. Also, because larceny is typically considered a minor
crime, prosecutors may be less likely to file felony charges against an arrested thief.
24 We try to include people charged with what is commonly understood to be a sexual assault in our definition of
rape. For example, we do not define felony charges for sodomy, incest, unnatural sexual acts, or fondling in our
definition of rape, but we do include charges for any type of sexual assault. The FBI records part one rape and part
two sexual offenses. The definition of UCR rape is exclusively carnal knowledge of a female against her will, which
is narrower than we would like to define sexual assault. At the same time, UCR sexual assault, a part two offense,
includes any offenses against “common decency,” which we think is too broad a definition. For more information,
see www2.fbi.gov/ucr/05cius/about/offense_definitions.html.
25 This point is specifically raised in Brantingham and Brantingham (1981). Alternatively, people with more money
might be more willing to cooperate with the police, which would lead to more arrests and charges filed. This is also
a policy relevant outcome, as it means that local economic development could place greater strain on the criminal
justice system.
15
Taking advantage of information on initially filed charges and a fingerprint-supported unique
identifier in the Bexar County District Court data, we identified individuals who were accused of
committing a crime that occurred between 2000 and 2010. We then used mapping software to
locate the census block group where each individual in the data lived at the time that charges
were filed against them.26 Block groups, which are the second smallest geographic unit identified
by the Census Bureau, are larger than a city block but smaller than a census tract. Census tracts
roughly correspond to homogenous “neighborhoods,” and in Bexar County, there are just under
four block groups per census tract on average. The median population of the 1,009 block groups
in Bexar County was 1,100 in 2000.27 The median land area was 0.2 square miles, well within
the range that most criminals travel.
For each census block group in Bexar County, we calculated “crime rates” for the seven
major index crimes (burglary, car theft, larceny, robbery, murder, rape, and assault) using block
group population estimates based on linear interpolations between the 2000 Decennial Census
and 2005-2009 American Community Survey (ACS).28 Our calculation of car theft is worth
some discussion, as Texas does not have a specific law against stealing motor vehicles. Someone
who takes another’s car with the “intent to deprive” the owner of that property is charged with
theft, with the sub-classification of theft of a vehicle. This is a different, and more serious, charge
than unauthorized use of a motor vehicle (UUM). In Texas, UUM means that someone operated
a vehicle without the consent of the owner, but without the intent to permanently deprive the
owner of that vehicle. In essence, Texas law differentiates between someone stealing a car for
acquisitive purposes and “joyriding.” Since we are primarily interested in acquisitive crime, we
focus on theft of a motor vehicle.
For our main measures of criminal activity, we simply calculated the number of offenses that
residents of each census block group were alleged to have committed each year. Next, we
divided our criminal activity rate into two additional groups: crimes allegedly committed by
people who had never appeared in the Bexar District Court before, and crimes allegedly
committed by those who had previously been accused of a crime that occurred at any time since
1976. Since most, if not all, of BRAC workers were required to have criminal background
26 We use 2000 Decennial Census geographic boundaries.
27 We drop seven block groups in Bexar County that had zero population in either the 2000 Decennial Census or the
2005-2009 American Community Survey data.
28 We also extrapolate population to 2010.
16
checks, we argue that these people are more likely to be ineligible to work on a BRAC contract.
We further refined this by calculating a fourth crime rate, based on crimes alleged to be
committed by people who had previously been found guilty of a felony in Bexar County. To the
extent that Directive 12 was enforced, any change in the criminal behavior of this group after
BRAC should only be related to the increase in criminal opportunities.
We then linked these block group crime rates to information on local demographic
characteristics and economic conditions using 2000 Decennial Census data and 2005-2009 ACS
data.29 We used the 2000 Decennial Census to calculate the fraction of jobs held by construction
workers in each block group in Bexar County. Obviously, this is only a coarse proxy for the
number of workers who will benefit from BRAC, and therefore our identification relies on the
assumption that there is a positive correlation between the share of adults working in
construction, in general, and the share that will be eligible to work on federal contracts. Notably,
measuring construction share as of 2000 means that our identification is not based on jobs that
were created, and then lost, during the construction boom years of 2005-2008 and the bust years
of 2009 and 2010. To the extent that only a fraction of construction workers in a block group will
be eligible for BRAC jobs, this will only sharpen the differential impact of BRAC on household
consumption and criminal opportunities within a neighborhood.
We also extracted from the 2000 Decennial Census a host of block group demographic
characteristics, including information on total population, racial and ethnic composition, the age
distribution, educational attainment levels, household and family income, poverty rates,
employment rates, and unemployment rates. The 2000 data also include a number of housing
variables, including total housing units, share vacant, share occupied, share owned, share rented,
median age of units, household turnover, median house values, and vehicle ownership.
To assess changes in neighborhood conditions later in the decade, we use recently released
small-area estimates from the 2005-2009 ACS. These estimates are based on interviews
conducted by the Census Bureau between January 1, 2005 and December 31, 2009. The ACS
block group estimates cannot be used to measure neighborhood characteristics in a given year;
they can only be used to measure average neighborhood characteristics over the entire five-year
period. Notably, though, the dates for which the ACS block group estimates are available bracket
29 The geographic boundaries in the 2005-2009 ACS for Bexar County match those used in the 2000 Decennial
Census, ensuring that no measurement error arises from changes in geographic boundaries driven by shifts in the
geographic distribution of the population.
17
the period during which the BRAC and Hardberger projects were underway and include the early
years of the Great Recession. We extract from the ACS information on population, poverty rates,
employment, household income, housing units, median house values, household turnover, and
vehicle ownership.
The magnitude of the impact of BRAC on the criminal opportunities of non-beneficiaries is
assumed to be proportional to the fraction of workers in that block group who work in
construction, or “construction share.” In Table 2, we present some basic descriptive statistics for
block groups with 2000 construction shares above and below the 50th percentile of construction
share (which is just over 7%). Block groups with higher construction shares also tend to have
more workers in the tourism industry and fewer workers in the health care sector. It is also clear
from the table that areas with higher construction shares are typically more disadvantaged along
a number of dimensions. Educational attainment levels, income levels, and house values are all
lower in areas with higher construction employment shares. Not surprisingly, each type of major
crime is more common in neighborhoods with higher construction shares.
In Figures 5 and 6, we provide some graphical evidence on the net effect of BRAC on crime
rates, dividing crimes by whether or not there is a clear economic return to the behavior. We
mark in the figures both when BRAC was announced and the Hardberger projects began (2005)
as well as when the first BRAC contract was awarded (2007). Larceny, the most common
acquisitive crime, appears to have been increasing in Bexar County for most of our sample
period. There is a slight upward trend in burglary, car theft, and robbery in the early 2000s, but
there is a sharp increase in these crimes between 2007 and 2008. Turning to crimes with less of a
clear financial motive, we see less evidence of a shock in the BRAC years. Rape and murder
rates, which are multiplied by ten for ease of comparison, are flat or slightly downward sloping
between 2000 and 2010. Assault has a strikingly different pattern, with the number of people
charged with this crime increasing sharply and continuously from 2006 to 2009.
6. Analytic Framework
The overall trends in felony charges suggest that BRAC coincided with an increase in some
crimes in San Antonio. If this actually was caused by an increase in criminal opportunities, we
would expect the increase in burglary, robbery, and car theft rates to be larger in neighborhoods
where more people plausibly benefited from BRAC. We therefore use a continuous difference-
18
in-difference strategy to identify the net effect of this localized economic shock on criminal
activity. The main outcome of interest is the natural log of the number of crimes committed by
residents of a block group in a given year divided by the estimated population of that block
group in that year. In our main specification, we examine how crime rates (technically the rate of
felony charges filed) vary with construction shares over the 2000s, controlling for year and block
group fixed effects:
(1)
btbtbttb
tb
bt
bt
BRACAwardonShareConstructi
rgerBRACHardbeonShareConstructi
Population
sargeCh
X
2
1
)(
)(ln
where b indexes census block groups and t years.30 The main coefficients of interest are β1 and
β2, the coefficients on the interactions of the share of block group workers who work in
construction and the two stages of BRAC. The first-order impact of BRAC on criminal behavior
is absorbed by year fixed effects ηt, and therefore β1 and β2 differentiate between block groups
where we expect the economic impact of BRAC to be larger. BRACHardbergert takes the value
of 1 in the years 2005 and 2006, when the BRAC decision was made public and the Hardberger
projects began. During this period, wages for both construction workers and tourism workers
rose due to Hardberger, and construction workers’ propensity to spend may have also increased
in anticipation of BRAC. BRACAwardt is equal to one in the years 2007, 2008, 2009, and 2010,
when BRAC construction took place. While BRAC did not directly benefit those in tourism, it is
plausible that construction workers were more likely to eat out or otherwise engage in activities
that would have benefited workers in that sector during this period.
We also include in Xbt interactions of the share of block group employment in tourism
(NAICS 71 and 72) and the share in health care (NAICS 62) with dummies for both stages of
BRAC. Finally, we control for time-invariant differences across the 1,009 block groups in our
sample with a vector of block group fixed effects, θb. In all regressions, we allow for arbitrary
correlation in crime rates within block groups by clustering our standard errors at the block
group level.
In a series of robustness checks, we test the sensitivity of our estimates to a number of
alternative modeling strategies. Specifically, we check the robustness of our results to including
30 We also considered specifications controlling for baseline (i.e. year 2000) block group demographic and housing
characteristics along with tract and year fixed effects. We show the results of these regressions, which are
quantitatively and qualitatively similar to those with block group fixed effects, in Appendix Tables A1-A9.
19
flexible controls for pre-BRAC trends in crime, using crime rates as the outcome, using
alternative measures of construction employment concentration, and replacing
BRACHardbergert and BRACAwardt with separate dummies for each year between 2004 and
2010. We also explore how the Great Recession may have interacted with BRAC spending to
affect outcomes. We discuss each of these robustness tests after presenting our main results in
the next section.
7. Results
7.1. BRAC and Neighborhood Characteristics
Before we present our estimates of the impact of BRAC on criminal behavior, we first must
establish that BRAC improved the economic circumstances of some San Antonio residents,
increasing the criminal opportunities for others. We do this by replacing the dependent variable
in equation (1) with a series of measures of block group economic conditions and neighborhood
characteristics.31 These outcomes are measured in the 2000 Decennial Census and again in the
2005-2009 ACS, such that we only effectively have two observations for each block group.
Further, to the extent that some of the surveys used to generate the 2005-2009 ACS estimates
were conducted prior to construction beginning on some of the Hardberger or BRAC projects,
we might expect our estimates to understate the degree of neighborhood change owing to the
projects. Still, if we see relative improvement in neighborhood conditions between 2000 and
2005-2009 in areas with relatively more construction workers, it would lend credence to our
claim that BRAC increased criminal opportunities relatively more in neighborhoods with a
disproportionate number of construction workers.
The results of our analysis of changes in the economic well-being of San Antonians appear in
Panel A of Table 3. To highlight the impact of BRAC and the extent to which it mediated some
of the effects of the recession of the late 2000s, we show results of the same regressions for
nearby Travis County in Panel B of the same table. We find that in contrast to trends in Travis
County, poverty rates in Bexar County were statistically significantly lower in block groups with
greater shares of construction workers after 2005. Note, however, that the average block group
resident in Bexar County did not appear to benefit from BRAC; as in Travis County, the impact
31 Observation counts in these regressions vary slightly across regressions due to missing information on
neighborhood characteristics.
20
on median household income is small and statistically indistinguishable from zero. This is
consistent with the idea that BRAC only benefited a subset of households in a neighborhood.
Meanwhile, median house values in affected San Antonio neighborhoods rose slightly, by
about 0.5% for each additional percentage point of employment in construction, an elasticity of
4%. This again contrasts with Travis County, which experienced no differential change in
median home values in communities with more construction workers.
Notably, a one percentage point increase in the share of employment in construction
increased the percentage of households in Bexar County with two or more vehicles by a
statistically significant 0.22 percentage points, compared to a statistically insignificant 0.04
percentage points in Travis County. Improved job opportunities for construction workers in
Bexar County are associated with more cars in neighborhoods where more construction workers
lived.32
These results suggest that neighborhoods most affected by BRAC and the Hardberger
projects witnessed important improvements in economic conditions. However, they also suggest
that those improvements were not enjoyed by all residents.33 Indeed, the effects appear to be
concentrated among lower income individuals and households. The fact that poverty rates fell,
but median household incomes remained essentially constant suggests that only a subset of
households gained from BRAC, and that many of those households may have been living below
the poverty line.34
It is plausible that some of the observed improvements in economic conditions were driven
by changing neighborhood composition. Also, a large literature in criminology links
neighborhood stability to crime rates (Wikström 1998). While the Decennial Census and ACS do
not explicitly measure disorder or instability, we do observe some indicators of neighborhood
growth and turnover. Regression results relating these indicators to neighborhood employment
32 In results not shown here for sake of space, we find substantively trivial (<0.08 percentage points) and statistically
insignificant (p>0.2) correlations between construction share than the fraction of households with one car or three or
more cars in Bexar County. This highlights who exactly benefited from BRAC: not the poorest households, but also
not particularly wealthy households. In comparison, there is a relatively large (1 percentage point), marginally
precise (p=0.11) increase in any car ownership in block groups with more tourism workers.
33 As we would expect given the requirements of Directive 12, the impacts on neighborhoods do not appear to be
driven by areas with large non-citizen populations. Block groups with more foreign born adults and more
construction workers in 2000 had slightly higher poverty rates by the end of the decade, but slightly smaller
increases in two-vehicle ownership. Both effects, though, are statistically indistinguishable from zero.
34 Notably, the ACS does not contain information on employment or unemployment rates at the block group level,
making it difficult to assess the labor market effects of BRAC.
21
composition are also presented in Table 3. During this time period, the population of block
groups in San Antonio grew by 4% on average, but this growth was roughly 0.9% slower on
average for each additional percentage point increase in construction share. Growth in housing
units was also slower in these neighborhoods. Meanwhile, we do not observe any statistically or
substantively significant differences between construction-intensive communities and other
neighborhoods in the change in the fraction of houses that were occupied or in household
turnover; we create a proxy for the latter using information on the year that residents report
moving into their current dwelling. We see very different patterns in Travis County, where
neighborhoods with more construction workers saw sizable increases in vacancy rates and
household turnover over the course of the decade.
Thus, there is little evidence that those in Bexar County who initially benefited from BRAC
moved to better areas in large numbers or that areas with relatively more construction workers in
2000 experienced differential growth or turnover during the 2000s.35 In fact, we find suggestive
evidence that communities with more construction workers in 2000 were slightly more stable
than other neighborhoods. Therefore, it is unlikely that inflows or outflows of more or less
criminal residents are a primary explanation for any observed changes in crime rates in
construction-intensive neighborhoods over this period. Rather, as the results in Table 3 indicate,
changes in the well-being of a subset of existing residents, and the associated changes in criminal
opportunities for their neighbors, appear to be a potentially more important driver behind
changes in crime.
7.2. BRAC and Criminal Opportunities
After 2005, neighborhoods in Bexar County with more construction workers had lower
poverty rates. Median house values and household income rose by only a small amount. There
was also an increase in the fraction of households with two or more cars. Along with improved
economic conditions in these neighborhoods came increased criminal opportunities.
35 In one robustness test, we use block group employment shares as measured in the 2005-2009 ACS as opposed to
the 2000 Decennial Census. The 2005-2009 shares will largely capture changes in the spatial distribution of workers
in response to BRAC. As discussed in Section 7.3.3, the results change little with this alternative measure, which is
not surprising given that there is a high correlation in the construction share between surveys.
22
Table 4 presents our baseline estimates of the impact of BRAC on property crimes at the
census block group level.36 We estimate that, after BRAC began, each percentage point increase
in construction workers in a block group increased the number of residents who were charged
with burglary by an imprecisely measured 1-2%.37 The overall effect on burglaries, however,
masks underlying heterogeneity in criminal behavior among different individuals within
neighborhoods. Indeed, we see no impact on the number of first-time offenders charged with
burglary, but much larger and statistically significant increases in burglaries committed by
people who were likely ineligible for BRAC jobs. Each percentage point increase in construction
workers in a block group is associated with a 4% increase in the number of burglaries committed
by neighbors who had been charged with felonies, and a nearly 5% increase in burglaries
committed by neighbors who had been previously convicted of felonies. Both estimates are
significant at the 1% level. The effects are typically much smaller and insignificant for burglaries
committed after the Hardberger projects began but before the first BRAC contracts were
awarded. We find more muted changes in burglaries post-2005 in tourism- and health care-
intensive areas; these results are reported in Appendix Table A2.
Car thefts also increased by roughly 2% after BRAC increased the purchasing power of
construction workers. As shown in the second panel of Table 4, while first-time offenders living
near construction workers were no more or less likely to steal a car after 2007, people with
criminal histories were. In particular, those who had been charged with felonies or were
previously convicted of a felony were nearly 3% more likely to steal a car for each percentage
point increase in construction jobs. We do not find evidence that car thefts increased after the
Hardberger projects began and BRAC was announced. Adding unauthorized use of a motor
vehicle to our definition of car theft (in the third panel) slightly reduces the magnitude of the
estimated coefficients. This is to be expected, since UUM is a non-acquisitive crime. Still, after
BRAC, each percentage point increase in the share of construction workers in the 2000 census is
36 We also conducted the analysis at the census tract level. Highlighting the importance of spatial disaggregation in
understanding local crime patterns, we tended to find weaker relationships between criminal activity and
construction employment concentration when we used the tract as the geographic unit of analysis, although the
estimates were qualitatively similar.
37 The estimates are very similar if, instead of including block group fixed effects, we include block group
characteristics and tract fixed effects.These results for each crime type appear in Appendix Tables A2-A9. Notably,
the relationship between block group characteristics and crime rates (not shown) generally conform to expectations;
average education and income levels are negatively related to crime, while the median age of the housing stock and
share of renters are positively related to crime.
23
associated with a roughly 2% increase in the rate of car theft, broadly defined, by people with
criminal histories. We estimate that people without criminal histories are less likely to steal cars
after BRAC, but this reduction in criminal behavior begins during the Hardberger projects.
The fourth panel of Table 4 provides evidence that people who lived near construction
workers were more likely to be charged with larceny after BRAC was announced, and some
evidence that this began with the Hardberger projects. This is again driven by people who had
previously been charged with or convicted of a felony. We also find that people living in
neighborhoods with tourism workers were more likely to be accused of stealing or burglary after
the Hardberger projects increased opportunities for workers in that sector. These impacts are
roughly half the size of those for construction workers (see Appendix Table A5).
Overall, these results are consistent with an increase in criminal opportunities when some
construction workers earn relatively more money. 38 While we do not know that BRAC
construction workers were actually victimized by their neighbors, our results are strongest for
crimes that criminology research suggests occur close to where offenders live: burglary and car
theft.
We turn to people charged with violent crimes in the bottom four panels of Table 4. Robbery
is clearly an acquisitive crime, which we should expect to see increase, while murders, rapes, and
assaults are typically non-acquisitive crimes.39 Overall, we observe less change in violent crime
in neighborhoods that benefited from BRAC. To the extent that there are increases, they are
entirely driven by crimes committed by people living in the same neighborhoods as construction
workers, but who are unlikely to pass the background check required to work on a BRAC
project.
In particular, after BRAC construction began, a one percentage point increase in the share of
neighborhood jobs in construction was associated with a 2% increase in robberies committed by
neighbors who had previously been accused of a felony. We observe a similar magnitude
increase in robberies committed by neighbors with a felony record.
38 These results do not appear to be driven by block groups with a larger number of non-citizen workers, who would
be ineligible to work on BRAC contracts. Coefficients on three-way interactions between construction share, citizen
share, and the Hardberger/BRAC dummies are generally positive, but never statistically significant and small in
magnitude relative to the first order effects.
39 For example, someone who injures another in the course of a robbery would have committed both assault and
robbery. Depending on the facts of the case, assault may be an easier case to prove than the intent to take property,
particularly if the robbery was unsuccessful.
24
There is also some impact on assaults, which rise by about 2% overall and by between 3%
and 4% for accused or convicted felons. People who have never been accused of a felony appear
to be slightly less likely to commit assault after the Hardberger projects began. However, we also
estimate that both tourism and health care concentrations are positively related to assault charges
after BRAC, and the point estimates are essentially the same size as the impact of construction
concentration (Appendix Table A9), making it difficult to interpret this result. Meanwhile,
consistent with these crimes being less driven by economic incentives, we find no evidence that
the neighbors of construction workers are any more or less likely to commit murder or rape after
BRAC spending began. Notably, though, the null effects for these non-acquisitive crimes suggest
that observed increases in property crimes are not merely driven by changes in policing or
cooperation with law enforcement in affected neighborhoods, as such changes would be
expected to affect all crimes in the same way.
7.3. Robustness Tests
7.3.1. Pre-Treatment Trends
With the exception of larceny, it is not obvious that crime rates were trending upwards in San
Antonio prior to the start of BRAC construction. However, if crime was differentially increasing
in neighborhoods with more construction workers, this might bias our results upwards. In order
to address this, we employ a strategy in the spirit of Evans and Owens (2007) and Freedman and
Owens (2011). For each block group, we estimate a model of total crime rates between 2000 and
2006 as a function of a linear time trend.40 We then divided block groups into ten groups of 100
to 101 block groups based on the deciles of their pre-BRAC crime trend and re-estimated
equation (1) with crime trend group-specific year fixed effects. Marginal effects in this
specification are identified off a differential change in the crime rates in construction-heavy
neighborhoods relative to neighborhoods with fewer construction workers, but similar trends in
crime prior to the increase in some construction workers’ spending power.41
As Table 5 shows, our estimates of the impact of BRAC on acquisitive crime are quite robust
to these fixed effects. However, it appears as if the changing relationship between assault and
40 There is a reasonable amount of variation across block groups in the pre-BRAC trends in crime; the block group
in the 1st percentile saw a decrease of 1.05 crimes per 1000 residents per year, and in the 99th percentile block group
saw an increase of 1.38 crimes per 1000 residents per year.
41 Conceptually, total crime trends may better capture changes in general social disorder or policing behavior than
trends in any one specific crime. In any case, our results are very similar when we use crime-specific trends.
25
construction workers was potentially driven by pre-BRAC trends. Overall, there is no
statistically significant correlation with construction share and assault after BRAC, and the
magnitude of the estimated change in criminal activity of those who were ineligible to work on
BRAC construction sites falls by an order of magnitude. In Appendix Table A9, we show that
the estimated relationship between assault and the share of adults working in tourism after
BRAC remains positive and statistically significant even with the inclusion of the crime trend
group-specific year fixed effects. Meanwhile, comparing block groups with similar trends in
crime reduces the estimated relationship between tourism workers and larceny and burglary by
15% to 30%, respectively, while the impact of construction workers on these crimes falls by only
2% to 6%.
7.3.2. Level Analysis
Next, we investigate whether our results are sensitive to functional form, and in particular our
use of logged crime rates. In Table 6, we replace our dependent variable with charges filed per
capita. We find qualitatively similar results for burglary, car theft, and robbery; BRAC was
associated with an increase in criminal behavior in neighborhoods with higher shares of
construction employment, driven by people who were unlikely to have directly benefited from
the development.
Using charges filed per capita, we again no longer observe a statistically significant increase
in the assault rate in construction-intensive neighborhoods. We also find weaker results for
larceny when we assume that changes are better described in rates than in logged rates. While it
is not obvious to us that neighborhoods with a higher share of construction workers should
experience a constant increase in criminal behavior per capita rather than a constant percentage
increase in criminal behavior, the fact that the observed increases in larceny and assault are
sensitive to this functional form assumption suggests that one should not put too much weight on
the results for these crimes. However, after BRAC, neighbors of BRAC beneficiaries were more
likely to engage in robbery, car theft, and burglary relative to those who did not live near
construction workers, in both percentage and level terms.
In additional results (available upon request), we find no relationship between tourism
employment and thefts per capita after BRAC, but do estimate that a one percentage point
increase in the share of tourism workers in 2000 increased burglary rates of convicted and
26
accused felons by 0.009 per 1000 people (se=0.004). This, however, is half the size of the effects
that we find for construction employment.
7.3.3. ACS Employment Shares
In the previous regressions, we used employment shares for construction and other industries
based on 2000 Decennial Census data. However, it is plausible that, potentially in part in
response to new construction projects, the geographic distribution of construction workers
changed by mid-decade, such that 2000 construction shares no longer reflected the communities
most impacted by the new projects. We found little evidence of increased mobility in
construction-heavy neighborhoods, but even a small movement of construction workers would
introduce measurement error into our estimates. Hence, as a robustness test, we calculate the
industry employment shares using the 2005-2009 ACS. These data represent averages between
2005 and 2009, and will capture any reshuffling in the geographic distribution of workers that
may have occurred mid-decade as BRAC projects got underway.
The results of this test, which appear in Table 7, are qualitatively and quantitatively similar to
the main results in Table 4. We continue to see sizable and statistically significant increases in
burglaries, car theft, and larcenies committed by accused and convicted felons. We do see some
increase in these crimes after the Hardberger projects in construction-intensive neighborhoods,
but the BRAC effects are consistently larger in magnitude and higher in precision. As before,
there are much more muted effects on murder and rape. When we use contemporaneous
measures of tourism employment, we no longer observe a statistically or substantively
meaningful change in the acquisitive criminal behavior of accused or convicted criminals. In
fact, we estimate a slight reduction in burglary by first-time offenders in neighborhoods where
more adults work in entertainment, accommodation, and food services.42
42 Because the ACS measures conditions between 2005 and 2009, it is possible that the data do not capture post-
recession patterns of neighborhood and employment change. In an additional robustness test, we instead use
recession-era public-use data from the OnTheMap database, which is constructed and maintained by the
Longitudinal Employer Household Dynamics (LEHD) Program at the U.S. Census Bureau. These data are derived
from state unemployment insurance records and capture over 98% of private sector employment. Confidentiality
protection measures introduce some noise into local estimates of employment, which makes us reluctant to rely on
these data exclusively (see Andersson et al. (2008) for details). However, using 2009 OnTheMap data on resident
employment, we arrive at qualitatively similar results as with the 2005-2009 ACS. Results are available upon
request.
27
7.3.4. Construction Employment per Capita
In our previous specifications, we identified neighborhoods where BRAC workers are most
likely to live using variation in the fraction of workers that are in construction. An alternative
way to identify neighborhoods that benefited from BRAC is to use the number of construction
workers as a share of the total population. While different from typical measures of employment
concentration, measuring construction intensity using workers per capita is more similar to the
focus in crime and economics on police or law enforcement expenditures per capita.
In Table 8, we replicate Table 4, but with all neighborhood employment shares replaced by
neighborhood employment per capita. While our results are slightly less precise, we still observe
statistically significant increases in the propensity of previously accused felons living near
construction workers to engage in burglary, car theft, larceny, and robbery. We observe no
change in the acquisitive behavior of people who were never previously accused of a felony, and
who therefore were plausibly eligible to be BRAC beneficiaries. We also observe no change in
the correlation between construction workers per capita and violent criminal behavior after
BRAC began. Notably, when we measure tourism and health care employment on a per capita
basis rather than as a share of total employment, the correlation between these types of
employment and crime rates is generally negative. In fact, block groups with more health care
workers in 2000 experienced statistically significantly lower rates of burglary and assault after
2007. Also, after BRAC spending began, fewer car thieves lived in block groups with more
tourism workers.
7.3.5. Relaxing the Timing of BRAC
The potential confounding of BRAC with the Hardberger projects also means that we want to
be particularly sensitive to the timing of our “shock” to construction workers. Additionally, one
might still be concerned that the observed patterns are driven by localized impacts of the Great
Recession, and in particular the housing collapse, which took hold nationally at the end of 2007
and dampened construction activity in Bexar county late in the decade (see Figure 3).
In Tables 9 and 10, we present results from a more flexible specification of equation (1).
Instead of dividing our sample into three time periods, we allow for the impact of the presence of
construction workers on the criminal behavior of neighborhood residents to vary in each year
between 2004 and 2010. As Table 9 shows, the relationship between construction workers and
acquisitive criminal behavior appears to have fundamentally changed starting in 2007, prior to
28
the recession but coincident with the start of BRAC spending. While there is some increase in
car theft and burglaries by repeat felons in 2006, the estimated coefficients are typically larger
after BRAC contracts were awarded, and are more likely to be statistically significant. In Table
10, we focus on violent crimes. We also see that the increase in robberies is driven by repeat
felons in 2007 and people who had previously been accused of felonies after 2009. Turning to
assault, we find the unusual result that first-time offenders in construction-heavy neighborhoods
became less likely to commit assault after 2004. There is no strong pre-BRAC trend in the
behavior of people with criminal histories, but the unusual result for first-time offenders suggests
that the observed changes in assaults may be driven by factors other than BRAC.
Based on these results, we conclude that there was an increase in acquisitive crime by
residents living in neighborhoods with more construction workers that only began in earnest as
BRAC awards began to be made and construction hiring increased in 2007. Further, we do not
find strong evidence that pre-treatment trends are driving the observed relationships in our
baseline results; there is no gradual increase in the coefficient estimates prior to BRAC. Instead,
the rise in acquisitive crime in neighborhoods with more construction workers began as federal
contract dollars started flowing to these areas later in the decade.
7.3.6. Differentiating BRAC from the Great Recession
The timing of our estimated impact of construction share on crime suggests that the Great
Recession, which was evident in the Bexar County construction industry by 2009, is not driving
the estimated effects on crime. In one additional test to address this possible concern, we run our
main regressions with an additional control for the interaction between total construction
employment in Bexar County, as measured in the QCEW each year, with the construction share
at the block group level measured in 2000.43 In this specification, we are explicitly controlling
for the fact that, as a result of the housing boom and bust, non-BRAC construction employment
overall rose and fell markedly over the course of the decade.44 The estimated coefficient on
BRAC x Construction Share is therefore interpreted as the change in crime associated with
43 Since the QCEW data only cover 2001-2010, we assume construction employment in 2000 was the same as in
2001. Alternative ways of imputing year 2000 construction employment (e.g., using January 2001 figures or
extrapolating linearly from the 2001-2010 data) yielded similar results, as did dropping year 2000 data.
44 In ongoing research, we are exploiting the Great Recession’s differential impact on neighborhoods due to the
varying composition of residents with respect to their industry and occupation to further explore how changes in
average income and the income distribution within communities affect crime.
29
BRAC spending over and above the change in behavior we might expect to see given other
fluctuations in the private construction market.
On the whole, the results, which appear in Table 11, are similar to those in Table 4. While the
impacts on car theft become slightly smaller and less precise, the effects on larceny and robbery
among accused and convicted felons become larger in magnitude. The impacts on burglary,
assault, murder, and rape are all very similar to the previous results, with the former two
increasing with the local construction share and the latter two unrelated to (or even decreasing
modestly with) the local construction share. Once again, the results point primarily to a rise in
acquisitive crime in parts of Bexar County where some, but not all, residents enjoyed improved
labor market conditions as a result of a surge in federal construction dollars under BRAC.
8. Conclusion
We take advantage of a positive economic shock to one particular group of workers in San
Antonio, Texas to provide new evidence on the relationship between relative income and crime.
The 2005 BRAC dramatically increased wages and employment opportunities for construction
workers in San Antonio who were in the United States legally and who did not have criminal
records. Using a unique data set of the residence of people accused of committing felonies in
Bexar County and detailed, block group-level information on employment and other
neighborhood characteristics from the Census Bureau, we provide evidence that an important
outcome of BRAC was an increase in criminal opportunities. Specifically, people living in block
groups with more construction workers were actually more likely to be accused of burglary and
car theft after the job prospects for these workers improved. These results are robust to using a
log or level specification and do not appear to be driven either by pre-treatment trends or by the
recession and associated collapse of the housing market later in the decade. While we do not
know whether or not these accused felons were construction workers, this increase in criminal
behavior is driven by people who, based on their criminal histories, were unlikely to be working
for federal contractors. We find some evidence that the neighbors of BRAC beneficiaries were
more likely to engage in larceny and robbery, but these results are less robust to variation in
modeling choice.
Importantly, acquisitive crime is not as socially costly as murder or sexual assault. Using
conservative estimates of the cost of crime in Miller et al. (1996), the official DoD estimates of
30
the economic impact of BRAC construction are only 0.07% too high; instead of an overall
economic impact of $2.6682 per federal dollar spent, Bexar County gained $2.6664 on net.45
More recent estimates of the cost of crime (Heaton 2010) increase the estimated multiplier gap to
0.2%, implying that each dollar spent on BRAC construction provided a $2.6629 boost to the
San Antonio economy.
While the costs associated with greater criminal activity in Bexar County pale in comparison
to the overall economic impact of BRAC, the fact that acquisitive crime rates increased in
neighborhoods where the economic conditions of residents were improving on average has
important policy implications. In particular, place-based economic programs that only benefit
certain residents may have perverse effects on crime rates. Though such effects appear to be
small for BRAC, they could be much larger for programs that induce longer lasting changes in
the economic circumstances of a particular subset of the population. Overall, our findings
suggest that income inequality, rather than simply average income, deserves careful attention
when estimating the criminal justice impacts of any policy that has implications for local
economies.
Acknowledgements
We would like to thank Julian Christia, Philip Cook, Dhaval Dave, William Evans, Beomsoo
Kim, Alejandro Gaveria, Naci Mocan, Timothy Moore, Daniel Ortega, Sarah Pearlman, Steven
Raphael, Daniel Rees, and Seth Sanders as well as participants in the 2012 LACEA AL
CAPONE Meetings, the 2012 Western Economic Association Conference, and the 2012 NBER
Summer Institute Crime Workshop for helpful comments. We would also like to thank the staff
of the Bexar County District Court for many patient explanations of the information in the felony
cases database. This research was made possible in part through the use of Cornell University’s
Social Science Gateway, which is funded through NSF Grant 0922005. All errors are our own.
45 The DoD awarded $2,514,410,000 in construction contracts (in 2006 dollars) and estimated that the total
economic benefit of that construction spending was $6,708,877,333, implying a local economic multiplier of 2.6682.
The estimated costs of burglary, car theft, and robbery are $1,745, $4,614, and $23,693, respectively, in Miller et al.
(1996), and $12,733, $8,826, and $65,414 in Heaton (2010). We estimate the total loss due to crime by multiplying
these values by the coefficient estimates in Table 6, then by the average 2000 construction share of 8.84, and then
the 2000 Bexar County population of 1.393 million.
31
References
Andersson, Fredrik, Matthew Freedman, Marc Roemer, and Lars Vilhuber. 2008. “LEHD
OnTheMap Technical Documentation.” Technical Paper DATA-OTM-2.0.3, U.S. Census
Bureau.
Becker, Gary. 1968. “Crime and Punishment: An Economic Approach.” Journal of Political
Economy 76(2): 169-217
Bernasco, Wim, Richard Block, and Stijn Ruiter. 2012. “Go Where the Money Is: Modeling
Street Robbers’ Location Choices.” Journal of Economic Geography 12(2): 1-25.
Bjerk, David. 2010. “Thieves, Thugs, and Neighborhood Poverty.” Journal of Urban Economics
68(3): 231-246.
Brantingham, Paul J. and Patricia L. Brantingham. 1981. “Notes on the Geometry of Crime.” In
Environmental Criminology, eds. Paul.J. Brantingham and Patricia.L. Brantingham. Beverly
Hills, CA: Sage.
Brush, Jesse. 2007. “Does Income Inequality Lead to More Crime? A Comparison of Cross-
Sectional and Time-Series Analyses of United States Counties.” Economics Letters 96(2):
264-268.
Bushway, Shawn and Peter Reuter. 2001. “Labor Markets and Crime.” In Crime, 3rd Edition,
eds. James Q. Wilson and Joan Petersilia. Richmond, CA: ICS Press.
Canter, David and Donna Youngs. 2009. Investigative Psychology: Offender Profiling and the
Analysis of Criminal Action. West Sussex: John Wiley & Sons.
Cantor, David and Kenneth.C. Land. 1985. “Unemployment and Crime Rates in the Post-World
War II United States: A Theoretical and Empirical Analysis.” American Sociological Review
50(3): 317-332.
Cohen, Lawrence and Marcus Felson. 1979. “Social Change and Crime Rate Trends: A Routine
Activity Approach.” American Sociological Review 44(4): 588-605.
Cook, Philip J. 1986. “The Demand and Supply of Criminal Opportunities.” Crime and Justice
7:1-27.
Cook, Philip J. and John MacDonald. 2011. “Public Safety through Private Action: An
Economic Assessment of BIDs.” Economic Journal 121(552): 445-462.
Cook, Philip J., Anthony A. Braga, and Mark H. Moore. 2011. “Gun Control” In Crime and
Public Policy, eds. James Q. Wilson and Joan Petersilia. Oxford: Oxford University Press.
Donohue, John and Steven Levitt. 2001. “The Impact of Legalized Abortion on Crime.”
Quarterly Journal of Economics 116(2): 379-420
Durlauf, Steven and Daniel Nagin. 2011. “Imprisonment and Crime: Can Both Be Reduced?”
Criminology and Public Policy 10(1): 13-54.
Ehrlich , Isaac. 1973. “Participation in Illegitimate Activities: A Theoretical and Empirical
Investigation.” Journal of Political Economy 81(3): 521-565.
Eriksen, Michael and Stuart Rosenthal. 2010. “Crowd Out Effects of Place-Based Subsidized
Rental Housing: New Evidence from the LIHTC Program.” Journal of Public Economics
94(11-12): 953-966.
Fagan, Jeffrey and Richard Freeman. 1999. “Crime and Work.” In Crime and Justice, Vol. 25,
ed. Michael Tonry. Chicago: University of Chicago.
Fajnzylber, Pablo, Daniel Lederman, and Norman Loayza. 2002. “What Causes Violent Crime?”
European Economic Review 46(7): 1323-1357.
Freedman, Matthew. 2012. “Teaching New Markets Old Tricks: The Effects of Subsidized
Investment on Low-Income Neighborhoods.” Journal of Public Economics, forthcoming.
32
Freedman, Matthew and Emily Owens. 2011. “Low-Income Housing Development and Crime.”
Journal of Urban Economics 70(2-3): 115-131.
Glaeser, Edward and Joshua Gottlieb. 2008. “The Economics of Place-Making Policies.”
Brookings Papers on Economic Activity Spring: 155-239.
Glaeser, Edward and Bruce Sacerdote. 1999. “Why Is There More Crime in Cities?” Journal of
Political Economy 107(6): S225-S258.
Greater San Antonio Chamber of Commerce. 2011. “2011 Quarterly Economic Trends, First
Quarter.” http://sachamber.org/cwt/external/wcpages/publications/Economic_Reports.aspx.
Grossman, Herschel and Minseong Kim. 1995. “Swords or Plowshares? A Theory of the
Security of Claims to Property.” Journal of Political Economy 103(6): 1275-1288.
Harbaugh, William, Naci Mocan, and Michael Visser. 2011. “Theft and Deterrence.” NBER
Working Paper 17059.
Heaton, Paul. 2010. “Hidden in Plain Sight: What Cost-of-Crime Research Can Tell Us About
Investing in Police.” RAND Corporation OP-279.
Heffetz, Ori. 2011. “A Test of Conspicuous Consumption: Visibility and Income Elasticities.”
The Review of Economics and Statistics 93(4): 1101-1117.
Hsieh, Ching-Chi and M.D. Pugh. 1993. “Poverty, Inequality, and Violent Crime: A Meta-
analysis of Recent Aggregate Data Studies.” Criminal Justice Review 18(2): 182-202
Kelly, Morgan. 2000. “Inequality and Crime.” Review of Economics and Statistics 82(4): 530-
539.
Kling, Jeffrey, Jens Ludwig, and Lawrence Katz. 2005. “Neighborhood Effects on Crime for
Female and Male Youth: Evidence from a Randomized Housing Voucher Experiment.”
Quarterly Journal of Economics 120(1): 87-130.
Lochner, Lance and Enrico Moretti. 2004. “The Effect of Education on Crime: Evidence from
Prison Inmates, Arrests, and Self-Reports.” American Economic Review 94 (1): 155-189.
Machin, Stephen and Olivier Marie. 2006. “Crime and Benefit Sanctions.” Portuguese Economic
Journal 5(2): 149-165.
Machin, Stephen and Costas Meghir. 2004. “Crime and Economic Incentives.” Journal of
Human Resources 39(4): 958-979.
Miller, Ted, Mark Cohen, and Brian Wiersema. 1996. “Victim Costs and Consequences: A New
Look.” National Institute of Justices, U.S. Department of Justice.
Mocan, Naci and Daniel Rees. 2005. “Economic Conditions, Deterrence and Juvenile Crime:
Evidence from Micro Data.” American Law and Economics Review 7(2): 319-349.
Mustard, David. 2010. “How Do Labor Markets Affect Crime? New Evidence on an Old
Puzzle.” IZA Discussion Paper 4856.
Nirvin, Steve. 2009. “Fort Sam Houston Economic Impact Update.” Office of Military Affairs,
Military Transformation Task Force. http://www.sanantonio.gov/oma/mttf.aspx.
Phillips, Phillip.1980. “Characteristics and Topology of the Journey to Crime.” In Crime: A
Spatial Perspective, eds. D.E. Georges-Abeyie and K.D. Harries. New York: Columbia
University Press.
Piehl, Anne. 1998. “Economic Conditions, Work and Crime.” In Handbook of Crime and
Punishment, ed. Michael Tonry. New York: Oxford University Press.
Puentes, Robert and Peter McFerrin. 2012. “The Intersection of Place and the Economy.” The
Brookings Institution, What Works Collaborative.
Raphael, Steven and Rudolf Winter-Ebmer. 2001. “Identifying the Effect of Unemployment on
Crime.” Journal of Law and Economics 44(1): 259-283.
33
Rhodes, William and Catherine Conley. 1981. “Crime and Mobility: An Empirical Study.” In
Environmental Criminology, eds. Paul.J. Brantingham and Patricia.L. Brantingham. Beverly
Hills, CA: Sage.
Saridakis, George. 2004. “Violent Crime in the United States of America: A Time-Series
Analysis Between 1960-2000." European Journal of Law and Economics 18(2): 203-221.
Sorenson, David. 2007. Military Base Closure: A Reference Handbook. Westport, CT:
Greenwood.
Thomas, Michael. 2009. “BRAC Work Providing City Multi-Billion Dollar Shot in the Arm.”
San Antonio Business Journal, September 3.
United States Government Accountability Office. 2007. “Selected Agencies Use of Criminal
Background Checks for Determining Responsibility” GAO-07-215R.
Wikström, Per-Olof H. 1998. “Communities and Crime.” In The Handbook of Crime and
Punishment, ed. Michael Tonry. New York, NY: Oxford University Press.
Wiles, Paul and Andrew Costello. 2000. “The Road to Nowhere: The Evidence for Travelling
Criminals.” UK Home Office Research Study 207.
Wilson, James Q. and Richard J. Herrnstein. 1985. Crime and Human Nature. New
York: Simon and Schuster.
34
Figure 1: Spatial Concentration of Construction Jobs in Bexar County, by Block Group
2000 (Decennial Census) 2005-2009 (ACS)
Average Share: 8.8% Average Share: 10.0%
Fraction with 15%+: 16.3% Fraction with 15%+: 22.7%
Construction Share < 2.5%
Construction Share 2.5-5%
Construction Share 5-7.5%
Construction Share 7.5-10%
Construction Share 10-12.5%
Construction Share 12.5%-15%
Construction Share 15%+
Block Group Boundaries
35
Figure 2: QCEW Relative Wages and Excess Jobs in Construction, Bexar vs. Travis County
Note: Includes private-sector employment in each county.
BRAC Announced,
Hardberger Elected
First BRAC Contract
8000 9000 10000 11000 12000
Bexar Jobs - Travis Jobs
.84 .86 .88 .9 .92
Bexar Wage / Travis Wage
2000 2002 2004 2006 2008 2010
Year
Relative Wages Excess Jobs
36
Figure 3: QCEW Employment in Bexar County in Construction, Tourism, and Health Care
Note: Includes private-sector employment in each industry.
Figure 4: QCEW Weekly Wages in Bexar County in Construction, Tourism, and Health Care
Note: Includes private-sector employment in each industry.
BRAC Announced,
Hardberger Elected
First BRAC Contract
60 70 80 90 100
Tourism and Health (1000s)
34 36 38 40 42 44
Construction (1000s)
2000 2002 2004 2006 2008 2010
Year
Construction Tourism Health
BRAC Announced,
Hardberger Elected
First BRAC Contract
200 400 600 800 1000
Average Weekly Wages ($)
2000 2002 2004 2006 2008 2010
Year
Const ruction Tourism He alt h
37
Figure 5: Acquisitive Crimes in Bexar County
Figure 6: Non-Acquisitive Crimes in Bexar County
Note: Murder and rape rates are multiplied by ten.
BRAC Announced,
Hardberger Elected
First BRAC Contract
0.2 .4 .6 .8 1
Offenses/1000 Population
2000 2002 2004 2006 2008 2010
Year
Burglary Car Theft Larceny Robbery
BRAC Announced,
Hardberger Elected
First BRAC Contract
.5 11.5 22.5
Offenses/1000 Population
2000 2002 2004 2006 2008 2010
Year
Murder Rape Assault
38
Table 1: Coverage of Bexar County Felony Data
A. Felony Charges Filed per 100 UCR Index Crimes
Property Crimes Violent Crimes
Burglary Car Theft1 Larceny Robbery Murder Assault Rape2
2000 3.66 7.04 0.70 17.32 68.69 10.17 38.17
2001 3.31 5.63 0.86 16.39 85.05 9.36 41.09
2002 3.23 5.32 0.87 13.09 64.91 7.88 37.29
2003 3.53 4.74 1.16 16.09 72.00 13.20 27.89
2004 3.51 6.70 1.10 14.62 75.00 11.83 28.44
2005 3.45 6.30 1.11 17.84 82.47 12.67 28.70
2006 3.61 7.25 1.38 17.21 49.23 12.75 29.10
2007 3.42 6.61 1.49 18.17 95.45 21.63 27.30
2008 3.92 8.16 1.53 20.81 73.91 17.58 28.86
2009 4.78 7.75 1.33 20.99 80.34 31.00 25.50
B. Felony Charges Filed per 100 Adults Arrested
Property Crimes Violent Crimes
Burglary Car Theft1 Larceny Robbery Murder Assault Rape2
2000 85.29 - 33.57 112.32 219.35 121.92 51.28
2001 93.01 - 37.54 111.64 260.00 130.80 62.97
2002 110.15 132.94 20.53 137.67 180.49 113.13 79.38
2003 86.82 87.63 16.37 93.65 124.14 91.08 43.03
2004 79.31 69.00 11.55 75.17 98.73 76.08 47.67
2005 72.75 71.83 11.61 83.88 121.21 73.47 29.12
2006 72.04 73.57 12.92 75.31 98.46 67.91 30.34
2007 82.25 77.80 12.57 80.61 150.00 93.52 40.37
2008 93.66 102.14 13.36 87.04 104.08 104.34 24.61
2009 85.69 85.14 10.29 84.33 91.26 115.84 31.26
Notes: Authors’ calculations from Uniform Crime Reports County-Level Detailed Arrest and Offense Data and
Bexar County District Court felony filings.
1 Felony car theft charges include unauthorized use of a motor vehicle. Data on car theft arrests for 2000 and 2001
are excluded due to clear underreporting in UCR arrest data.
2 Felony rape charges include sexual assault. Rape arrests include arrests for sexual offenses.
39
Table 2: Descriptive Statistics
Low Construction Share High Construction Share
Employment Shares (2000)
Share in Construction 0.04 0.14
Share in Tourism 0.10 0.12
Share in Health Care 0.13 0.11
Demographic & Housing Characteristics (2000)
Population 1553 1209
Share Black 0.09 0.06
Share Hispanic 0.45 0.73
Share Male 0.47 0.49
Share Under Age 30 0.43 0.47
Share Age 65 or Over 0.13 0.12
Share HHs Speak Spanish* 0.39 0.66
Share Foreign Born 0.09 0.15
Share in Same House 1 Year Ago 0.51 0.59
Share with HS Degree 0.23 0.27
Share with Some College 0.25 0.19
Share with College Degree 0.34 0.14
Unemployment Rate* 0.06 0.08
Labor Force Participation Rate 0.64 0.56
Poverty Rate 0.14 0.24
Median HH Income 44,959 30,352
Employment to Pop. Ratio† 0.45 0.38
Housing Units 608 425
Share Units Vacant* 0.06 0.07
Share Units Owner-Occupied* 0.63 0.63
Median House Value* 92,975 54,435
Median House Age 33.22 37.59
Share HHs with 2+ Vehicles 0.54 0.47
Demographic & Housing Characteristics (2005-2009)
Poverty Rate 0.17 0.24
Median HH Income* 52,489 35858
Employment to Pop. Ratio† 0.45 0.41
Median House Value* 136,413 81,859
Share HHs with 2+ Vehicles 0.54 0.49
Observations (2000, 2005-2009) 504 505
Crime Rates (2000-2010)
Burglary Rate 0.472 0.770
Car Theft Rate 0.119 0.184
Larceny Rate 0.638 0.921
Robbery Rate 0.330 0.470
Murder Rate 0.060 0.098
Assault Rate 0.735 1.017
Rape Rate 0.135 0.189
Observations (2000-2010) 5544 5555
Notes: * Missing one or more observations in 2000 Decennial Census and/or 2005-2009 ACS data.
†Employment to population ratio calculated as total employment divided by total population
(including persons with ages less than 16).
40
Table 3: Neighborhood Outcomes and Construction Workers in Bexar and Travis Counties, 2000 to 2005-2009
Poverty
Rate
(%)
Log
Median
HH
Income
Log
Median
House
Value
HHs with
2+
Vehicles
(%)
Log
Population
Log
Housing
Units
Housing
Units
Occupied
(%)
HHs
Moving in
> 5 Years
Ago (%)
A. Bexar County
Percentage in Construction -0.259** 0.002 0.005** 0.215* -0.009*** -0.006** -0.069 -0.065
x BRAC/Hardberger [0.110] [0.002] [0.002] [0.120] [0.004] [0.003] [0.084] [0.120]
Percentage in Tourism 0.045 0.002 0.006** 0.122 -0.012** -0.008** 0.003 -0.149
x BRAC/Hardberger [0.118] [0.003] [0.003] [0.126] [0.005] [0.004] [0.101] [0.144]
Percentage in Health Care -0.033 0.001 0.002 0.074 -0.009 -0.004 0.007 -0.242
x BRAC/Hardberger [0.151] [0.003] [0.003] [0.148] [0.006] [0.005] [0.101] [0.171]
Observations 2018 2016 1972 2017 2018 2017 2017 2017
R-Squared 0.803 0.909 0.948 0.846 0.912 0.935 0.662 0.835
B. Travis County
Percentage in Construction 0.055 -0.001 0.001 0.038 0.00002 0.005 -0.168* 0.342**
x BRAC/Hardberger [0.120] [0.003] [0.004] [0.179] [0.004] [0.004] [0.100] [0.142]
Percentage in Tourism 0.166 0.001 0.016*** 0.054 -0.020*** -0.015** -0.205* 0.0001
x BRAC/Hardberger [0.159] [0.005] [0.005] [0.203] [0.006] [0.007] [0.124] [0.191]
Percentage in Health Care 0.018 -0.002 0.004 0.063 -0.009 -0.009 -0.017 0.078
x BRAC/Hardberger [0.175] [0.007] [0.005] [0.252] [0.006] [0.011] [0.180] [0.320]
Observations 1012 1005 966 1010 1016 1010 1010 1010
R-Squared 0.898 0.926 0.955 0.893 0.922 0.936 0.643 0.872
Block Group Fixed Effects Y Y Y Y Y Y Y Y
Year Fixed Effects Y Y Y Y Y Y Y Y
Notes: Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05, ***p<0.01.
41
Table 4: Fixed Effects Estimates of Crime and Construction Workers in Bexar
County
All First Time
Accused
Felons Felons
Burglary
Percentage in Con. x Hardberger -0.008 -0.013 0.004 0.015
[0.013] [0.012] [0.010] [0.010]
Percentage in Con. x BRAC 0.015 -0.008 0.043*** 0.047***
[0.013] [0.011] [0.010] [0.009]
Car Theft
Percentage in Con. x Hardberger -0.005 -0.011 0.007 0.009
[0.009] [0.007] [0.007] [0.007]
Percentage in Con. x BRAC 0.018** -0.003 0.028*** 0.026***
[0.009] [0.006] [0.007] [0.007]
Car Theft + Unauthorized Use of a Motor Vehicle
Percentage in Con. x Hardberger -0.027** -0.031*** 0.002 0.002
[0.011] [0.009] [0.008] [0.008]
Percentage in Con. x BRAC -0.002 -0.020** 0.020*** 0.017**
[0.009] [0.008] [0.007] [0.007]
Larceny
Percentage in Con. x Hardberger 0.005 -0.021* 0.022* 0.020*
[0.014] [0.012] [0.012] [0.012]
Percentage in Con. x BRAC 0.029** -0.004 0.045*** 0.047***
[0.012] [0.012] [0.010] [0.010]
Robbery
Percentage in Con. x Hardberger 0.015 0.010 0.0005 -0.0004
[0.012] [0.011] [0.009] [0.008]
Percentage in Con. x BRAC 0.025** 0.001 0.023*** 0.020***
[0.010] [0.009] [0.008] [0.007]
Murder
Percentage in Con. x Hardberger -0.008 -0.008 0.000001 -0.00003
[0.006] [0.005] [0.005] [0.004]
Percentage in Con. x BRAC -0.003 -0.005 0.001 0.005
[0.006] [0.005] [0.004] [0.004]
Rape
Percentage in Con. x Hardberger -0.011 -0.013 0.002 -0.0005
[0.010] [0.008] [0.005] [0.005]
Percentage in Con. x BRAC -0.008 -0.011 0.002 -0.005
[0.007] [0.007] [0.004] [0.004]
Assault
Percentage in Con. x Hardberger -0.024 -0.026* 0.003 0.004
[0.016] [0.014] [0.011] [0.011]
Percentage in Con. x BRAC 0.017 0.007 0.029** 0.035***
[0.012] [0.011] [0.011] [0.011]
Employment Interactions Y Y Y Y
Year Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y
Observations 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population).
Employment interactions include tourism employment share interacted with Hardberger and BRAC
dummies as well as health care employment share interacted with Hardberger and BRAC dummies.
Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05,
***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal history.
42
Table 5: Fixed Effects Estimates of Crime and Construction Workers in Bexar
County, Including Block Group Pre-Treatment Trend by Year Fixed Effects
All First Time
Accused
Felons Felons
Burglary
Percentage in Con. x Hardberger -0.014 -0.015 -0.003 0.007
[0.013] [0.013] [0.010] [0.011]
Percentage in Con. x BRAC 0.015 -0.004 0.037*** 0.041***
[0.013] [0.011] [0.010] [0.009]
Car Theft
Percentage in Con. x Hardberger -0.010 -0.014** 0.004 0.007
[0.009] [0.007] [0.007] [0.007]
Percentage in Con. x BRAC 0.015* -0.003 0.024*** 0.023***
[0.009] [0.006] [0.007] [0.007]
Larceny
Percentage in Con. x Hardberger 0.007 -0.017 0.02 0.019
[0.014] [0.013] [0.012] [0.012]
Percentage in Con. x BRAC 0.032*** -0.001 0.045*** 0.046***
[0.012] [0.012] [0.010] [0.010]
Robbery
Percentage in Con. x Hardberger 0.007 0.005 -0.006 -0.007
[0.012] [0.011] [0.009] [0.008]
Percentage in Con. x BRAC 0.024** 0.002 0.021*** 0.019**
[0.010] [0.009] [0.008] [0.008]
Murder
Percentage in Con. x Hardberger -0.010 -0.009* -0.001 -0.001
[0.006] [0.005] [0.005] [0.004]
Percentage in Con. x BRAC -0.003 -0.003 0.000 0.005
[0.006] [0.005] [0.004] [0.004]
Rape
Percentage in Con. x Hardberger -0.011 -0.013 0.002 0.001
[0.010] [0.009] [0.005] [0.005]
Percentage in Con. x BRAC -0.008 -0.01 0.001 -0.005
[0.008] [0.007] [0.004] [0.004]
Assault
Percentage in Con. x Hardberger 0.012 0.009 0.003 0.002
[0.017] [0.016] [0.004] [0.003]
Percentage in Con. x BRAC 0.022 0.019 0.003 0.004
[0.019] [0.017] [0.007] [0.005]
Employment Interactions Y Y Y Y
Year Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y
2000-2006 Crime Trend x Year Fixed
Effects Y Y Y Y
Observations 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population).
Employment interactions include tourism employment share interacted with Hardberger and BRAC
dummies as well as health care employment share interacted with Hardberger and BRAC dummies.
Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05,
***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal history.
43
Table 6: Fixed Effects Estimates of Crime and Construction Workers in Bexar
County, Level Analysis
All First Time
Accused
Felons Felons
Burglary
Percentage in Con. x Hardberger -0.003 -0.005 0.002 0.003
[0.005] [0.004] [0.003] [0.003]
Percentage in Con. x BRAC 0.015*** -0.004 0.019*** 0.017***
[0.006] [0.004] [0.004] [0.004]
Car Theft
Percentage in Con. x Hardberger 0.001 -0.002 0.004 0.004
[0.003] [0.001] [0.002] [0.003]
Percentage in Con. x BRAC 0.005* -0.001 0.006*** 0.006***
[0.003] [0.002] [0.002] [0.002]
Larceny
Percentage in Con. x Hardberger 0.001 -0.007* 0.008 0.008
[0.007] [0.004] [0.005] [0.005]
Percentage in Con. x BRAC 0.010 -0.002 0.012 0.011
[0.011] [0.005] [0.008] [0.008]
Robbery
Percentage in Con. x Hardberger 0.006 0.004 0.002 0.001
[0.005] [0.004] [0.002] [0.002]
Percentage in Con. x BRAC 0.013** 0.001 0.011** 0.009*
[0.006] [0.003] [0.006] [0.006]
Murder
Percentage in Con. x Hardberger -0.001 -0.001 0.0003 -0.0004
[0.002] [0.001] [0.001] [0.001]
Percentage in Con. x BRAC 0.002 0.001 0.001 0.002**
[0.002] [0.002] [0.001] [0.001]
Rape
Percentage in Con. x Hardberger -0.003 -0.003 -0.0004 -0.001
[0.002] [0.002] [0.001] [0.001]
Percentage in Con. x BRAC -0.004 -0.004 -0.0002 -0.002*
[0.003] [0.002] [0.001] [0.001]
Assault
Percentage in Con. x Hardberger 0.009 0.005 0.003 0.003
[0.014] [0.013] [0.004] [0.004]
Percentage in Con. x BRAC 0.024 0.018 0.006 0.006
[0.016] [0.013] [0.007] [0.006]
Employment Interactions Y Y Y Y
Year Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y
Observations 11099 11099 11099 11099
Notes: Dependent variables are people charged with felonies committed in year/1000 population.
Employment interactions include tourism employment share interacted with Hardberger and BRAC
dummies as well as health care employment share interacted with Hardberger and BRAC dummies.
Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05,
***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal history.
44
Table 7: Fixed Effects Estimates of Crime and Construction Workers in Bexar
County, ACS-Based Estimates
All First Time
Accused
Felons Felons
Burglary
Percentage in Con. x Hardberger 0.012 0.004 0.011* 0.013**
[0.008] [0.008] [0.006] [0.006]
Percentage in Con. x BRAC 0.024*** 0.004 0.033*** 0.034***
[0.008] [0.007] [0.006] [0.006]
Car Theft
Percentage in Con. x Hardberger 0.004 -0.005 0.009* 0.009*
[0.006] [0.004] [0.005] [0.005]
Percentage in Con. x BRAC 0.011* 0.005 0.009* 0.010**
[0.006] [0.004] [0.005] [0.005]
Larceny
Percentage in Con. x Hardberger -0.001 -0.017* 0.017** 0.014*
[0.009] [0.009] [0.008] [0.008]
Percentage in Con. x BRAC 0.009 -0.008 0.024*** 0.024***
[0.007] [0.007] [0.008] [0.007]
Robbery
Percentage in Con. x Hardberger 0.001 0.001 0.001 0.002
[0.007] [0.008] [0.007] [0.006]
Percentage in Con. x BRAC 0.004 0.003 -0.008 0.009
[0.007] [0.007] [0.006] [0.005]
Murder
Percentage in Con. x Hardberger -0.002 -0.004 0.003 0.002
[0.005] [0.004] [0.003] [0.003]
Percentage in Con. x BRAC 0.0001 -0.002 0.001 0.001
[0.004] [0.003] [0.003] [0.002]
Rape
Percentage in Con. x Hardberger 0.003 -0.002 0.006* 0.003
[0.007] [0.006] [0.004] [0.003]
Percentage in Con. x BRAC -0.003 -0.005 0.002 0.0002
[0.006] [0.005] [0.003] [0.003]
Assault
Percentage in Con. x Hardberger -0.024 -0.026* 0.003 0.004
[0.016] [0.014] [0.011] [0.011]
Percentage in Con. x BRAC 0.017 0.007 0.029** 0.035***
[0.012] [0.011] [0.011] [0.011]
Employment Interactions Y Y Y Y
Year Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y
Observations 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population).
Employment interactions include tourism employment share interacted with Hardberger and BRAC
dummies as well as health care employment share interacted with Hardberger and BRAC dummies.
Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05,
***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal history.
45
Table 8: Fixed Effects Estimates of Crime and Construction Workers in Bexar
County, Employment per Capita
All First Time
Accused
Felons Felons
Burglary
Con. Per Capita x Hardberger -0.041 -0.031 -0.012 0.025
[0.039] [0.036] [0.030] [0.029]
Con. per Capita x BRAC 0.004 -0.014 0.066** 0.087***
[0.037] [0.032] [0.029] [0.027]
Car Theft
Con. Per Capita x Hardberger -0.046* -0.043** -0.004 -0.003
[0.026] [0.020] [0.020] [0.019]
Con. per Capita x BRAC 0.009 -0.03 0.049** 0.043**
[0.025] [0.019] [0.020] [0.019]
Larceny
Con. Per Capita x Hardberger 0.052 -0.031 0.066* 0.053
[0.043] [0.038] [0.035] [0.034]
Con. per Capita x BRAC 0.090*** 0.009 0.096*** 0.101***
[0.034] [0.032] [0.029] [0.028]
Robbery
Con. Per Capita x Hardberger 0.051 0.046 -0.010 -0.006
[0.037] [0.034] [0.025] [0.024]
Con. per Capita x BRAC 0.065** 0.012 0.043* 0.030
[0.029] [0.024] [0.022] [0.020]
Murder
Con. Per Capita x Hardberger -0.018 -0.018 -0.004 -0.003
[0.018] [0.014] [0.013] [0.011]
Con. per Capita x BRAC -0.001 -0.007 0.004 0.012
[0.017] [0.015] [0.010] [0.010]
Rape
Con. Per Capita x Hardberger -0.018 -0.018 -0.004 -0.003
[0.018] [0.014] [0.013] [0.011]
Con. per Capita x BRAC -0.001 -0.007 0.004 0.012
[0.017] [0.015] [0.010] [0.010]
Assault
Con. Per Capita x Hardberger -0.074 -0.058 -0.036 -0.016
[0.047] [0.041] [0.030] [0.030]
Con. per Capita x BRAC 0.017 0.020 0.026 0.041
[0.039] [0.034] [0.034] [0.032]
Employment Interactions Y Y Y Y
Year Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y
Observations 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population).
Employment interactions include tourism employment per capita interacted with Hardberger and
BRAC dummies as well as health care employment per capita interacted with Hardberger and BRAC
dummies. Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10,
**p<0.05, ***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal history.
46
Table 9: Fixed Effects Estimates of Property Crime and Construction Workers in Bexar County, Relaxing the Timing of BRAC
Burglary Car Theft Larceny
All First
Time
Accused
Felons Felons All First
Time
Accused
Felons Felons All First
Time
Accused
Felons Felons
Percentage in Con. x 2004 0.009 -0.003 0.026** 0.021 -0.004 -0.010 0.007 0.004 0.018 0.000 0.026* 0.034**
[0.018] [0.017] [0.013] [0.013] [0.011] [0.010] [0.008] [0.008] [0.017] [0.017] [0.016] [0.015]
Percentage in Con. x 2005 -0.009 -0.011 -0.001 0.007 -0.020* -0.017* -0.004 -0.002 0.001 -0.017 0.018 0.02
[0.018] [0.017] [0.013] [0.012] [0.011] [0.009] [0.008] [0.008] [0.016] [0.016] [0.016] [0.015]
Percentage in Con. x 2006 -0.004 -0.016 0.02 0.032** 0.008 -0.009 0.020* 0.021* 0.016 -0.025 0.036** 0.032**
[0.017] [0.016] [0.014] [0.014] [0.015] [0.010] [0.012] [0.012] [0.019] [0.018] [0.016] [0.016]
Percentage in Con. x 2007 0.033* 0.004 0.048*** 0.055*** 0.005 -0.01 0.018* 0.017** 0.041** 0.015 0.035** 0.040**
[0.019] [0.018] [0.015] [0.015] [0.013] [0.010] [0.010] [0.009] [0.019] [0.019] [0.016] [0.016]
Percentage in Con. x 2008 0.009 -0.01 0.054*** 0.051*** 0.042*** -0.007 0.059*** 0.057*** 0.025 -0.007 0.051*** 0.057***
[0.018] [0.016] [0.016] [0.016] [0.015] [0.010] [0.014] [0.014] [0.018] [0.018] [0.017] [0.017]
Percentage in Con. x 2009 0.017 -0.009 0.045** 0.054*** -0.003 -0.015 0.014 0.014 0.048*** -0.012 0.073*** 0.071***
[0.019] [0.017] [0.017] [0.017] [0.014] [0.010] [0.010] [0.011] [0.018] [0.018] [0.016] [0.016]
Percentage in Con. x 2010 0.008 -0.021 0.045*** 0.042*** 0.026* 0.011 0.026** 0.017* 0.016 -0.012 0.041** 0.045***
[0.020] [0.018] [0.016] [0.016] [0.014] [0.012] [0.011] [0.010] [0.019] [0.016] [0.017] [0.017]
Employment Interactions Y Y Y Y Y Y Y Y Y Y Y Y
Year Effects Y Y Y Y Y Y Y Y Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Observations 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Employment interactions include tourism employment share
interacted with Hardberger and BRAC dummies as well as health care employment share interacted with Hardberger and BRAC dummies. Standard errors in
brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05, ***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal
history.
47
Table 10: Fixed Effects Estimates of Violent Crime and Construction Workers in Bexar County, Relaxing the Timing of BRAC
Robbery Murder Rape Assault
All First
Time
Accused
Felons Felons All First
Time
Accused
Felons Felons All First
Time
Accused
Felons Felons All First
Time
Accused
Felons Felons
Percentage in
Con. x 2004
-0.024* -0.021* -0.012 -0.013 0.008 0.006 0.002 0.002 0.008 0.005 0.005 0.004 -0.033*-0.048*** 0.004 0.001
[0.013] [0.012] [0.010] [0.010] [0.009] [0.008] [0.004] [0.004] [0.015] [0.014] [0.008] [0.008] [0.017] [0.017] [0.015] [0.012]
Percentage in
Con. x 2005
0.016 0.011 -0.004 -0.003 -0.001 -0.008 0.007 0.004 -0.014 -0.016 0.001 -0.004 -0.033*-0.035*0.003 0.004
[0.016] [0.014] [0.012] [0.011] [0.010] [0.007] [0.008] [0.006] [0.012] [0.011] [0.007] [0.007] [0.019] [0.018] [0.013] [0.014]
Percentage in
Con. x 2006
0.005 0.001 0.00003 -0.003 -0.012 -0.006 -0.006 -0.003 -0.005 -0.008 0.004 0.005 -0.027 -0.037** 0.004 0.004
[0.017] [0.015] [0.012] [0.012] [0.007] [0.006] [0.004] [0.004] [0.012] [0.011] [0.007] [0.007] [0.020] [0.018] [0.014] [0.013]
Percentage in
Con. x 2007
0.032** 0.012 0.019 0.030** 0.011 -0.002 0.012 0.021** -0.001 0.0003 0.001 -0.010 -0.00002 -0.014 0.017 0.031**
[0.016] [0.014] [0.014] [0.013] [0.011] [0.008] [0.008] [0.009] [0.014] [0.012] [0.008] [0.007] [0.019] [0.018] [0.016] [0.016]
Percentage in
Con. x 2008
0.015 0.011 0.001 -0.003 0.015 0.010 0.005 0.006 -0.018 -0.024** 0.002 -0.008 0.001 -0.010 0.033** 0.027*
[0.017] [0.015] [0.014] [0.014] [0.011] [0.009] [0.006] [0.005] [0.011] [0.010] [0.006] [0.005] [0.019] [0.018] [0.016] [0.015]
Percentage in
Con. x 2009
0.012 -0.023* 0.032** 0.021 -0.012 -0.005 -0.007 -0.001 0.003 -0.005 0.010 0.008 0.043** 0.029 0.058*** 0.059***
[0.017] [0.013] [0.015] [0.015] [0.008] [0.007] [0.005] [0.004] [0.013] [0.012] [0.008] [0.008] [0.018] [0.018] [0.018] [0.017]
Percentage in
Con. x 2010
0.021 -0.012 0.031** 0.019 -0.019** -0.018*** -0.004 -0.002 -0.009 -0.009 -0.00001 -0.007 -0.004 -0.015 0.010 0.022
[0.018] [0.015] [0.015] [0.014] [0.007] [0.005] [0.006] [0.005] [0.012] [0.011] [0.006] [0.005] [0.018] [0.018] [0.017] [0.016]
Employment
Interactions Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Year Effects Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Block Group
Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y Y
Observations 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Employment interactions include tourism employment share
interacted with Hardberger and BRAC dummies as well as health care employment share interacted with Hardberger and BRAC dummies. Standard errors in
brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05, ***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal
history.
48
Table 11: Fixed Effects Estimates of Crime and Construction Workers in Bexar
County , Including QCEW Construction Employment Interactions
All First Time
Accused
Felons Felons
Burglary
Percentage in Con. x Hardberger -0.009 -0.016 0.005 0.016
[0.014] [0.014] [0.011] [0.011]
Percentage in Con. x BRAC 0.014 -0.013 0.043*** 0.048***
[0.016] [0.014] [0.012] [0.012]
Car Theft
Percentage in Con. x Hardberger -0.013 -0.011 -0.001 0.0004
[0.011] [0.008] [0.008] [0.008]
Percentage in Con. x BRAC 0.005 -0.004 0.015* 0.011
[0.011] [0.009] [0.008] [0.008]
Larceny
Percentage in Con. x Hardberger 0.001 -0.026** 0.024* 0.022*
[0.014] [0.013] [0.013] [0.013]
Percentage in Con. x BRAC 0.022 -0.013 0.048*** 0.051***
[0.015] [0.014] [0.014] [0.014]
Robbery
Percentage in Con. x Hardberger 0.014 0.0004 0.008 0.005
[0.013] [0.012] [0.009] [0.009]
Percentage in Con. x BRAC 0.022* -0.016 0.037*** 0.028***
[0.013] [0.012] [0.011] [0.010]
Murder
Percentage in Con. x Hardberger -0.014** -0.013** -0.001 -0.002
[0.007] [0.006] [0.005] [0.004]
Percentage in Con. x BRAC -0.012* -0.013** -0.001 0.002
[0.006] [0.005] [0.004] [0.004]
Rape
Percentage in Con. x Hardberger -0.009 -0.012 0.003 0.00004
[0.010] [0.009] [0.005] [0.005]
Percentage in Con. x BRAC -0.004 -0.008 0.004 -0.004
[0.010] [0.010] [0.005] [0.005]
Assault
Percentage in Con. x Hardberger -0.026 -0.031** 0.003 0.006
[0.017] [0.016] [0.012] [0.011]
Percentage in Con. x BRAC 0.012 -0.001 0.029** 0.038***
[0.015] [0.015] [0.013] [0.013]
Employment Interactions Y Y Y Y
Year Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y
QCEW Construction Employment
Interaction Y Y Y Y
Observations 11099 11099 11099 11099
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population).
Employment interactions include tourism employment share interacted with Hardberger and BRAC
dummies as well as health care employment share interacted with Hardberger and BRAC dummies.
Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05,
***p<0.01. See Appendix Tables A2-A9 for mean crime rates by criminal history.
Appendix
49
Table A1: Socio-Economic Outcomes and Construction Workers in Bexar County, 2000 to 2005-2009
Poverty
Rate
(%)
Log
Median
HH
Income
Emp. to
Pop. Ratio
(%)
Log
Median
House
Value
HHs with
2+
Vehicles
(%)
Log
Population
Log
Housing
Units
Housing
Units
Occupied
(%)
HHs
Moving in
> 5 Years
Ago (%)
Percentage in Construction -0.314*** 0.003* 0.396*** 0.004** 0.235*** -0.007*** -0.005** -0.059 -0.047
x BRAC/Hardberger [0.082] [0.002] [0.061] [0.002] [0.090] [0.002] [0.002] [0.057] [0.087]
Percentage in Construction 0.122** -0.001 -0.121*** -0.001 -0.042 0.004** 0.004*** -0.015 0.054
[0.061] [0.001] [0.036] [0.002] [0.063] [0.002] [0.001] [0.038] [0.065]
Percentage in Tourism 0.002 0.002 -0.011 0.007*** 0.150* -0.008** -0.007*** 0.047 -0.102
x BRAC/Hardberger [0.082] [0.002] [0.067] [0.002] [0.091] [0.003] [0.003] [0.060] [0.109]
Percentage in Tourism 0.057 0.000 -0.020 -0.003 -0.098* 0.002 0.004** 0.060 0.179***
[0.056] [0.001] [0.033] [0.002] [0.058] [0.002] [0.002] [0.044] [0.068]
Percentage in Health Care -0.177* 0.003* 0.224*** 0.002 0.140 -0.004 -0.002 -0.035 -0.184
x BRAC/Hardberger [0.098] [0.002] [0.066] [0.002] [0.092] [0.004] [0.004] [0.066] [0.125]
Percentage in Health Care -0.018 -0.0001 -0.012 0.003 0.018 -0.001 -0.00002 0.061 0.130*
[0.058] [0.002] [0.037] [0.002] [0.061] [0.002] [0.002] [0.039] [0.067]
Demographic & Housing Controls Y Y Y Y Y Y Y Y Y
Tract Fixed Effects Y Y Y Y Y Y Y Y Y
Year Fixed Effects Y Y Y Y Y Y Y Y Y
Observations 1982 1982 1982 1964 1982 1982 1982 1982 1982
R-Squared 0.669 0.861 0.645 0.917 0.729 0.878 0.914 0.455 0.718
Notes: Demographic and housing controls measured in 2000 at the block group level include log population, share black, share Hispanic, share male, share under
age 30, share age 65+, share of households that speak Spanish, share foreign born, share who lived in the same house 1 year ago, share with only a HS degree,
share with some college, share with a college degree, unemployment rate, labor force participation rate, log household income, log number of housing units,
share of units vacant, share of units owner occupied, median house age, and log house value. The relevant dependent variable is excluded from the set of
controls. Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05, ***p<0.01.
Appendix
50
Table A2: Fixed Effects Estimates of Burglary and Construction Workers in Bexar County, 2000-2010
Burglary
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
-0.002 -0.008 -0.014 -0.007 -0.013 -0.015 0.008 0.004 -0.003 0.019+ 0.015 0.007
[0.013] [0.013] [0.013] [0.012] [0.012] [0.013] [0.010] [0.010] [0.010] [0.010] [0.010] [0.011]
Percentage in
Construction x BRAC
0.021+ 0.015 0.015 -0.003 -0.008 -0.004 0.045** 0.043** 0.037** 0.049** 0.047** 0.041**
[0.012] [0.013] [0.013] [0.010] [0.011] [0.011] [0.010] [0.010] [0.010] [0.009] [0.009] [0.009]
Percentage in Construction -0.009 0.004 -0.018* -0.026**
[0.011] [0.009] [0.008] [0.007]
Percentage in Tourism x
Hardberger
-0.013 -0.015 -0.021 -0.022+ -0.023+ -0.025** 0.018+ 0.019+ 0.009 0.014 0.014 0.004
[0.014] [0.014] [0.013] [0.012] [0.012] [0.012] [0.010] [0.010] [0.010] [0.011] [0.010] [0.010]
Percentage in Tourism x
BRAC
0.007 0.002 0.002 -0.021+ -0.020+ -0.014 0.028** 0.022* 0.016 0.032** 0.025* 0.019*
[0.013] [0.012] [0.013] [0.011] [0.011] [0.011] [0.010] [0.010] [0.010] [0.009] [0.010] [0.010]
Percentage in Tourism -0.006 0.016+ -0.024** -0.023**
[0.010] [0.009] [0.006] [0.006]
Percentage in Health Care x
Hardberger
-0.009 -0.009 -0.017 -0.001 -0.003 -0.009 -0.005 -0.003 -0.009 -0.011 -0.01 -0.016
[0.014] [0.015] [0.014] [0.012] [0.013] [0.013] [0.011] [0.011] [0.011] [0.010] [0.011] [0.011]
Percentage in Health
Care x BRAC
-0.012 -0.006 -0.01 -0.018 -0.016 -0.019+ 0.004 0.01 0.007 0.003 0.008 0.005
[0.013] [0.014] [0.013] [0.011] [0.011] [0.011] [0.010] [0.011] [0.010] [0.011] [0.011] [0.011]
Percentage in Health Care 0.00 0.004 -0.004 0.001
[0.011] [0.009] [0.006] [0.007]
Mean Rate, 2000-2010 0.621 0.358 0.263 0.240
Year Fixed Effects Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099 11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.166 0.259 0.277 0.115 0.200 0.215 0.136 0.221 0.237 0.131 0.220 0.237
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table A1.
Standard errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
Appendix
51
Table A3: Fixed Effects Estimates of Car Theft and Construction Workers in Bexar County, 2000-2010
Car Theft
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
-0.007 -0.005 -0.01 -0.011 -0.011 -0.014* 0.005 0.007 0.004 0.007 0.009 0.007
[0.009] [0.009] [0.009] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007]
Percentage in
Construction x BRAC
0.021* 0.018* 0.015+ -0.002 -0.003 -0.003 0.030** 0.028** 0.024** 0.028** 0.026** 0.023**
[0.009] [0.009] [0.009] [0.006] [0.006] [0.006] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007]
Percentage in Construction -0.008 0.005 -0.017** -0.016**
[0.007] [0.005] [0.004] [0.004]
Percentage in Tourism x
Hardberger
0.022* 0.018+ 0.014 0.006 0.003 0 0.018* 0.020* 0.017* 0.009 0.011 0.009
[0.010] [0.011] [0.011] [0.008] [0.008] [0.009] [0.008] [0.008] [0.008] [0.008] [0.008] [0.008]
Percentage in Tourism x
BRAC
0.002 -0.003 -0.006 0.002 -0.003 -0.003 0.001 0.002 -0.002 0.003 0.004 0.001
[0.009] [0.010] [0.010] [0.007] [0.008] [0.008] [0.006] [0.006] [0.006] [0.006] [0.006] [0.006]
Percentage in Tourism 0.003 0.005 -0.003 -0.001
[0.006] [0.005] [0.004] [0.004]
Percentage in Health Care x
Hardberger
0.014 0.014 0.01 0.013 0.014 0.011 0.006 0.006 0.003 0.002 0.001 -0.001
[0.011] [0.011] [0.011] [0.009] [0.009] [0.009] [0.008] [0.008] [0.008] [0.008] [0.008] [0.008]
Percentage in Health
Care x BRAC
0.022* 0.022* 0.019* 0.015* 0.015* 0.014+ 0.016* 0.016* 0.014* 0.015* 0.015* 0.013*
[0.009] [0.009] [0.009] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007] [0.006] [0.006] [0.006]
Percentage in Health Care -0.007 -0.009+ 0.00 0.00
[0.006] [0.004] [0.004] [0.004]
Mean Rate, 2000-2010 0.152 0.076 0.075 0.070
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099 11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.092 0.179 0.191 0.056 0.133 0.142 0.083 0.174 0.187 0.072 0.168 0.18
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table A1. Standard
errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
Appendix
52
Table A4: Fixed Effects Estimates of Car Theft + Unauthorized Use of a Motor Vehicle and Construction Workers in Bexar County,
2000-2010
Car Theft + Unauthorized Use of a Motor Vehicle
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
-0.028** -0.027* -0.026* -0.031** -0.031** -0.030** 0.001 0.002 0.001 -0.001 0.002 0.002
[0.011] [0.011] [0.011] [0.009] [0.009] [0.009] [0.008] [0.008] [0.009] [0.008] [0.008] [0.009]
Percentage in
Construction x BRAC
0.002 -0.002
-0.001 -0.017* -0.020* -0.016+ 0.022** 0.020** 0.017* 0.019* 0.017* 0.016*
[0.009] [0.009] [0.009] [0.008] [0.008] [0.008] [0.007] [0.007] [0.007] [0.007] [0.007] [0.007]
Percentage in Construction -0.001 0.009 -0.011+ -0.008
[0.009] [0.008] [0.006] [0.006]
Percentage in Tourism x
Hardberger
0.015 0.009 0.011 -0.001 -0.005 -0.003 0.015 0.016+ 0.015 0.001 0.004 0.005
[0.011] [0.011] [0.012] [0.009] [0.009] [0.010] [0.009] [0.009] [0.009] [0.009] [0.009] [0.009]
Percentage in Tourism x
BRAC
0.001 -0.004 -0.001 -0.001 -0.005 -0.001 0.00 0.001 0.00 0.00 0.001 0.001
[0.011] [0.010] [0.011] [0.008] [0.008] [0.008] [0.008] [0.008] [0.008] [0.008] [0.007] [0.008]
Percentage in Tourism 0.003 0.004 0.00 0.003
[0.008] [0.006] [0.005] [0.005]
Percentage in Health Care x
Hardberger
-0.001 -0.003 -0.006 0.009 0.009 0.008 -0.002 -0.005 -0.007 -0.005 -0.006 -0.008
[0.014] [0.014] [0.013] [0.010] [0.010] [0.010] [0.010] [0.010] [0.010] [0.010] [0.010] [0.010]
Percentage in Health
Care x BRAC
0.01 0.009 0.007 0.007 0.005 0.005 0.007 0.007 0.005 0.009 0.009 0.008
[0.011] [0.011] [0.011] [0.008] [0.008] [0.008] [0.009] [0.009] [0.009] [0.009] [0.009] [0.009]
Percentage in Health Care -0.002 -0.009 0.008 0.006
[0.008] [0.007]
[0.006] [0.006]
Mean Rate, 2000-2010 0.276 0.149 0.126 0.119
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099
11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.086 0.208
0.216 0.085 0.170 0.179 0.086 0.175 0.184 0.080 0.175 0.184
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table A1.
Standard errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
Appendix
53
Table A5: Fixed Effects Estimates of Larceny and Construction Workers in Bexar County, 2000-2010
Larceny
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
0.006 0.005 0.007 -0.017 -0.021+ -0.017 0.021* 0.022+ 0.02 0.019+ 0.020+ 0.019
[0.013] [0.014] [0.014] [0.012] [0.012] [0.013] [0.011] [0.012] [0.012] [0.011] [0.012] [0.012]
Percentage in
Construction x BRAC
0.031* 0.029* 0.032** -0.002 -0.004 -0.001 0.046** 0.045** 0.045** 0.048** 0.047** 0.046**
[0.012] [0.012] [0.012] [0.012] [0.012] [0.012] [0.011] [0.010] [0.010] [0.010] [0.010] [0.010]
Percentage in Construction -0.019+ -0.012 -0.020+ -0.017
[0.011] [0.009] [0.010] [0.011]
Percentage in Tourism x
Hardberger
-0.015 -0.017 -0.017 -0.034** -0.032* -0.030* 0.011 0.014 0.011 0.01 0.013 0.011
[0.014] [0.013] [0.014] [0.013] [0.013] [0.013] [0.013] [0.013] [0.013] [0.013] [0.013] [0.012]
Percentage in Tourism x
BRAC
0.002 -0.002 0.00 -0.025* -0.026* -0.023+ 0.026* 0.025* 0.024* 0.028* 0.026* 0.023+
[0.013] [0.013] [0.013] [0.012] [0.012] [0.012] [0.013] [0.012] [0.012] [0.013] [0.012] [0.012]
Percentage in Tourism -0.001 0.009 -0.013 -0.01
[0.011] [0.009] [0.009] [0.009]
Percentage in Health Care x
Hardberger
-0.017 -0.013 -0.02 -0.02 -0.019 -0.022 -0.002 0.001 -0.005 0.007 0.009 0.003
[0.016] [0.016] [0.017] [0.015] [0.015] [0.016] [0.012] [0.012] [0.012] [0.012] [0.012] [0.012]
Percentage in Health
Care x BRAC
0.013 0.016 0.012 0.004 0.005 0.003 0.007 0.009 0.006 0.005 0.008 0.005
[0.014] [0.014] [0.014] [0.012] [0.012] [0.012] [0.012] [0.012] [0.013] [0.012] [0.013] [0.013]
Percentage in Health Care -0.015 -0.005 -0.008 -0.007
[0.012] [0.009]
[0.010] [0.010]
Mean Rate, 2000-2010 0.780 0.406 0.373 0.351
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099
11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.189 0.304
0.323 0.109 0.195 0.212 0.181 0.309 0.325 0.180 0.315 0.330
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table A1.
Standard errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
Appendix
54
Table A6: Fixed Effects Estimates of Robbery and Construction Workers in Bexar County, 2000-2010
Robbery
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
0.011 0.015 0.007 0.006 0.010 0.005 0.002 0.0005 -0.006 0.0004 -0.0004 -0.007
[0.012] [0.012] [0.012] [0.011] [0.011] [0.011] [0.009] [0.009] [0.009] [0.008] [0.008] [0.008]
Percentage in
Construction x BRAC
0.025** 0.025* 0.024* 0.00003 0.001 0.002 0.025** 0.023** 0.021** 0.021** 0.020** 0.019*
[0.010] [0.010] [0.010] [0.009] [0.009] [0.009] [0.007] [0.008] [0.008] [0.007] [0.007] [0.008]
Percentage in Construction -0.020* -0.005 -0.007 -0.005
[0.009] [0.007] [0.007] [0.007]
Percentage in Tourism x
Hardberger
0.008 0.011 0.003 -0.001 -0.002 -0.008 0.009 0.015 0.009 0.007 0.015 0.01
[0.013] [0.013] [0.012] [0.011] [0.011] [0.011] [0.010] [0.010] [0.010] [0.010] [0.010] [0.010]
Percentage in Tourism x
BRAC
0.003 -0.001 -0.001 -0.001 -0.003 -0.002 0.01 0.009 0.007 0.006 0.007 0.006
[0.012] [0.011] [0.012] [0.010] [0.010] [0.010] [0.009] [0.009] [0.009] [0.008] [0.008] [0.008]
Percentage in Tourism -0.0005 0.001 0.001 -0.002
[0.008] [0.007] [0.006] [0.005]
Percentage in Health Care x
Hardberger
0.010 0.012 0.006 -0.004 -0.0001 -0.005 0.009 0.009 0.005 0.006 0.006 0.002
[0.015] [0.015] [0.014] [0.012] [0.012] [0.012] [0.012] [0.012] [0.012] [0.011] [0.012] [0.011]
Percentage in Health
Care x BRAC
0.013 0.014 0.011 -0.001 -0.00003 -0.002 0.015 0.015 0.014 0.013 0.014 0.013
[0.012] [0.012] [0.013] [0.010] [0.010] [0.011] [0.010] [0.010] [0.010] [0.010] [0.010] [0.010]
Percentage in Health Care 0.008 0.007 0.008 0.009
[0.009] [0.007]
[0.007] [0.006]
Mean Rate, 2000-2010 0.400 0.207 0.193 0.169
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099
11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.125 0.203
0.220 0.094 0.170 0.184 0.097 0.172 0.186 0.091 0.166 0.180
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table
A1. Standard errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
Appendix
55
Table A7: Fixed Effects Estimates of Murder and Construction Workers in Bexar County, 2000-2010
Murder
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
-0.009 -0.008 -0.01 -0.008 -0.008 -0.009+ -0.001 0.000 -0.001 -0.0002 -0.000 -0.001
[0.006] [0.006] [0.006] [0.005] [0.005] [0.005] [0.005] [0.005] [0.005] [0.004] [0.004] [0.004]
Percentage in
Construction x BRAC
-0.003 -0.003 -0.003 -0.006 -0.005 -0.003 0.001 0.001 0.00 0.005 0.005 0.005
[0.006] [0.006] [0.006] [0.004] [0.005] [0.005] [0.004] [0.004] [0.004] [0.004] [0.004] [0.004]
Percentage in Construction -0.003 -0.004 0.00002 -0.003
[0.004] [0.004] [0.003] [0.002]
Percentage in Tourism x
Hardberger
0.003 0.002 0.00 0.005 0.003 0.002 0.005 0.004 0.002 0.010+ 0.007 0.006
[0.008] [0.008] [0.008] [0.007] [0.007] [0.007] [0.006] [0.006] [0.005] [0.005] [0.005] [0.005]
Percentage in Tourism x
BRAC
-0.003 -0.003 -0.003 -0.003 -0.004 -0.002 -0.0001 -0.0003 -0.001 0.001 -0.001 -0.001
[0.006] [0.006] [0.006] [0.005] [0.005] [0.005] [0.004] [0.004] [0.004] [0.003] [0.003] [0.003]
Percentage in Tourism -0.0002 -0.003 0.001 -0.002
[0.004] [0.003] [0.002] [0.002]
Percentage in Health Care x
Hardberger
0.004 0.004 0.002 -0.001 -0.0003 -0.002 0.0004 0.0002 -0.001 0.003 0.003 0.002
[0.008] [0.009] [0.009] [0.008] [0.008] [0.008] [0.006] [0.006] [0.006] [0.006] [0.006] [0.006]
Percentage in Health
Care x BRAC
-0.007 -0.007 -0.008 -0.005 -0.005 -0.005 -0.007+ -0.007 -0.008+ -0.002 -0.002 -0.003
[0.006] [0.007] [0.007] [0.005] [0.005] [0.005] [0.004] [0.004] [0.004] [0.004] [0.004] [0.004]
Percentage in Health Care 0.001 0.001 0.002 0.002
[0.005] [0.004]
[0.003] [0.003]
Mean Rate, 2000-2010 0.079 0.046 0.033 0.027
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099
11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.045 0.118
0.129 0.011 0.108 0.117 0.038 0.105 0.116 0.039 0.103 0.113
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table A1.
Standard errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
Appendix
56
Table A8: Fixed Effects Estimates of Rape and Construction Workers in Bexar County, 2000-2010
Rape
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
-0.011 -0.011 -0.011 -0.014+ -0.013 -0.013 0.003 0.002 0.002 0.0001 -0.0005 0.001
[0.010] [0.010] [0.010] [0.009] [0.008] [0.009] [0.005] [0.005] [0.005] [0.005] [0.005] [0.005]
Percentage in
Construction x BRAC
-0.008 -0.008 -0.008 -0.011 -0.011 -0.01 0.002 0.002 0.001 -0.006 -0.005 -0.005
[0.008] [0.007] [0.008] [0.007] [0.007] [0.007] [0.004] [0.004] [0.004] [0.004] [0.004] [0.004]
Percentage in Construction 0.011+ 0.012* 0.001 0.006
[0.007] [0.006] [0.004] [0.004]
Percentage in Tourism x
Hardberger
0.003 -0.0001 0 -0.005 -0.005 -0.005 0.009 0.007 0.007 0.008 0.006 0.007
[0.011] [0.011] [0.011] [0.010] [0.010] [0.010] [0.006] [0.006] [0.006] [0.006] [0.006] [0.006]
Percentage in Tourism x
BRAC
-0.009 -0.004 -0.005 -0.014+ -0.008 -0.008 0.002 0.002 0.001 -0.003 -0.004 -0.003
[0.009] [0.009] [0.009] [0.008] [0.009] [0.009] [0.004] [0.004] [0.004] [0.003] [0.003] [0.004]
Percentage in Tourism 0.010 0.014* -0.003 -0.001
[0.007] [0.006] [0.003] [0.003]
Percentage in Health Care x
Hardberger
-0.0002 -0.0003 -0.002 -0.003 -0.001 -0.004 -0.001 -0.002 -0.002 -0.011 -0.01 -0.01
[0.011] [0.011] [0.011] [0.010] [0.010] [0.009] [0.006] [0.006] [0.006] [0.007] [0.007] [0.007]
Percentage in Health
Care x BRAC
-0.009 -0.008 -0.010 -0.01 -0.009 -0.010 -0.004 -0.004 -0.004 -0.014** -0.014** -0.014**
[0.009] [0.009] [0.009] [0.008] [0.008] [0.008] [0.004] [0.004] [0.004] [0.005] [0.005] [0.005]
Percentage in Health Care 0.009 0.008 0.004 0.010*
[0.006] [0.006]
[0.004] [0.004]
Mean Rate, 2000-2010 0.162 0.124 0.038 0.034
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects
Y
Y Y Y
Observations 10901 11099
11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.080 0.156
0.164 0.071 0.145 0.152 0.051 0.125 0.135 0.047 0.124 0.132
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table A1.
Standard errors in brackets allow for arbitrary correlation within block groups. *p<0.10, **p<0.05, ***p<0.01.
Appendix
57
Table A9: Fixed Effects Estimates of Assault and Construction Workers in Bexar County, 2000-2010
Assault
All First Time Accused Felons Felons
Percentage in Construction
x Hardberger
-0.017 -0.024 0.012 -0.024+ -0.026+ 0.009 0.007 0.003 0.003 0.008 0.004 0.002
[0.015] [0.016] [0.017] [0.014] [0.014] [0.016] [0.011] [0.011] [0.004] [0.011] [0.011] [0.003]
Percentage in
Construction x BRAC
0.022* 0.017 0.022 0.007 0.007 0.019 0.037** 0.029* 0.003 0.042** 0.035** 0.004
[0.011] [0.012] [0.019] [0.011] [0.011] [0.017] [0.010] [0.011] [0.007] [0.010] [0.011] [0.005]
Percentage in Construction -0.009 0.001 -0.018* -0.015*
[0.010] [0.009] [0.007] [0.007]
Percentage in Tourism x
Hardberger
-0.019 -0.019 0.012 -0.007 -0.005 0.02 -0.015 -0.016 -0.008* -0.007 -0.007 -0.004
[0.014] [0.015] [0.020] [0.014] [0.014] [0.020] [0.011] [0.011] [0.004] [0.011] [0.010] [0.004]
Percentage in Tourism x
BRAC
0.009 0.007 0.055 0.0003 -0.0001 0.023 0.039** 0.037** 0.033 0.042** 0.039** 0.030+
[0.012] [0.012] [0.037] [0.011] [0.011] [0.021] [0.012] [0.012] [0.024] [0.011] [0.011] [0.018]
Percentage in Tourism 0.004 0.004 -0.006 -0.008
[0.010] [0.009] [0.008] [0.007]
Percentage in Health Care x
Hardberger
0.016 0.017 0.027 0.012 0.013 0.027 0.002 0.002 -0.001 0.0001 -0.0004 0.001
[0.015] [0.016] [0.024] [0.014] [0.014] [0.023] [0.014] [0.014] [0.004] [0.012] [0.012] [0.003]
Percentage in Health
Care x BRAC
0.011 0.015 0.026 0.002 0.006 0.025 0.024* 0.024* 0.001 0.025* 0.025* 0.001
[0.013] [0.013] [0.025] [0.013] [0.013] [0.024] [0.012] [0.012] [0.006] [0.011] [0.011] [0.005]
Percentage in Health Care -0.012 -0.007
-0.008 -0.008
[0.010] [0.009]
[0.007] [0.007]
Mean Rate, 2000-2010 0.876 0.569 0.307 0.252
Year Fixed Effects Y Y Y Y Y Y Y Y Y Y Y Y
Demo. & Housing Controls Y Y Y Y
Tract Fixed Effects Y Y Y Y
Block Group Fixed Effects Y Y Y Y Y Y Y
Y
2000-2006 Crime Trend x
Year Fixed Effects Y Y Y Y
Observations 10901 11099
11099 10901 11099 11099 10901 11099 11099 10901 11099 11099
R-Squared 0.222 0.311
0.189 0.164 0.248 0.178 0.190 0.266 0.216 0.161 0.240 0.196
Notes: Dependent variables are ln(people charged with felonies committed in year/1000 population). Demographic and housing controls are listed in the notes to Table
A1. Standard errors in brackets allow for arbitrary correlation within block groups. +p<0.10, *p<0.05, **p<0.01.
