METROPOLITAN OPPORTUNITY SERIES

City and Suburban

Crime Trends in
Metropolitan America
Elizabeth Kneebone and Steven Raphael

Analysis of FBI Uniform Crime Report and U.S. Census Bureau data from 1990 to 2008 in the cit-
“The narrowing ies, towns, and counties of the 100 largest U.S. metropolitan areas finds that:

of the gap n Both violent and property crime declined significantly between 1990 and 2008 in the
100 largest metro areas, with the largest decreases occurring in cities. Violent crime rates
between city and dropped by almost 30 percent in cities, while property crime fell by 46 percent. Though city
crime rates remain considerably above those in suburbs, smaller decreases in suburban violent
suburban crime and property crime rates over this time period (7 and 37 percent, respectively) narrowed the
gap.
rates underscores
n The gap between city and suburban violent crime rates declined in nearly two-thirds of
that crime is not metro areas. In 90 of the 100 largest metro areas, the gap between city and suburban prop-
erty crime rates narrowed from 1990 to 2008. In most metro areas, city and suburban crime
solely an urban rates rose or fell together.

challenge, but n Among suburban communities, older high-density suburbs registered the largest declines
in crime rates. All types of suburban communities saw property crime rates fall over this time
a metropolitan- period. Cities and high-density suburbs also saw violent crime rates decline, but low-density
exurban communities experienced slight increases that are not explained by their changing
wide issue.” demographics.

n As crime rates fell and communities diversified, relationships between crime and com-
munity demographic characteristics weakened significantly. The association between crime
and community characteristics—like the proportion of the population that is black, Hispanic,
poor, or foreign-born—diminished considerably over time. For example, the strength of the rela-
tionship between share of black residents and property crime decreased by half between 1990
and 2008, while the association between the share of Hispanic residents and violent crime all
but disappeared.

In general, the nation’s largest metropolitan areas are much safer today than they were in years
past. Within metropolitan areas, older, more urbanized, poorer, and more minority communities
have benefited the most from these trends, narrowing the disparities between cities and suburbs
and underscoring that crime is not a uniquely urban issue, but a metropolitan one. As such, juris-
dictions that lagged in reducing crime rates since 1990 may benefit from looking to neighboring
communities and similar regions for lessons learned and successful policies that helped signifi-
cantly reduce property and violent crime over the last two decades.

BROOKINGS | May 2011 1

 Introduction

T
he impact of crime on general well-being is profound. Those most directly impacted are the
victims of crime. By one estimate, the combination of direct monetary losses and the costs
of pain and suffering among crime victims in the U.S. amounts to nearly 6 percent of gross
domestic product.1 Beyond these direct costs are substantial indirect costs associated with
reducing the threat of crime. In 2006, federal, state, and local government criminal justice expendi-
tures amounted to $214 billion.2 Many households pay significant premiums, either in terms of housing
prices or commute costs, to live in neighborhoods with lower probabilities of victimization. Many also
purchase security devices and insurance to minimize the likelihood and costs of being criminally vic-
timized. Moreover, fear of crime often impacts the most mundane personal decisions, such as whether
to walk down a given street or through a particular neighborhood, whether to let one’s children play
outside, or whether to leave one’s home after dark.3
While all communities are affected by crime and the criminal justice system, residents in large urban
areas are particularly impacted. Moreover, within large metropolitan areas, the residents of poor,
largely minority neighborhoods suffer disproportionately. Crime rates are generally higher in more
urbanized areas and the young, male, and minority residents of the nation’s central cities contribute
disproportionately to the growing prison population.4
Yet, in recent decades, U.S. crime rates have fallen sharply. By 2008 the sexual assault rate stood
at only 23 percent of its peak value in 1991, while robbery, aggravated assault, and simple assault had
fallen to 37, 33, and 42 percent of their 1991 levels, respectively. Similarly, homicide rates dropped from
10.5 per 100,000 in 1991 to 6.2 per 100,000 by 2006. Between 1991 and 2008 the number of burglaries
per 1,000 households declined by 59 percent, while rates of theft and motor vehicle theft dropped by
62 and 70 percent, respectively.5
Though much has been written about the precipitous declines in crime since the 1990s, less is
known about trends within the nation’s big cities and suburbs.6 Two-thirds of the nation’s population
lives in the 100 largest metropolitan areas, but crime levels vary greatly across—and even within—these
regions. To what extent have decreases in crime been shared across these communities? Moreover,
crime fell over a period that coincided with considerable changes in the makeup and distribution of the
country’s metropolitan population.7 Do those changes help explain the steep declines in community-
level crime?
In this paper, we explore these questions by analyzing crime data compiled by the Federal Bureau of
Investigation (FBI) and data from the U.S. Census Bureau to provide a geographically-focused assess-
ment of how crime rates have changed between 1990 and 2008. Specifically, we analyze data for
the roughly 5,400 communities located within the 100 largest U.S. metropolitan areas. We estimate
changes in metropolitan crime, as well as city and suburban trends within these regions. We then con-
sider the relationship between community-level demographic characteristics and crime, and analyze
how those relationships may have changed over time.

Methodology

T
his analysis uses data from the FBI and various U.S. Census Bureau surveys to assess crime
trends and their relationship to demographic characteristics across and within the nation’s
100 largest metropolitan areas.

Crime Data
To measure crime rates, we use data from the 1990, 2000, and 2008 Uniform Crime Report (UCR)
Offenses Known and Cleared by Arrest data sets compiled by the FBI. The UCR data provide counts of
crimes reported to the police for each police agency (referred to as a reporting unit in the UCR data)
by month.
We use the UCR data to estimate rates of serious felony crimes. Felony criminal incidents involving
victims are officially categorized into the following seven mutually exclusive categories:

2 BROOKINGS | May 2011

 ➤ Murder and non-negligent manslaughter: defined as the willful killing of one human being by
another.
➤ Rape/sexual assault: rape refers to forced sexual intercourse, inclusive of psychological coer-
cion and physical force. Sexual assault is distinct from rape and includes any unwanted sexual
contact between victim and offender.
➤ Robbery: a completed or attempted theft directly from a person by force of threat with or with-
out a weapon and with or without an injury.
➤ Assault: an attack with or without a weapon and with or without an injury. Attack with a
weapon or an attack without a weapon resulting in a serious injury is referred to as aggravated
assault. An attack without a weapon with no or minor injuries to the victim is referred to as
simple assault.
➤ Burglary: the unlawful or attempted or forcible entry of a residence, often but not necessarily
involving theft.
➤ Larceny/theft: the taking of property without personal contact.
➤ Motor vehicle theft: the stealing or unauthorized taking of a motor vehicle, including
attempted theft.8
In addition to providing select results for each of the individual crimes listed above, we aggregate
incident types to present findings for two general categories of crime. Conventional aggregations
generally group the first four felonies under the banner of violent crimes, since each felony involves
direct coercive or violent contact between the offender and victim. The latter three felony offenses
are commonly referred to as property crimes, since the objective of each is to unlawfully acquire the
property of another without physically encountering the victim.
To calculate crime rates, we aggregate 12 months of crime data to create annual estimates for
1990, 2000, and 2008. In all tabulations, crime rates are measured as criminal incidents per 100,000
residents.

Geographies
We restrict our analysis to reporting units located within the country’s 100 largest metropolitan
areas, as defined by the U.S. Office of Management and Budget in 2008 and based on official popula-
tion estimates in that year.
To determine the location of each reporting unit in the 100 largest metro areas, we first identify
the unit of geography with which the police agency is associated. Crime data from reporting units
generally can be matched to three types of geographies: places, Minor Civil Divisions, and unincorpo-
rated portions of counties.
For the most part, reporting units/police agencies correspond to places. “Places” include incorpo-
rated jurisdictions (such as cities, towns, and villages) as well as census-designated places (unin-
corporated areas delineated by the U.S. Census Bureau for statistical purposes). For example, the
Oakland Police Department is a single reporting unit. In instances where there are multiple police
agencies within a place, we aggregate crime data from all reporting units to create a place-level
total.9
Reporting units may also correspond to a Minor Civil Division (MCD). The Census Bureau uses
MCDs to designate the primary governmental and/or administrative divisions of a county, such as a
civil township, precinct, or magisterial district. MCDs exist in 28 states and the District of Columbia.
For the remaining states, the Census Bureau designates MCD equivalents, called Census County
Divisions (CCDs), for statistical purposes.10 Police agencies covering areas not located within a place
but located within an identifiable MCD/CCD are aggregated to the MCD level.
Finally, police agencies covering unincorporated areas of counties that lie outside of these two
geography types are combined into a balance-of-county aggregate.11
After matching reporting units to the relevant geography, we identified roughly 5,400 separate
geographic units within the 100 largest metropolitan areas that appear in the UCR data.
To analyze differences in crime trends within regions, we assign each of these geographic units a
community type. We first establish whether a community is a primary city or a suburb. Primary cities
are those that appear first in the official metropolitan statistical area (MSA) name, as well as any

BROOKINGS | May 2011 3

 other city in the MSA title with a population of at least 100,000. Suburbs make up the remainder of the
MSA outside the primary city or cities.
We also further classify suburban communities according to their degree of urbanization to explore
differences in crime trends across types of suburbs in more detail. Based on a community’s urbaniza-
tion rate (or the share of people in the community who live in an urbanized area, as designated by the
Census Bureau), we assign suburban geographies to one of four categories.12 A high-density suburb
has an urbanization rate of 99.5 percent or more; a mature suburb has an urbanization rate between
75 and 99.5 percent; an emerging suburb has an urbanization rate greater than 0 but less than 75
percent; and an exurb has no urbanized population whatsoever. High-density suburbs are often quite
similar to central cities; they are usually inner-ring suburbs with relatively large minority populations,
and often markedly differ from newer emerging suburbs and exurbs on metropolitan area fringes.
Mature suburbs lie somewhere in the middle.

Demographic Data
Finally, we match our community-level crime data to data from the decennial census and the American
Community Survey (ACS) to analyze the relationship between crime rates and community-level demo-
graphic characteristics. Specifically, we employ data from the 1990 Census of Population and Housing
Summary Tape File 3A, the 2000 Census of Population and Housing Summary File 3, and the 2005-
2009 ACS five-year estimates.13 We use these data to estimate the proportion of community residents
that are black, Hispanic, foreign-born, or poor in each year.
For identifiable census places and MCDs, we match corresponding estimates from the decennial cen-
sus or ACS directly to the UCR data. Roughly 75 percent of the population of the metropolitan areas
included in this study resides within a definable place or MCD. For the unincorporated balance-of-
county observations, we assign the county-level average values.14 The goal of this additional analysis
is to assess the degree to which the empirical relationships between community-level crime rates and
average characteristics have changed with the sharp declines in crime that have occurred since 1990.

Findings

A. Both violent and property crime declined significantly between 1990 and 2008 in
the 100 largest metro areas, with the largest decreases occurring in cities.
Documenting how crime rates have changed in the nation’s 100 largest metropolitan areas over the
last two decades Figure 1 presents estimates of the annual number of violent crimes and property
crimes per 100,000 residents for 1990, 2000, and 2008. In effect, these averages represent the crime
rate experienced by the typical resident of a given community type (metropolitan, city, suburban) in
each year.15
Taken together, the 100 largest metro areas saw both violent and property crimes rates decline
substantially between 1990 and 2008. Over that time period, violent crime fell by 21 percent, while
property crime declined by twice that rate (42 percent). However, these changes did not occur evenly
over time; more than 80 percent of the net decline in both types of crime occurred between 1990 and
2000 alone.
Nor did these changes occur evenly within metropolitan areas. Primary city residents experienced
more pronounced decreases in crime, both absolutely and in percentage terms, than their suburban
counterparts. Between 1990 and 2008, the primary city violent crime rate fell by 879 incidents per
100,000 (a 29 percent decrease) while property crime dropped by 3,848 incidents per 100,000 (a 46
percent decline). By contrast, the suburban violent crime rate dropped by 86 incidents per 100,000 (a
7 percent change) while the property crime fell by 1,508 incidents per 100,000 (a 37 percent decline).
While city crime rates remain higher than suburban levels, the pace of urban declines between 1990
and 2008 markedly narrowed the city/suburban gap. Specifically, between 1990 and 2008, the violent
crime rate in primary cities dropped from 2.8 times the comparable rate for the suburbs to double the
suburban rate, and the disparity in the average property crime rate dropped from twice the suburban
rate to 1.7.

4 BROOKINGS | May 2011

 Figure 1A. Violent Crimes per 100,000 Residents
in the 100 Largest U.S. Metropolitan Areas, by City and Suburb

3,500
■ 1990 ■ 2000 ■ 2008
Violent Crimes per 100,000 Residents

3,000

2,500

2,000

1,500

1,000

500

0
Metro Area Primary City Suburb

Source: Authors’ analysis of UCR data

Figure 1B. Property Crimes per 100,000 Residents

in the 100 Largest U.S. Metropolitan Areas, by City and Suburb

9,000
■ 1990 ■ 2000 ■ 2008
8,000
Property Crimes per 100,000 Residents

3500
7,000
2008
6,000
3000
2000
5,000
2500
4,000 1990
2000
3,000

2,000
1500

1,000
1000
0
Metro Area Primary City Suburb
500

0
Metro
Source: Authors’ analysis of Area
UCR data Primary City Suburb

BROOKINGS | May 2011 5

 Table 1. Violent and Property Crime Rates for the 100 Largest Metropolitan Areas, by City and Suburb

Metro Area City Suburb

Change, % Change Change, % Change Change, % Change
1990 to 1990 to 1990 to 1990 to 1990 to 1990 to
1990 2000 2008 2008 2008 1990 2000 2008 2008 2008 2000 2008 2008 2008
Violent 1,776 1,474 1,402 -374* -21% 3,008 2,330 2,129 -879* -29% 1,061 1,062 -86* -7%
Murder 11 6 6 -5* -46% 23 13 12 -12* -50% 3 3 -2* -32%
Rape 41 29 26 -16* -38% 69 44 36 -34* -49% 21 21 -6* -23%
Robbery 353 183 182 -171* -48% 773 391 362 -410* -53% 82 97 -41* -30%
Aggravated assault 477 340 281 -196* -41% 833 592 465 -368* -44% 219 195 -100* -33%
Simple assault 894 917 908 14 2% 1309 1291 1242 -67 -5% 736 746 64* 9%

Property 5,544 3,535 3,210 -2,334* -42% 8,326 5,357 4,477 -3,848* -46% 2,653 2,616 -1,508* -37%
Burglary 1,314 683 706 -608* -46% 1,984 1,024 1,005 -979* -49% 517 566 -407* -42%
Larceny 3,359 2,357 2,131 -1,227* -37% 4,772 3,447 2,853 -1,919* -40% 1,830 1,793 -844* -32%
MV theft 871 495 373 -498* -57% 1,570 887 619 -950* -61% 306 257 -257* -50%

Source: Author’s analysis of UCR data

*Difference statistically significant at the one percent level of confidence.
Note: All figures measure the number of criminal incidents per 100,000 residents. Averages are weighted by community-level population for each year.

Moreover, the larger absolute and percentage crime declines in primary cities occurred broadly
across offense categories (Table 1). With the exception of simple assault, the 100 largest metro areas
saw crime decline significantly across incident types. In each case primary city declines outpaced
those in suburbs. In other words, the aggregate trends documented in Figure 1 were not driven by
changes in specific offenses. Rather between 1990 and 2008 there was a broad-based decline in crimi-
nal offending that disproportionately impacted primary cities.

B. The gap between city and suburban violent crime rates declined in nearly two-thirds
of metro areas.
Though average declines in city crime rates outpaced suburban decreases, the extent to which indi-
vidual metro areas experienced these trends—and the related narrowing of the city/suburban crime
rate differential—varied considerably.
Among individual metro areas, about half (49) experienced declines in their metro-wide violent
crime rates. Within the largest metro areas, cities were more likely to see violent crime fall than
suburbs; more than half of the metro areas studied (56) experienced drops in city violent crime rates,
while only 39 saw suburban violent crime decline.16 Metro areas that were home to the greatest drops
in city and suburban crime between 1990 and 2008 were diverse in location and size, and several
place among the top 10 for greatest declines in both city and suburban violent crime rates, including
Chicago, Los Angeles, Miami, New York, and Portland, OR (Table 2).
In fact, city and suburban crime rates generally tended to follow the same trajectory over this time
period. In 59 metro areas, city and suburban crime trends moved together, whether trending upward
or down. Of these, 44 metro areas saw city rates drop faster (or increase less) than suburban levels. In
the remaining regions, only 11 experienced declines in suburban crime while city rates grew, while more
than a quarter of regions saw city crime rates fall as suburban violent crime rates increased.
Together, these trends meant that the gap between city and suburban violent crime rates nar-
rowed in 62 of the largest metro areas. The average change in the gap between city and suburban
crime rates was -474 crimes per 100,000, though metro areas that started with the largest disparities
between cities and suburbs in 1990 saw that gap narrow the most.17
Property crime declines were more broadly shared across individual metro areas between 1990 and
2008. Rates of property crime decreased in 95 metro areas, with 95 regions experiencing a decline in

6 BROOKINGS | May 2011

 Table 2. Cities and Suburbs with Largest Declines in Violent Crime Rates,
100 Largest U.S. Metropolitan Areas, 1990 to 2008

Primary City Violent Crime Suburban Violent Crime

Rates per 100,000 Residents Rates per 100,000 Residents
% %
1990 2008 Change 1990 2008 Change
Chicago-Naperville-Joliet, IL-IN-WI 5,750 1,135 -80.3 Chicago-Naperville-Joliet, IL-IN-WI 1,557 269 -82.7
New York-Northern New Jersey- 2,489 636 -74.4 Riverside-San Bernardino-Ontario, CA 1,758 970 -44.8
Long Island, NY-NJ-PA
Atlanta-Sandy Springs-Marietta, GA 6,975 2,920 -58.1 Poughkeepsie-Newburgh-Middletown, NY 1,161 668 -42.5
Hartford-West Hartford-East Hartford, CT 7,022 3,015 -57.1 Bakersfield, CA 1,853 1,128 -39.1
Portland-South Portland-Biddeford, ME 3,533 1,598 -54.8 Los Angeles-Long Beach-Santa Ana, CA 1,564 971 -37.9
Portland-Vancouver-Beaverton, OR-WA 3,113 1,439 -53.8 Sacramento—Arden-Arcade—Roseville, CA 1,343 855 -36.3
Miami-Fort Lauderdale-Pompano Beach, FL 4,789 2,246 -53.1 Portland-Vancouver-Beaverton, OR-WA 950 607 -36.1
Los Angeles-Long Beach-Santa Ana, CA 3,071 1,494 -51.4 Miami-Fort Lauderdale-Pompano Beach, FL 2,464 1,592 -35.4
Dallas-Fort Worth-Arlington, TX 4,535 2,460 -45.8 New York-Northern New Jersey-Long Island, 986 644 -34.7
NY-NJ-PA
Seattle-Tacoma-Bellevue, WA 3,280 1,812 -44.8 San Jose-Sunnyvale-Santa Clara, CA 1,183 785 -33.6

Source: Author’s analysis of UCR data

Table 3. Cities and Suburbs with Largest Declines in Property Crime Rates,
100 Largest U.S. Metropolitan Areas, 1990 to 2008

Primary City Violent Crime Suburban Violent Crime

Rates per 100,000 Residents Rates per 100,000 Residents
% %
1990 2008 Change 1990 2008 Change
Tucson, AZ 11,338 2,294 -79.8 Chicago-Naperville-Joliet, IL-IN-WI 4,189 1,398 -66.6
New York-Northern New Jersey- 7,503 1,870 -75.1 New Haven-Milford, CT 5,638 2,045 -63.7
Long Island, NY-NJ-PA
El Paso, TX 10,281 3,277 -68.1 Boise City-Nampa, ID 4,182 1,876 -55.1
Miami-Fort Lauderdale-Pompano Beach, FL 14,448 5,379 -62.8 Bridgeport-Stamford-Norwalk, CT 3,551 1,596 -55.1
Los Angeles-Long Beach-Santa Ana, CA 6,961 2,629 -62.2 New York-Northern New Jersey- 3,509 1,611 -54.1
Long Island, NY-NJ-PA
Boston-Cambridge-Quincy, MA-NH 9,466 3,660 -61.3 Indianapolis-Carmel, IN 2,792 1,348 -51.7
San Diego-Carlsbad-San Marcos, CA 8,177 3,225 -60.6 Colorado Springs, CO 3,496 1,688 -51.7
Palm Bay-Melbourne-Titusville, FL 6,022 2,382 -60.4 Los Angeles-Long Beach-Santa Ana, CA 4,869 2,452 -49.6
Bridgeport-Stamford-Norwalk, CT 8,200 3,261 -60.2 Lakeland-Winter Haven, FL 7,221 3,658 -49.3
Portland-South Portland-Biddeford, ME 10,995 4,419 -59.8 Denver-Aurora, CO 5,389 2,754 -48.9

Source: Author’s analysis of UCR data

city rates and 85 seeing suburban rates fall. The regions that posted the largest decreases in city and
suburban crime rates are once again diverse in their geographic location and population size, though
there is only limited overlap with the regions that saw the greatest drops in violent crime, like New
York and Los Angeles (Table 3).
Compared to violent crime, less overlap exists between metro areas experiencing the largest
declines in city and suburban property crime, but metro areas were even more likely to see city and

BROOKINGS | May 2011 7

 suburban trends for property crime move together over this time period. In 85 metro areas, changes
in property crime rates trended in the same direction between 1990 and 2008, and in all but one of
these regions (Scranton) property crime declined. In two-thirds of these metro areas, cities saw prop-
erty crime rates fall faster than suburban rates.
As a result of these changes, fully 90 metro areas saw the gap between city and suburban property
crime rates narrow between 1990 and 2008. And given the broad-based nature of property crime
declines, the average change in the disparity between city and suburban rates was more than four
times greater than for violent crime (-2,075).18

C. Among suburban communities, older high-density suburbs registered the largest

declines in crime rates.
In our tabulations so far, we have used “suburbs” to encompass all metropolitan communities that fall
outside primary cities; however, a great deal of variation exists among these areas. This large category
includes inner-ring suburbs proximate to primary cities (and that often have characteristics and social
problems comparable to those of the central city), high-growth emerging suburbs, and exurban areas on
the metropolitan area fringe with very low population density. In order to paint a more thorough portrait
of crime trends within metropolitan areas, in this section we stratify suburbs into four categories based
on their level of urbanization: high-density suburbs, mature suburbs, emerging suburbs, and exurbs.
To better understand the composition of these communities, we first analyze the distribution and
demographic characteristics of U.S. metropolitan area residents in each category. Over the eighteen
years studied, population movements continued the post-WWII trends towards greater population
decentralization within metropolitan areas, with population share declining in primary cities and high-
density suburbs while growing in mature suburbs and lower-density suburbs (Figure 2).19 Moreover,
as total population shifted outward in these metro areas, key demographic groups also decentralized.
In 1990, 64 percent of African Americans lived in primary cities, but by 2005-2009 that figure had
dropped to 51 percent. In addition, poor residents and the Hispanic population passed a notable tipping
point over this time period; despite being majority urban in 1990, by 2005-2009 more than half of
each population was suburban.20

Figure 2. Distribution of Population Across Community Types,

100 Largest U.S. Metropolitan Areas

0.40
■ 1990 ■ 2000 ■ 2005–2009
0.35
Proportion of Metro Area Population

0.30

0.25

0.20

0.15

0.10

0.05

0.00
Primary High-Density Mature Emerging Exurb
City Suburb Suburb Suburb

Source: Authors’ analysis of 1990 and 2000 decennial census data and 2005-2009 ACS five-year estimates

8 BROOKINGS | May 2011

 Table 4. Demographic Characteristics Within the 100 Largest Metropolitan Areas, by Community Type

Share Black Share Poor Share Hispanic Share Foreign-Born

1990 2008 Change 1990 2008 Change 1990 2008 Change 1990 2008 Change
Primary City 24.8 22.1 -2.7 18.1 18.3 0.2 16.4 24.9 8.4a 14.9 20.4 5.9a
High-Density Suburb 8.8 10.9 2.1 a
8.3 10.3 1.9a
10.2 18.4 8.1 a
11.9 18.4 6.5a
Mature Suburb 7.6 11.5 3.9 a
7.9 9.1 1.2a
7.4 13.6 6.2 a
8.1 13.3 5.1a
Emerging Suburb 6.7 8.3 1.6 a
9.5 10.0 0.5c
6.8 12.1 5.3 a
5.2 8.4 3.2a
Exurb 8.3 8.0 -0.3 14.1 13.4 -0.6b
6.0 11.4 5.4 a
3.0 5.6 2.6a

Source: Authors’ analysis of 1990 decennial census data and 2005-2009 ACS five-year estimates
Note: All averages are weighted by community population.
a. Change statistically significant at the one percent level of confidence.
b. Change statistically significant at the five percent level of confidence.
c. Change statistically significant at the 10 percent level of confidence.

With this decentralization, the demographic makeup of community types also shifted (Table 4).
In general, primary cities post the highest shares for the proportion of residents that is black, poor,
Hispanic, and foreign born, yet suburbs significantly diversified over this period. Between 1990 and
2005-2009, all community types showed an increase in their proportion of Hispanic and foreign-born
residents. Most suburbs also saw their proportion of African Americans grow, though exurbs, along
with primary cities, held statistically steady on this measure. In addition, high-density, mature, and
emerging suburbs saw significant increases in their share of poor residents over this period. Exurbs
were the only community type to experience a slight decrease in the proportion of poor residents,
though they continued to have the highest poverty rate among suburban types in 2005-2009.
Just as demographic distributions and trends varied within the 100 largest metro areas, crime
trends also played out differently across community types. Following the primary city trend, high-
density suburbs saw relatively large drops in violent crime rates between 1990 and 2008. At the
same time, emerging suburbs and exurban communities experienced slight increases (Figure 3A).
Despite these upticks, the population distributions shown in Figure 2 suggest that the overwhelming
majority of metropolitan area residents (82 percent) experienced declines in their community’s violent
crime rate.
Similar to the findings for violent crime rates, we generally observe lower property crime rates in
areas with lower urbanization. But by 2008 the crime-community type profile had flattened consider-
ably, driven primarily by relatively large property crime declines in central cities and high-density sub-
urbs (Figure 3B). In contrast to the violent crime results, all areas experience pronounced decreases in
property crime over the eighteen-year period, and there is little evidence of an increase in crime in the
emerging suburbs and exurbs between 2000 and 2008.
How do the demographic shifts outlined above relate to the differing crime trends seen across
community types? To explore the extent to which these shifts explain the crime patterns observed in
Figure 3A and 3B, we use a series of linear regression models (Appendix C). The dependent variable is
the change in community-level crime between 1990 and 2008, and the key explanatory variables are
indicator variables for community type and community-level changes in the demographic characteris-
tics presented in Table 4.21
Starting with violent crime, the baseline model confirms the statistical significance of the large
average crime declines in central cities and high-density suburbs as well as the slight increases in
emerging suburbs and exurbs noted above. We then add controls for the change in the proportion of
residents that is black, poor, Hispanic, and foreign born. In terms of specific characteristics, we find
a positive association between a change in violent crime rates and both the proportion of residents
that is African-American and that is poor. In contrast, crime declines significantly with increases in the
proportion that is foreign-born and Hispanic.

BROOKINGS | May 2011 9

 Figure 3A. Violent Crimes per 100,000 Residents by Community Type,
100 Largest Metropolitan Areas

3,500
■ 1990 ■ 2000 ■ 2008

3,000

2,500

2,000

1,500

1,000

500

0
Primary High-Density Mature Emerging Exurb
City Suburb Suburb Suburb

Figure 3B. Property Crimes per 100,000 Residents by City Type for Cities
in the 100 Largest Metropolitan Areas

9,000
■ 1990 ■ 2000 ■ 2008
8,000
3500
7,000

3000
6,000

5,000
2500
4,000
2000
3,000

1500
2,000

1,000
1000

0
500 Primary High-Density Mature Emerging Exurb
City Suburb Suburb Suburb

0
Primary City High-Density Suburb Mature Suburb Emerging Suburb Exurb

10 BROOKINGS | May 2011

 Figure 4. Estimated Change in Violent Crime by Community Type, with and without Accounting for Demographic Change

■ Baseline
■ Controlling for Demographics Emerging Suburb

Exurb

Mature Suburb

High-Density Suburb

Primary City

-800 -700 -600 -500 -400 -300 -200 -100 0 100 200 300 400

Source: Authors’ analysis of UCR data, 1990 decennial census, and 2005-2009 ACS five-year estimates

Figure 5. Estimated Change in Property Crime by Community Type, with and without Accounting for Demographic Change

■ Baseline
Emerging Suburb ■ Controlling for Demographics

Controlling for D
Emerging Suburb
Exurb
Baseline

Exurb
Mature Suburb

Mature Suburb
High-Density Suburb

High-Density Suburb
Primary City

-4,000 -3,500 -3,000 -2,500 -2,000 -1,500 -1,000 -500 0

Primary City

-800 -700 -600 -500 -400 -300 -200 -100 0 100 200 300 400
Source: Authors’ analysis of UCR data, 1990 decennial census, and 2005-2009 ACS five-year estimates

BROOKINGS | May 2011 11

 Yet the ability of these controls to explain the observed changes in crime varies widely across com-
munity types (Figure 4). For primary cities, roughly 30 percent of the decline in violent crime can be
explained by the demographic control variables.22 However, for high-density suburbs, these controls
(at least the demographics we control for in Table 4) explain only about 12 percent of the decline, and
do not at all explain the changes seen in emerging suburbs and exurbs. In fact, since the estimated
changes in violent crime are larger when we adjust for demographics, the regressions suggest that
had the demography of these areas not changed in the manner that they had, crime would have
increased by even more than it actually did.
The results for property crime are qualitatively similar to those for violent crime, although the
observed changes are larger (Figure 5). The first model for property crime confirms the statistical sig-
nificance of the declines seen in property crime for all community types, with the greatest decreases
occurring in more urbanized communities. When we control for demographic changes, the coefficients
for each community type variable become smaller, suggesting that within each community type, demo-
graphic change has served to lower crime rates.

D. As crime rates fell and communities diversified, relationships between crime and
community demographic characteristics weakened significantly.
As crime rates declined between 1990 and 2008, and the gap between city and suburban rates nar-
rowed, the results from Finding C suggest that the empirical relationships between community demo-
graphics and crime rates weakened. In this section, we directly explore that proposition.
Tables 6 and 7 present the results of a series of regressions in which we test the relationships
between crime rates and specific demographic characteristics in each year. We start with a simple
regression of community-level crime rates on a selected demographic characteristic (Specification
(1) in the tables). By comparing the results estimated in each year, this model indicates whether
the empirical relationship between crime rates and the demographic characteristic in question has

Table 6. Results from Regressing Violent Crime Rates on Various City Level Demographic
Characteristics: Separate Model Estimates for 1990, 2000, and 2008

Coefficient Estimates
1990 2000 2008
Proportion Black
Specification (1) 5,619* 3,832* 3,802*
Specification (2) 6,500* 4,459* 4,285*

Proportion Poor
Specification (1) 12,492* 9,859* 9,789*
Specification (2) 13,936* 10,662* 9,771

Proportion Hispanic
Specification (1) 2,763* 1,189* 514*
Specification (2) 5,342* 2,904* 2,334*

Proportion Foreign-Born
Specification (1) 3,230* 333* -707*
Specification (2) 6,356* 2,673* 1,942*

Source: Authors’ analysis of UCR, decennial census data, and 2005-2009 ACS five-year estimates
All regressions are weighted by the community-level population for the given year of the dependent variable. Specification (1)
is a simple bivariate regression of the crime rate on the demographic characteristic. Specification (2) adds a dummy variable
for non-incorporated areas in the balance of counties as well as 100 dummy variables for each metropolitan area.
*Coefficient statistically significant at the one percent level of confidence.

12 BROOKINGS | May 2011

changed. (For example, if a coefficient in 2008 is smaller than the comparable coefficient in 1990,
then the association between these two variables has weakened over time.)
We then add a complete set of dummy variables for each metropolitan area, as well as a dummy
variable indicating communities for which we used average county-level characteristics (Specification
(2)).23 Including the complete set of metropolitan area dummy variables ensures that coefficients are
estimated using only the variation in the demographic variable and crime rates occurring across com-
munities within each of the metropolitan areas.24
The results of the simple regression model in Table 6 shows a substantial weakening between
violent crime and each demographic variable, with near total attenuation for the proportion Hispanic
and an eventually negative coefficient for the proportion of foreign-born residents. The results for
specification (2) suggest stronger relationships within each year and metropolitan area. However, we
still observe a substantial decrease in the strength of relationships between these demographic traits
and crime over time.
Property crime exhibits very similar patterns. In the first specification, the coefficient on the propor-
tion black declines by almost half—from 10,006 in 1990 to 5,421 in 2008. We find a similar weakening
of the relationship between poverty and property crime. Property crime and the proportion Hispanic
are positively associated in all years, though the coefficient in 2008 is only 22 percent of the value of
the comparable coefficient in 1990. Finally, by 2008 there is no measurable relationship between the
property crime rate and the proportion foreign-born. Adding dummy variables for each metropolitan
area and a dummy indicating a balance-of-county observation increases all estimates. However, the
patterns of weakening relationships between these characteristics and crime over time still hold.
In sum, the disproportionate declines in crime in more urbanized areas coincided with dispropor-
tionate declines in crime in areas with higher shares of minority and poor residents. Consequently,
such average community-level characteristics are less predictive of crime today than in years past.

Table 7. Results from Regressing Property Crime Rates on Various City Level Demographic
Characteristics: Separate Model Estimates for 1990, 2000, and 2008

Coefficient Estimates
1990 2000 2008
Proportion Black
Specification (1) 10,006* 6,790* 5,421*
Specification (2) 11,077* 7,803* 6,289*

Proportion Poor
Specification (1) 24,902* 17,657* 14,813*
Specification (2) 23,361* 17,774* 13,850*

Proportion Hispanic
Specification (1) 6,793* 2,296* 1,469*
Specification (2) 8,853* 4,796* 3,535*

Proportion Foreign-Born
Specification (1) 6,952* 858* -123
Specification (2) 13,124* 5,810* 3,875*

Source: Authors’ analysis of UCR, decennial census data, and 2005-2009 ACS five-year estimates
All regressions are weighted by the community-level population for the given year of the dependent variable. Specification (1)
is a simple bivariate regression of the crime rate on the demographic characteristic. Specification (2) adds a dummy variable
for non-incorporated areas in the balance of counties as well as 100 dummy variables for each metropolitan area.
*Coefficient statistically significant at the one percent level of confidence.

BROOKINGS | May 2011 13

 Conclusion

U
.S. metropolitan areas today are considerably safer than they were twenty years ago, and
residents of these regions are less likely to be victimized by violent and property crimes
of all sorts. Declining crime rates have had disproportionate impacts on urbanized, poor,
and relatively minority areas where crime levels were the highest at the beginning of our
study period.
The implications of these trends for the relative attractiveness of centralized urban locations are
profound. Not only do cities become a more attractive residential choice; they also become more
attractive locations for leisure and recreation, tourism, and other important sources of economic
activity that benefit both the residents of central cities as well as the residents from all communities
within a metropolitan area. One might expect such concentrated crime declines to be capitalized into
central city property values, and perhaps contribute to a slowing of population decentralization (with
subsequent, albeit small, implications for sprawl and metropolitan area traffic congestion).
At the same time, we find very little evidence that the decentralization of poverty and minority
households has contributed to higher crime in distant suburbs. Our results clearly indicate that demo-
graphic changes in the nation’s suburbs, if anything, lower crime rates all else held equal. Indeed, we
document substantial property crime rate declines in all community types throughout the 100 largest
metropolitan areas, though we do observe modest increases in violent crime in emerging suburbs
and exurbs. While changing demographics do not explain these statistically significant, though slight,
increases in violent crime, future research should explore other factors that may be contributing to
these trends.
Moreover, the narrowing of the gap between city and suburban crime rates, coupled with the fact
that intra-metropolitan crime trends tend to move together (whether up or down), underscore that
crime is not solely an urban challenge, but a metropolitan-wide issue. Thinking about these trends
from a metropolitan perspective affords opportunities for policymakers and public safety officials
to learn from neighboring communities and regions about what crime-reduction policies might be
effective in their region. In particular, growing and urbanizing suburbs and communities where crime
reduction stalled or lost ground since the 1990s stand to benefit from lessons learned and successful
policies adopted in communities that achieved significant declines in both violent and property crime
over the past two decades.

Appendix A. Violent Crime per 100,000 Residents, 100 Largest Metropolitan Areas, by City and Suburb

Violent Crime Rates Violent Crime Violent Crime

in 1990 in 2000 in 2008
All City Suburb All City Suburb All City Suburb
Akron, OH 1,738 3,932 636 1,660 3,075 1,022 1,861 4,398 810
Albany-Schenectady-Troy, NY 950 2,957 672 1,543 4,457 1,175 1,077 3,219 824
Albuquerque, NM 2,251 3,277 358 2,318 2,999 1,202 2,141 2,556 1,443
Allentown-Bethlehem-Easton, PA-NJ 621 528 638 882 2,164 663 855 2,527 586
Atlanta-Sandy Springs-Marietta, GA 1,801 6,975 1,119 1,488 5,724 1,060 1,461 2,920 1,309
Augusta-Richmond County, GA-SC 1,104 1,189 1,041 1,991 2,558 1,632 2,353 3,681 1,609
Austin-Round Rock, TX 2,274 3,339 975 1,860 2,488 1,178 1,660 2,512 950
Bakersfield, CA 1,609 1,077 1,853 1,020 499 1,300 1,231 1,400 1,128
Baltimore-Towson, MD 2,829 5,225 1,759 2,653 5,030 1,841 1,979 3,146 1,615
Baton Rouge, LA 1,937 4,071 685 1,970 2,932 1,521 2,226 4,193 1,437
Birmingham-Hoover, AL 962 2,232 477 1,302 1,216 1,328 2,127 3,603 1,743
Boise City-Nampa, ID 1,061 1,266 929 1,389 1,587 1,259 1,100 1,247 1,027
Boston-Cambridge-Quincy, MA-NH 971 3,693 453 1,045 2,437 787 1,150 2,166 961
Bradenton-Sarasota-Venice, FL 1,951 4,121 1,738 1,828 2,578 1,759 1,796 2,825 1,710

14 BROOKINGS | May 2011

 Appendix A. Violent Crime per 100,000 Residents, 100 Largest Metropolitan Areas, by City and Suburb (continued)

Violent Crime Rates Violent Crime Violent Crime

in 1990 in 2000 in 2008
All City Suburb All City Suburb All City Suburb
Bridgeport-Stamford-Norwalk, CT 1,571 4,138 462 1,140 2,517 552 1,163 2,465 631
Buffalo-Niagara Falls, NY 1,398 2,882 856 1,653 3,460 1,075 1,393 3,217 837
Cape Coral-Fort Myers, FL 751 1,100 650 1,431 730 1,642 1,203 1,143 1,225
Charleston-North Charleston-Summerville, SC 2,141 1,873 2,190 3,224 2,780 3,318 2,897 2,490 2,979
Charlotte-Gastonia-Concord, NC-SC 2,476 4,283 1,340 2,160 2,839 1,562 1,997 2,577 1,520
Chattanooga, TN-GA 1,183 2,275 589 2,524 4,680 1,479 1,972 2,990 1,469
Chicago-Naperville-Joliet, IL-IN-WI 3,077 5,750 1,557 673 1,507 231 555 1,135 269
Cincinnati-Middletown, OH-KY-IN 1,384 4,530 649 1,130 3,651 652 1,409 4,159 923
Cleveland-Elyria-Mentor, OH 1,619 4,537 725 1,318 4,774 361 1,528 4,989 637
Colorado Springs, CO 734 736 729 1,040 1,198 717 875 1,077 556
Columbia, SC 2,183 4,121 1,764 2,512 3,858 2,217 2,288 2,399 2,265
Columbus, OH 1,522 2,704 575 2,116 4,044 606 2,446 4,383 1,024
Dallas-Fort Worth-Arlington, TX 2,711 4,535 1,351 2,180 3,488 1,309 1,667 2,460 1,191
Dayton, OH 1,917 6,794 599 1,396 4,767 589 1,674 4,088 1,142
Denver-Aurora, CO 1,460 1,993 1,090 862 1,211 648 955 1,236 788
Des Moines-West Des Moines, IA 1,235 2,065 516 1,311 2,029 795 1,705 3,355 773
Detroit-Warren-Livonia, MI 1,787 2,889 1,374 1,780 3,030 1,384 1,913 3,777 1,358
El Paso, TX 2,740 2,861 1,929 2,449 2,612 1,657 1,959 1,917 2,156
Fresno, CA 1,951 2,505 1,355 1,954 2,714 1,121 1,551 2,141 938
Grand Rapids-Wyoming, MI 1,889 4,096 975 2,052 3,761 1,432 1,834 3,068 1,424
Greensboro-High Point, NC 991 1,281 736 1,053 1,184 933 1,250 1,469 1,036
Greenville-Mauldin-Easley, SC 1,602 2,370 1,449 2,065 3,518 1,904 2,165 2,798 2,099
Harrisburg-Carlisle, PA 929 4,449 496 805 2,617 615 1,231 4,325 934
Hartford-West Hartford-East Hartford, CT 1,392 7,022 602 1,026 3,261 710 1,182 3,015 926
Honolulu, HI 1,237 1,237 — 1,040 1,040 — 1,282 1,282 —
Houston-Sugar Land-Baytown, TX 1,893 2,697 1,279 2,246 3,317 1,498 2,121 2,943 1,603
Indianapolis-Carmel, IN 1,102 2,277 404 1,097 1,838 300 1,673 3,058 426
Jackson, MS 1,018 1,932 300 1,546 3,418 447 1,161 2,393 574
Jacksonville, FL 2,629 3,299 1,192 2,409 2,896 1,507 2,106 2,568 1,385
Kansas City, MO-KS 2,669 5,875 1,146 1,795 3,441 1,177 1,872 3,504 1,344
Knoxville, TN 840 1,904 365 1,958 2,627 1,694 1,715 2,691 1,359
Lakeland-Winter Haven, FL 2,738 2,655 2,756 2,271 2,034 2,316 1,904 1,729 1,937
Las Vegas-Paradise, NV 1,239 1,394 474 2,035 2,088 1,855 2,377 2,711 1,502
Little Rock-North Little Rock-Conway, AR 2,090 3,722 1,294 1,948 3,414 1,322 3,211 5,265 2,415
Los Angeles-Long Beach-Santa Ana, CA 2,127 3,071 1,564 1,422 2,017 1,081 1,161 1,494 971
Louisville-Jefferson County, KY-IN 584 869 487 697 1,668 426 1,124 1,687 555
Madison, WI 973 1,442 605 688 1,137 371 697 985 500
McAllen-Edinburg-Mission, TX 1,833 2,226 1,723 1,993 2,046 1,980 1,883 1,837 1,893
Memphis, TN-MS-AR 1,409 2,071 559 2,529 3,523 1,365 4,397 6,443 2,171
Miami-Fort Lauderdale-Pompano Beach, FL 2,797 4,789 2,464 2,134 3,384 1,966 1,684 2,246 1,592
Milwaukee-Waukesha-West Allis, WI 1,698 3,345 411 1,125 2,289 356 1,295 2,709 395
Minneapolis-St. Paul-Bloomington, MN-WI 931 1,453 754 1,267 3,289 685 961 2,561 577
Modesto, CA 1,664 1,859 1,508 1,748 1,639 1,827 1,272 1,486 1,131
Nashville-Davidson—Murfreesboro—Franklin, TN 2,203 3,937 644 3,172 4,888 1,915 2,826 4,357 1,852
New Haven-Milford, CT 2,196 7,573 1,162 964 4,344 607 1,256 5,440 826
New Orleans-Metairie-Kenner, LA 1,879 2,638 1,388 1,418 1,097 1,605 1,835 2,428 1,635
New York-Northern New Jersey-Long Island, NY-NJ-PA 1,615 2,489 986 827 1,038 675 641 636 644

BROOKINGS | May 2011 15

 Appendix A. Violent Crime per 100,000 Residents, 100 Largest Metropolitan Areas, by City and Suburb (continued)

Violent Crime Rates Violent Crime Violent Crime

in 1990 in 2000 in 2008
All City Suburb All City Suburb All City Suburb
Ogden-Clearfield, UT 1,155 3,497 650 1,255 3,121 869 1,250 3,732 810
Oklahoma City, OK 1,555 2,414 918 1,450 2,136 952 1,522 2,298 980
Omaha-Council Bluffs, NE-IA 1,359 1,997 800 1,724 2,426 1,009 1,348 1,653 1,030
Orlando-Kissimmee, FL 1,935 3,729 1,656 2,173 4,844 1,834 1,981 3,887 1,742
Oxnard-Thousand Oaks-Ventura, CA 974 1,130 815 946 1,171 708 798 944 643
Palm Bay-Melbourne-Titusville, FL 1,486 1,383 1,504 1,748 2,464 1,608 1,916 1,557 1,998
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD 1,463 2,038 1,228 1,693 3,284 1,124 1,569 3,228 1,028
Phoenix-Mesa-Scottsdale, AZ 1,841 2,173 1,289 1,493 1,673 1,236 1,276 1,464 1,063
Pittsburgh, PA 1,027 4,296 484 985 3,572 584 980 3,109 671
Portland-South Portland-Biddeford, ME 1,172 3,533 791 938 2,048 778 984 1,598 903
Portland-Vancouver-Beaverton, OR-WA 1,634 3,113 950 1,188 2,017 748 875 1,439 607
Poughkeepsie-Newburgh-Middletown, NY 1,367 5,332 1,161 952 2,799 862 794 3,618 668
Providence-New Bedford-Fall River, RI-MA 1,324 3,293 1,090 1,276 1,959 1,192 1,446 2,143 1,362
Provo-Orem, UT 463 585 406 585 908 457 550 989 402
Raleigh-Cary, NC 712 1,101 379 1,102 1,554 715 935 1,362 553
Richmond, VA 1,346 2,471 1,041 1,949 3,874 1,526 1,766 3,160 1,494
Riverside-San Bernardino-Ontario, CA 1,963 2,800 1,758 1,376 2,294 1,170 1,113 1,854 970
Rochester, NY 1,855 6,170 580 1,235 3,639 605 1,289 3,491 748
Sacramento—Arden-Arcade—Roseville, CA 1,558 2,110 1,343 1,150 1,707 943 1,124 1,734 855
Salt Lake City, UT 1,704 4,343 1,026 1,556 2,466 1,360 1,714 3,748 1,325
San Antonio, TX 1,260 1,452 881 2,014 2,326 1,385 1,979 2,160 1,610
San Diego-Carlsbad-San Marcos, CA 1,546 1,793 1,367 1,300 1,433 1,198 1,151 1,194 1,119
San Francisco-Oakland-Fremont, CA 1,943 2,808 1,493 1,356 1,762 1,154 1,354 2,086 994
San Jose-Sunnyvale-Santa Clara, CA 1,368 1,469 1,183 1,080 1,151 948 875 922 785
Scranton—Wilkes-Barre, PA 549 1,249 437 383 203 411 606 1,008 547
Seattle-Tacoma-Bellevue, WA 1,922 3,280 1,328 1,437 2,263 1,107 1,237 1,812 1,019
Springfield, MA 1,204 4,422 241 2,117 5,416 1,185 2,134 4,742 1,407
St. Louis, MO-IL 1,925 6,323 1,129 1,186 4,723 665 1,186 3,837 805
Stockton, CA 1,728 2,338 1,251 1,818 2,823 1,051 1,968 2,926 1,246
Syracuse, NY 953 1,914 646 1,286 2,998 804 1,141 3,594 558
Tampa-St. Petersburg-Clearwater, FL 2,130 3,167 1,689 2,589 3,880 2,099 1,967 2,666 1,734
Toledo, OH 3,056 5,495 672 2,796 5,545 456 2,546 4,335 952
Tucson, AZ 2,338 3,016 1,313 2,193 2,954 1,155 1,970 2,876 975
Tulsa, OK 1,487 2,208 829 1,570 2,413 859 1,688 2,593 1,026
Virginia Beach-Norfolk-Newport News, VA-NC 1,305 1,247 1,383 2,079 2,299 1,828 2,205 2,311 2,094
Washington-Arlington-Alexandria, DC-VA-MD-WV 1,516 2,661 1,213 1,403 2,094 1,251 1,215 2,104 1,030
Wichita, KS 2,907 4,222 961 2,476 3,462 986 3,070 4,196 1,353
Worcester, MA* — — 340 1,120 2,219 793 1,275 2,199 1,007
Youngstown-Warren-Boardman, OH-PA 938 3,496 480 771 3,559 339 1,622 4,353 1,226

*Data for the city of Worcester is missing in 1990.

16 BROOKINGS | May 2011

 Appendix B. Property Crime per 100,000 Residents, 100 Largest Metropolitan Areas, by City and Suburb

Property Crime Rates Property Crime Rates Property Crime Rates

in 1990 in 2000 in 2008
All City Suburb All City Suburb All City Suburb
Akron, OH 3,589 6,952 1,901 2,310 2,524 2,214 3,057 5,410 2,082
Albany-Schenectady-Troy, NY 3,416 6,459 2,995 2,933 6,564 2,474 2,454 4,782 2,179
Albuquerque, NM 6,206 8,955 1,133 5,851 7,838 2,595 4,784 6,187 2,427
Allentown-Bethlehem-Easton, PA-NJ 2,913 5,882 2,366 2,402 4,528 2,039 2,624 5,267 2,198
Atlanta-Sandy Springs-Marietta, GA 7,145 17,207 5,819 4,189 12,299 3,370 4,001 8,356 3,546
Augusta-Richmond County, GA-SC 5,443 8,277 3,339 4,143 5,954 2,994 4,721 7,290 3,282
Austin-Round Rock, TX 7,944 11,277 3,878 4,168 5,798 2,396 3,955 6,252 2,040
Bakersfield, CA 5,387 7,132 4,588 3,166 3,905 2,767 3,735 4,985 2,964
Baltimore-Towson, MD 5,563 8,339 4,323 4,501 7,831 3,363 3,553 4,915 3,128
Baton Rouge, LA 5,909 10,975 2,936 5,783 8,899 4,326 3,868 5,746 3,115
Birmingham-Hoover, AL 4,942 9,560 3,181 3,502 7,347 2,340 4,621 8,807 3,533
Boise City-Nampa, ID 4,572 5,180 4,182 3,495 4,160 3,060 2,220 2,917 1,876
Boston-Cambridge-Quincy, MA-NH 3,689 9,466 2,590 2,298 4,971 1,803 2,159 3,660 1,881
Bradenton-Sarasota-Venice, FL 6,524 9,132 6,268 4,003 5,941 3,825 4,261 4,453 4,245
Bridgeport-Stamford-Norwalk, CT 4,954 8,200 3,551 2,559 3,778 2,037 2,079 3,261 1,596
Buffalo-Niagara Falls, NY 4,351 7,613 3,159 2,865 6,056 1,844 2,990 5,961 2,084
Cape Coral-Fort Myers, FL 4,362 4,794 4,237 4,249 3,341 4,523 3,509 3,097 3,662
Charleston-North Charleston-Summerville, SC 5,965 7,071 5,760 5,189 6,735 4,861 4,060 4,729 3,925
Charlotte-Gastonia-Concord, NC-SC 7,083 10,371 5,015 4,957 6,703 3,420 4,620 6,231 3,293
Chattanooga, TN-GA 5,450 10,502 2,698 5,162 9,970 2,833 4,390 7,647 2,782
Chicago-Naperville-Joliet, IL-IN-WI 5,611 8,112 4,189 2,196 5,379 508 2,413 4,477 1,398
Cincinnati-Middletown, OH-KY-IN 3,707 6,606 3,029 2,702 6,065 2,065 2,939 6,295 2,346
Cleveland-Elyria-Mentor, OH 3,580 7,359 2,422 2,162 5,610 1,208 2,280 5,854 1,361
Colorado Springs, CO 5,724 7,062 2,792 3,668 4,545 1,875 3,134 4,268 1,348
Columbia, SC 5,371 11,589 4,027 4,264 8,292 3,379 3,977 5,986 3,558
Columbus, OH 5,594 9,114 2,772 4,916 8,320 2,250 4,239 6,777 2,376
Dallas-Fort Worth-Arlington, TX 8,960 12,836 6,069 4,957 7,226 3,446 4,235 5,889 3,242
Dayton, OH 4,771 9,992 3,361 3,888 8,844 2,702 3,211 6,280 2,535
Denver-Aurora, CO 6,153 7,252 5,389 3,681 4,796 2,999 3,036 3,513 2,754
Des Moines-West Des Moines, IA 5,814 8,432 3,547 4,299 6,529 2,695 3,093 5,110 1,953
Detroit-Warren-Livonia, MI 5,779 9,005 4,571 3,767 7,082 2,716 3,151 5,594 2,423
El Paso, TX 9,360 10,281 3,138 4,874 5,428 2,182 3,110 3,277 2,320
Fresno, CA 7,058 9,465 4,469 5,158 6,875 3,279 3,990 4,724 3,227
Grand Rapids-Wyoming, MI 4,581 7,072 3,550 3,415 5,518 2,652 2,936 4,866 2,295
Greensboro-High Point, NC 5,032 7,452 2,904 4,736 6,304 3,296 4,661 6,341 3,019
Greenville-Mauldin-Easley, SC 4,586 7,052 4,096 3,861 8,505 3,347 3,775 6,482 3,492
Harrisburg-Carlisle, PA 3,172 8,057 2,570 2,191 4,458 1,953 2,198 5,658 1,866
Hartford-West Hartford-East Hartford, CT 4,271 12,467 3,121 3,384 8,024 2,728 2,826 5,117 2,507
Honolulu, HI 5,814 5,814 — 5,062 5,062 — 3,506 3,506 —
Houston-Sugar Land-Baytown, TX 7,245 10,081 5,078 4,258 5,826 3,165 3,842 5,105 3,048
Indianapolis-Carmel, IN 4,277 5,594 3,496 2,712 3,900 1,433 3,795 6,134 1,688
Jackson, MS 5,389 10,100 1,688 4,531 9,910 1,373 3,466 7,674 1,463
Jacksonville, FL 7,176 8,648 4,018 4,883 5,844 3,102 4,652 5,761 2,921
Kansas City, MO-KS 6,319 12,188 3,530 4,241 8,545 2,627 3,953 7,659 2,754
Knoxville, TN 3,837 7,238 2,318 3,526 5,466 2,764 3,744 6,776 2,640
Lakeland-Winter Haven, FL 7,844 10,800 7,221 5,196 7,086 4,832 3,972 5,664 3,658
Las Vegas-Paradise, NV 5,712 6,442 2,120 3,824 3,903 3,557 3,611 3,972 2,665

BROOKINGS | May 2011 17

 Appendix B. Property Crime per 100,000 Residents, 100 Largest Metropolitan Areas, by City and Suburb (continued)

Property Crime Rates Property Crime Rates Property Crime Rates

in 1990 in 2000 in 2008
All City Suburb All City Suburb All City Suburb
Little Rock-North Little Rock-Conway, AR 7,346 12,954 4,613 5,421 8,729 4,011 5,367 8,233 4,257
Los Angeles-Long Beach-Santa Ana, CA 5,650 6,961 4,869 2,872 3,470 2,531 2,516 2,629 2,452
Louisville-Jefferson County, KY-IN 3,885 5,730 3,259 2,312 5,082 1,538 3,475 4,727 2,212
Madison, WI 4,589 6,915 2,763 2,732 3,787 1,989 2,774 3,785 2,084
McAllen-Edinburg-Mission, TX 6,097 9,814 5,055 5,191 7,848 4,580 4,632 6,056 4,321
Memphis, TN-MS-AR 5,607 8,420 1,993 5,517 7,773 2,875 5,705 8,055 3,148
Miami-Fort Lauderdale-Pompano Beach, FL 9,427 14,448 8,589 5,759 7,976 5,461 4,672 5,379 4,558
Milwaukee-Waukesha-West Allis, WI 5,503 8,363 3,268 3,953 6,473 2,288 3,835 6,134 2,373
Minneapolis-St. Paul-Bloomington, MN-WI 3,793 3,721 3,817 3,631 6,006 2,948 3,081 5,047 2,609
Modesto, CA 5,732 5,962 5,548 4,643 5,134 4,283 4,483 5,285 3,953
Nashville-Davidson—Murfreesboro—Franklin, TN 4,190 6,500 2,115 4,813 7,352 2,953 3,585 5,489 2,374
New Haven-Milford, CT 7,021 14,214 5,638 2,791 7,055 2,340 2,492 6,843 2,045
New Orleans-Metairie-Kenner, LA 7,347 10,177 5,515 4,845 6,035 4,152 3,953 5,404 3,465
New York-Northern New Jersey-Long Island, NY-NJ-PA 5,181 7,503 3,509 2,047 2,768 1,528 1,720 1,870 1,611
Ogden-Clearfield, UT 4,308 8,777 3,345 3,431 6,730 2,748 2,640 5,135 2,198
Oklahoma City, OK 7,334 10,724 4,818 6,006 10,201 2,965 4,609 7,108 2,866
Omaha-Council Bluffs, NE-IA 4,414 6,115 2,921 4,533 6,065 2,974 3,415 4,298 2,497
Orlando-Kissimmee, FL 6,957 12,065 6,163 5,032 10,071 4,392 4,466 8,810 3,921
Oxnard-Thousand Oaks-Ventura, CA 3,496 4,302 2,673 2,047 2,462 1,606 1,994 2,260 1,708
Palm Bay-Melbourne-Titusville, FL 6,437 6,022 6,512 2,960 4,115 2,734 3,400 2,382 3,632
Philadelphia-Camden-Wilmington, PA-NJ-DE-MD 4,064 5,843 3,335 3,168 4,955 2,530 2,944 4,342 2,489
Phoenix-Mesa-Scottsdale, AZ 7,743 8,652 6,234 5,596 6,238 4,675 4,247 4,739 3,688
Pittsburgh, PA 2,583 7,433 1,776 1,882 4,891 1,415 1,990 4,200 1,669
Portland-South Portland-Biddeford, ME 4,528 10,995 3,484 2,506 4,139 2,270 2,330 4,419 2,053
Portland-Vancouver-Beaverton, OR-WA 5,381 8,988 3,715 4,478 6,321 3,498 3,207 4,978 2,365
Poughkeepsie-Newburgh-Middletown, NY 2,980 6,892 2,777 1,987 3,867 1,897 1,973 3,790 1,892
Providence-New Bedford-Fall River, RI-MA 4,606 10,383 3,920 2,917 6,903 2,428 2,764 5,471 2,439
Provo-Orem, UT 3,876 4,350 3,652 3,287 3,998 3,007 2,470 2,816 2,353
Raleigh-Cary, NC 4,354 5,970 2,974 4,049 5,485 2,821 2,620 3,125 2,166
Richmond, VA 4,980 10,349 3,525 3,834 7,886 2,942 2,838 4,561 2,503
Riverside-San Bernardino-Ontario, CA 5,675 7,064 5,334 3,287 4,559 3,001 3,067 3,989 2,891
Rochester, NY 4,573 9,802 3,027 3,371 7,106 2,391 2,633 5,420 1,948
Sacramento—Arden-Arcade—Roseville, CA 5,904 7,992 5,094 3,775 5,713 3,056 3,339 4,660 2,758
Salt Lake City, UT 7,177 12,190 5,889 4,685 9,094 3,735 4,491 9,447 3,543
San Antonio, TX 9,358 11,888 4,379 5,625 6,947 2,958 5,953 7,307 3,197
San Diego-Carlsbad-San Marcos, CA 5,961 8,177 4,352 2,867 3,246 2,575 2,829 3,225 2,534
San Francisco-Oakland-Fremont, CA 5,893 7,831 4,884 3,566 4,461 3,119 3,606 4,532 3,152
San Jose-Sunnyvale-Santa Clara, CA 4,145 4,294 3,872 2,209 2,079 2,452 2,420 2,386 2,485
Scranton—Wilkes-Barre, PA 1,683 3,110 1,455 997 602 1,056 2,200 3,639 1,989
Seattle-Tacoma-Bellevue, WA 6,405 10,458 4,631 5,030 7,301 4,120 4,149 5,865 3,496
Springfield, MA 2,957 7,157 1,701 3,214 6,495 2,287 2,808 4,900 2,224
St. Louis, MO-IL 4,593 11,291 3,381 3,388 12,433 2,056 3,064 8,621 2,265
Stockton, CA 7,500 10,258 5,342 4,444 5,861 3,365 4,923 6,307 3,880
Syracuse, NY 3,701 6,261 2,883 2,553 5,165 1,817 2,129 4,557 1,551
Tampa-St. Petersburg-Clearwater, FL 6,847 11,483 4,876 4,940 7,448 3,988 4,100 5,271 3,709
Toledo, OH 6,005 8,620 3,449 4,312 7,067 1,968 4,029 5,882 2,377
Tucson, AZ 8,650 11,338 4,589 6,339 8,503 3,388 2,703 2,294 3,153

18 BROOKINGS | May 2011

 Appendix B. Property Crime per 100,000 Residents, 100 Largest Metropolitan Areas, by City and Suburb (continued)

Property Crime Rates Property Crime Rates Property Crime Rates

in 1990 in 2000 in 2008
All City Suburb All City Suburb All City Suburb
Tulsa, OK 5,732 8,342 3,352 4,010 5,815 2,489 3,735 6,054 2,039
Virginia Beach-Norfolk-Newport News, VA-NC 5,896 6,693 4,845 4,046 4,792 3,196 3,470 3,855 3,069
Washington-Arlington-Alexandria, DC-VA-MD-WV 4,635 7,589 3,853 3,051 4,526 2,726 2,917 4,091 2,673
Wichita, KS 6,060 8,282 2,773 4,242 5,754 1,957 4,278 5,576 2,298
Worcester, MA* — — 1,529 2,269 4,353 1,649 2,045 3,618 1,590
Youngstown-Warren-Boardman, OH-PA 2,714 6,887 1,966 1,731 6,676 965 3,219 5,757 2,851

*Data for the city of Worcester is missing in 1990.

Appendix C. Ordinary Least Squared Regression of the Community-Level Changes in Crime Rates
Between 1990 and 2008 on Community Type and Demographic Change

Dependent Variable = Change in Violent Crime Dependent Variable = Change in Property Crime
(1) (2) (3) (4)
City -799a -575a -3,850a -3,205a
(28) (35) (51) (64)
High density suburb -246a -217a -1,934a -1,520a
(31) (38) (56) (72)
Mature suburb -40 -54 -1,626a -1,279a
(33) (41) (60) (76)
Emerging suburb 107b 196a -910a -519a
(47) (52) (86) (96)
Exurb 313a 399a -530a -246c
(70) (69) (128) (127)
Δ, proportion black - 5,068a - 4,424a
(312) (577)
Δ, proportion poor - 1,086a - 6,510a
(603) (1114)
Δ, proportion foreign-born - -948b - -3,604a
(431) (797)
Δ, proportion Hispanic - -1,873a - -5,742a
(308) (568)
R2 0.146 0.223 0.591 0.623

N 5,383 5,251 5,383 5,251

Source: Authors’ analysis of UCR, decennial census data, and 2005-2009 ACS five-year estimates
All regressions omit a constant term. Hence, the coefficients on the community type indicator variables provide the average change for each type.
All regressions are weighted by the average of the 1990 and 2005-2009 community-level population. The difference in the number of observations between the
regressions with and without the demographic controls is driven by missing values for 132 observations for the proportion poor either in 1990 or 2005-2009.
a. Change statistically significant at the one percent level of confidence.
b. Change statistically significant at the five percent level of confidence.
c. Change statistically significant at the 10 percent level of confidence.

BROOKINGS | May 2011 19

 References

Butcher, Kirsten F. and Anne Morrison Piehl (1998a), “Recent Immigrants: Unexpected Implications for
Crime and Incarceration,” Industrial and Labor Relations Review, 51(4): 654-679.

Butcher, Kirsten F. and Anne Morrison Piehl (1998b), “Cross-City Evidence on the Relationship Between
Immigration and Crime,” Journal of Policy Analysis and Management, 17(3): 457-493.

Freeman, Richard B. (1996), “Why Do So Many Young American Men Commit Crimes and What Might
We Do About It?” Journal of Economic Perspectives, 10(1): 25-42.

Raphael, Steven and Melissa Sills (2005) “Urban Crime in the United States,” in Richard Arnott and
Dan McMillen (eds.) A Companion to Urban Economics, Blackwell Publishing.

Raphael, Steven and Michael Stoll (2009), Do Prisons Make Us Safer? The Benefits and Costs of the
Prison Boom, Russell Sage Foundation, New York (2009).

Rennison, Callie (2002), “Criminal Victimization 2001: Changes 2000-01 with Trends 1993-2001,” Bu-
reau of Justice Statistics National Crime Victimization Survey, NCJ 194610.

Endnotes
1. See the Senate testimony of Jens Ludwig, September 6. See e.g., The Crime Drop in America, 2nd edition, Alfred
19, 2006 and reference cited therein. Testimony tran- Blumstein and Joel Wallman, eds. (Cambridge: Cambridge
script at http://judiciary.senate.gov/hearings/testimony. University Press, 2006); Steven Levitt, “Understanding
cfm?id=2068&wit_id=5749, accessed on April 11, 2011. Why Crime Fell in the 1990s: Four Factors that Explain
the Decline and Six that Do Not” Journal of Economic
2. Figures from Bureau of Justice Statistics, http://bjs.ojp. Perspectives 18(1) (2004): 163-190.
usdoj.gov/index.cfm?ty=tp&tid=5, accessed on September
8, 2010. 7. Metropolitan Policy Program at Brookings, “State of
Metropolitan America: On the Front Lines of Demographic
3. Ludwig (see note 1) estimates the total costs of crime to be Transformation” (Washington: 2010).
slightly over $2 trillion per year in 2006.
8. See Rennison (2002) for a detailed discussion of these defi-
4. See Steven Raphael and Melissa Sills, “Urban Crime in nitions.
the United States,” in Richard Arnott and Dan McMillen,
eds, A Companion to Urban Economics (Boston: Blackwell 9. For example, many universities have their own police
Publishing, 2005), and Steven Raphael and Michael Stoll, departments. Hence, a city that has within its boundaries
Do Prisons Make Us Safer? The Benefits and Costs of the a single university with its own department will have two
Prison Boom (New York: Russell Sage Foundation, 2009). reporting units: one for the city’s police department and
one for the university police department.
5. These figures come from various years of the National
Criminal Victimization Survey (NCVS), accessed on April 11, 10. 28 states, mostly on the east coast, use the MCD system,
2011 at the Bureau of Justice Statistics webpage http://bjs. while CCSD’s are found mostly on the west coast.
ojp.usdoj.gov/. Here we report figures from the National
Criminal Victimization Survey since this national level data 11. For all three geography types, we use the 2005 Law
does not suffer from under-reporting of criminal incidents Enforcement Agency Identifier Crosswalk to mach each
to the police (as it is a survey of victims/potential victims). police agency in the UCR data to Census Bureau data.
Throughout our analysis we employ data from the Federal
Bureau of Investigation’s Uniform Crime Reports, a data 12. The Census Bureau defines “urbanized area” as an area
series that is based on crimes reported to the authorities. with a population density of at least 1,000 people per
While victimization surveys provide rate estimates that are square mile of land that has a minimum residential popula-
not biased by under-reporting, the NCVS does not permit tion of 50,000 people.
sub-national geographical analysis.

20 BROOKINGS | May 2011

13. Post-Census 2000, five-year estimates from the ACS 19. Of course, these changing distributions can be driven
represent the only demographic data source with sample either by decentralization of the existing population or dif-
sizes sufficient to produce estimates for geographies with ferential population growth rates, with higher growth in the
populations under 20,000. more distant suburban cities.

14. In the model estimates presented below, we explored the 20. Authors’ analysis of 1990 decennial census data and 2005-
sensitivity of our results to this particular imputation for 2009 American Community Survey.
balance-of-county observations. First, we reran all models
omitting these observations. Second, we reran all models 21. Specifically, for violent crime rates (models (1) and (2)) and
including a dummy variable indicating counties with such property crime rates (models (3) and (4)) we first estimate
an imputation. All of the results are robust to these specifi- models where the specification only includes the com-
cation changes. munity type indicators. This baseline regression serves to
document the average change in crime rates in communi-
15. Average crime rates in Figures 1 and 2 are weighted by ties of each type (and also provides a test for the statistical
population to take into account differences in size across significance of these changes). We then add the changes in
geographic units. For example, more weight is placed on demographic variables to the specification.
New York City than Long Beach in tabulating the average
primary city crime rate. Appendix A and B present violent 22. The coefficient on the primary city declined from -799 to
and property crime rates overall and for primary cities and -575 after the controls were added.
suburbs in 1990 and 2008 for each of the metropolitan
areas analyzed in this study. 23. Recall, for areas that do not match to either a FIPS place
code or a census county subdivision code, we assign the
16. Worcester and Honolulu are not included in these summary average value for each demographic characteristic for the
counts because the city of Worcester did not report crime entire county.
data in 1990, and Honolulu does not have separate subur-
ban data in any year. 24. The indicator for non-place, non-MCD suburbs ensures that
the estimation results are not being driven by our imputa-
17. This change is statistically significant at the five percent tion of region-level values from the county average. All
level of confidence. regressions are weighted by community-level population.

18. This change is statistically significant at the one percent

level of confidence.

BROOKINGS | May 2011 21

 Acknowledgments
The authors thank Alan Berube, Meagan Cahill, P. Mitchell Downey, Joshua Markman,
Carey Nadeau, John Roman, and Martha Ross for their helpful comments on an earlier draft
of the report.

The Metropolitan Policy Program at Brookings thanks the Ford Foundation for its generous sup-
port of the program’s research on city and suburban poverty and opportunity, the Annie E. Casey
Foundation for its support of the program’s research on low-income working families, and the
John D. and Catherine T. MacArthur Foundation, the George Gund Foundation, the F.B. Heron
Foundation, and the Heinz Endowments for their general support of the program, as well as
members of the Metropolitan Leadership Council.

For More Information

Elizabeth Kneebone
Senior Research Associate
Metropolitan Policy Program at Brookings
ekneebone@brookings.edu
(202) 797-6108

Steven Raphael
Goldman School of Public Policy
University of California, Berkeley
stevenraphael@berkeley.edu
(510) 643-0536

For General Information

Metropolitan Policy Program at Brookings
(202) 797-6139
www.brookings.edu/metro

The Brookings Institution is a private non-profit organization. Its mission is to conduct high qual-
ity, independent research and, based on that research, to provide innovative, practical recommen-
dations for policymakers and the public. The conclusions and recommendations of any Brookings
publication are solely those of its author(s), and do not reflect the views of the Institution, its
management, or its other scholars.

Brookings recognizes that the value it provides to any supporter is in its absolute commitment to
quality, independence and impact. Activities supported by its donors reflect this commitment and
the analysis and recommendations are not determined by any donation.

22 BROOKINGS | May 2011

 About the Brookings Metropolitan
Opportunity Series
Launched in 2009, the Metropolitan Opportunity
Series documents the changing geography of
poverty and opportunity in metropolitan America,
analyzes its drivers and implications, and offers
policy recommendations to enhance the well-being
of lower-income families and communities in both
cities and suburbs. This study and other publications,
speeches, presentations, and commentary in the
series are available at: www.brookings.edu/metro/
Metropolitan-Opportunity.aspx

In the Series
• Missed Opportunity: Transit and Jobs in
Metropolitan America
• Job Sprawl and the Suburbanization of Poverty

Forthcoming
• Philanthropy’s Role in Building a Strong Regional
Safety Net
• Immigration and Shifting Metropolitan Poverty

Brookings 1775 Massachusetts Avenue, NW

Washington D.C. 20036-2188
telephone 202.797.6000
fax 202.797.6004
web site www.brookings.edu

telephone 202.797.6139
fax 202.797.2965
web site www.brookings.edu/metro