Abstract
Stark ethno-racial differences in reported neighborhood crime are a major facet of contemporary U.S. inequality. However, the most generalizable research on neighborhood inequality in crime across cities is only for 2000. Many of the underpinnings of crime have changed since 2000—increases in socioeconomic segregation, the Great Recession and attendant housing crisis, the continuation of the crime decline, shifting trends in incarceration and other types of social control, and small decreases in racial residential segregation. We provide a much-needed assessment of whether ethno-racial reported neighborhood crime disparities have increased, remained stable, or decreased in the contemporary period. We invoke a racial structural perspective that traces ethno-racial disparities in neighborhood crime to the divergent community conditions emblematic of the U.S. racial hierarchy. Using newly collected data for 8,557 neighborhoods in 71 large U.S. cities for 2010–2013, we demonstrate that violent and property crime is lower in White, African American, Latino, minority, and multiethnic neighborhoods than in 2000. However, smaller relative decreases in African American neighborhoods widened the relative crime gap from other ethno-racial communities. Supporting the racial structural perspective, large ethno-racial inequalities in neighborhood well-being account for most of the crime gaps, with disadvantage and residential lending being most important. This suggests that non-White neighborhoods need economic investments to reduce the harmful and inequitable consequences of neighborhood crime.
Disparities in neighborhood crime rates, particularly violent crime, are a major dimension of racial and ethnic stratification in the United States. The largest study of urban neighborhood crime to date for more than 8,000 neighborhoods in 87 large U.S. cities showed that average rates of violent crime reported to the police in 2000 were five times higher in predominantly African American than in White neighborhoods (Peterson and Krivo 2010). Latino, minority, and multiethnic neighborhoods had violence rates approximately two and a half to three and a half times those in White neighborhoods. 1
Peterson and Krivo (2010) argue that ethno-racial inequality in serious crime is a product of the racialized social structure of U.S. society rather than group differences in cultural orientations or individual criminal proclivities. As scholars contend, a racial hierarchy—fueled by discriminatory institutional practices and routine racialized interactions—characterizes the U.S. social order (see Bonilla-Silva 2001; Golash-Boza 2016). Within this structure, residential segregation and differential criminal justice practices are fundamental drivers of unequal crime levels across neighborhoods of different ethno-racial compositions. Whites commonly reside in segregated neighborhoods with more resources and opportunities, while African Americans and other non-Whites live in much more disadvantaged communities. African American and some other non-White neighborhoods also experience uniquely high levels of criminal justice contact including police surveillance, coercive social control, and incarceration. These divergent ethno-racial neighborhood environments set the stage for dramatic inequality in reported serious crime (Sampson and Wilson 1995; Sampson, Wilson, and Katz 2018; Shaw and McKay 1942). By emphasizing the root causes of racially structured inequalities, a racial structural perspective challenges accounts that pathologize Black and Brown communities as inherently or culturally prone to crime and violence.
Peterson and Krivo’s (2010) examination of neighborhood crime for 2000 provides a convincing case for a racial structural perspective. However, their study requires updating given dramatic shifts since 2000. The Great Recession and associated housing crisis destabilized local populations and housing markets with particularly deleterious consequences for communities of color (see Faber 2018; Woodstock Institute 2010). Reported crime also is lower nationally and for most cities and states than at the turn of the century (Baumer, Vélez, and Rosenfeld 2018). Further, mass incarceration and other formal social controls continued to grow though most of the 2000s, although imprisonment began declining around 2008 (Sentencing Project 2017). Finally, racial residential segregation continued to decline slowly while income segregation rose. Given these shifts, we assess whether inequality in reported violent and property crime between non-White and White neighborhoods in 2010–2013 persisted, expanded, or declined relative to 2000.
A lack of crime data for neighborhoods across multiple U.S. cities has stymied research on how ethno-racial inequality in serious crime changed after 2000. Police departments do not make data about known crimes publicly available for neighborhoods systematically. We thus engaged in an extensive collection of crime data from police departments in a sample of large cities. We use these newly collected reported crime data for 8,557 neighborhoods in 71 U.S. cities across the country for 2010–2013 to evaluate contemporary patterns of ethno-racial inequality in neighborhood crime. We minimize concerns about biases in officially reported data by focusing on serious violent and property crimes known to the police, which reflect realities of neighborhood crime much more reliably than reports for other offenses such as drug or traffic crimes (Beck and Blumstein 2018). 2 We explore how crime differentials relate to racialized patterns of socioeconomic disadvantage, residential instability, immigration, and housing investments. Although we cannot measure policing or other criminal justice practices, our racial structural perspective highlights the significance of racialized justice in reproducing criminal inequality. The unprecedented national scope of our data across large cities allows the broadest conclusions to date about ethno-racial differences in neighborhood crime in the new millennium.
Racial Structure and Inequality in Neighborhood Crime
Decades of neighborhood research show higher reported crime rates in non-White than in White neighborhoods (Peterson and Krivo 2010; Sampson and Wilson 1995; Shaw and McKay 1942). This observation dates back to Du Bois’s ([1899] 1973) work on Philadelphia and Shaw and McKay’s (1942) analyses of White and Black juvenile delinquency in early-nineteenth-century Chicago. Contemporary studies in a few cities also show that patterns of higher crime in African American and Latino than in White neighborhoods persist (e.g., Hipp 2007; Krivo and Peterson 1996). The National Neighborhood Crime Study (NNCS) demonstrates that such differences hold across a large set of U.S. cities and neighborhood types in 2000 (Peterson and Krivo 2010). Within the 87 NNCS cities, White communities averaged only about one fifth as much reported violent crime as African American neighborhoods and one third to one half as much violence as Latino, minority, and multiethnic neighborhoods.
Criminologists have long documented ethno-racial inequality in reported neighborhood crime while also questioning how much these inequalities reflect differences in offending or some combination of racialized enforcement (e.g., differential policing or profiling of non-White neighborhoods) and willingness to report crimes to the police. These dynamics certainly influence official police statistics. However, for serious reported violent and property crimes, such as homicide, robbery, and burglary, research demonstrates that the overrepresentation of Blacks in official statistics is not reducible to biases in police surveillance or citizens’ decisions to report crime to authorities (Beck and Blumstein 2018; Xie and Baumer 2019). Most officially counted crimes come from victims’ and witnesses’ reports, not from police finding or catching offenders. Further, serious offenses are less subject to reporting biases regardless of race and neighborhood characteristics (Baumer and Lauritsen 2010; Xie and Baumer 2019; Xie and Lauritsen 2012). Thus, although police data are social products, the weight of evidence maintains that reported serious crimes represent inequalities in the realities of crime across urban landscapes. 3
These large ethno-racial inequalities in reported neighborhood crime are rooted in racialized social structures that produce and maintain dramatically divergent social conditions in White and non-White areas (e.g., Peterson and Krivo 2010; Sampson et al. 2018; Shaw and McKay 1942). As Du Bois ([1899] 1973:241–42) noted more than 100 years ago, high crime in African American communities is “a symptom of countless wrong social conditions.” Shaw and McKay (1942) observed nearly 50 years later that highly racialized patterns of neighborhood social conditions underlie inequality in reported crime. Thus, theorists contend that uniquely high concentrations of disadvantage, considerable residential turnover, and a dearth of housing investments, along with disproportionate legal surveillance and incarceration, render non-White communities less able to control crime (Bursik and Grasmick 1993; Clear 2007; Sampson and Wilson 1995; Shaw and McKay 1942). Restricted access to elites and the attendant resources that make or break neighborhoods (Vélez and Lyons 2014) compound the vulnerability of non-White areas to criminal involvement. Ethnographers caution, however, that non-White neighborhoods that face racialized disadvantages do not lack social organization (e.g., Duck 2015; Jackson 2003; Pattillo 2007). Rather, structurally disadvantaged communities actively respond to the challenges emanating from the racial order. Duck (2015) shows that residents of structurally isolated African American communities create, maintain, and enforce a local social order to make life safer and more predictable. Pattillo (2007) demonstrates that residents on Chicago’s South Side engaged in concerted efforts to confront threats posed by gentrification, including managing ongoing struggles with crime and disorder. Nonetheless, some adaptations to structured exclusions can encourage violence in the absence of trustworthy law enforcement and/or contribute to cultural orientations that foster crime (e.g., Anderson 1999; Duck 2015; Rios 2011).
While the racially structured social hierarchy is multifaceted, ethno-racial socioeconomic inequality is a core element of this system. Compared to Whites, African Americans have the highest prevalence of poverty, joblessness, and low-status jobs as well as the least wealth and income (Bonilla-Silva 2001). Other ethno-racial groups such as Latinos and Asians generally fall between these extremes. Such inequalities are geographically concentrated in distinct neighborhoods by social and institutional mechanisms that create and maintain racial residential segregation (e.g., Massey and Denton 1993; Peterson and Krivo 2010; Quillian 2017). Historic and contemporary discriminatory policies and practices are primary forces fueling segregation. These include the building of segregated public housing and Federal Housing Administration and Veterans Administration lending criteria that supported White suburban development and redlined African American neighborhoods (Massey and Denton 1993; Rothstein 2017). White racism, discrimination by realtors, and racially disparate knowledge, experiences, and perceptions of communities further perpetuate segregation (e.g., Krysan and Crowder 2017).
Residential segregation is a linchpin for the reproduction of racial inequality because it ensures persistent differences across neighborhoods in disadvantages, investments, wealth, services, and social control (Liska and Yu 1992; Massey 2016). Segregation separates the disparate resources of African Americans and Whites into distinct neighborhoods (Massey and Denton 1993; Quillian 2017). Predominantly African American areas typically have the very highest levels of disadvantage present in this population; conditions in White neighborhoods reflect the resource-rich characteristics of Whites. Segregation of Latinos and Asians from Whites also concentrates their differential levels of disadvantage, but more modest segregation of these groups makes community socioeconomic inequality less extreme than between African Americans and Whites. Ethno-racially segregated spaces are also targets for differential (dis)investments by mortgage lenders and businesses that can stabilize or harm communities (Vélez and Lyons 2014).
Deep historical and contemporary processes of formal social control also permeate life within communities of color, particularly African American neighborhoods (e.g., Franklin 2018; Nunn 2002). Standing on the roots of slavery, Jim Crow laws, and many other now technically illegal practices, Black and Brown communities endure over- and underpolicing, wildly disproportionate stop-and-frisk rates, disrespectful and harassing police treatment, and greater use of force (Holmes, Painter, and Smith 2019; Rios 2011; Ward 2018). Some residents of non-White neighborhoods have such experiences, but many more hear about them from friends, family members, and neighbors and thus experience them vicariously (Brunson 2007). High levels of incarceration add to these everyday encounters and represent a form of “coercive mobility” that undermines community organization (Clear 2007). Together, this context of mistreatment strains relationships between communities and formal social control agents (Kirk and Matsuda 2011).
Highly inequitable levels of (dis)advantage, investments, and social control reinforce differential crime across ethno-racial communities. Peterson and Krivo (2010) provide the broadest support for the racial structural argument that varying levels of structural conditions, especially concentrated disadvantage, account for much of the crime disparities between distinct ethno-racial neighborhoods. Using data for neighborhoods in 87 large cities for 2000, they find dramatic ethno-racial segregation; 60 percent of Whites, 51 percent of African Americans, and 33 percent of Latinos lived in neighborhoods where most of their neighbors were of their same race-ethnicity. These segregated neighborhoods had extremely divergent levels of disadvantage; trivial proportions of African American, Latino, and minority neighborhoods had disadvantage levels as low as those in virtually all White neighborhoods. Peterson and Krivo (2010) also document that differences in disadvantage are the most important factor accounting for gaps in violent and property crime for African American, Latino, minority, and multiethnic compared to White neighborhoods. Other factors theorized to influence communities’ abilities to control crime—residential instability, which should increase crime (Shaw and McKay 1942), and immigration and home mortgage lending, which should decrease crime (Lyons, Vélez, Santoro 2013; Vélez, Lyons, Boursaw 2012)—have significant relationships with neighborhood violent and property crime. However, they make much smaller contributions than disadvantage to neighborhood crime gaps.
Although Peterson and Krivo (2010) were unable to test the contribution of legal social control—an important component of the racial order—to ethno-racial criminal inequality, other studies demonstrate its significance for crime. For example, incarceration is concentrated in impoverished communities of color where it exacerbates crime (Clear 2007). Profound distrust in legal authorities in non-White neighborhoods also undermines their ability to keep crime at bay (Anderson 1999; Kirk and Papachristos 2011). As such, differential formal social control is clearly part of the context that contributes to ethno-racial inequality in neighborhood crime.
Since 2000, the United States has experienced significant changes that shape neighborhood well-being: greater residential segregation by income, particularly among African Americans and Latinos; the Great Recession and housing crisis; lower violent and property crime; and lower Black-White residential segregation. These dynamics necessitate an updated assessment of the patterns and sources of ethno-racial neighborhood inequality in crime, particularly because the changing conditions have contradictory implications for criminal inequality.
On the one hand, greater income segregation, greater housing instability, and less prime mortgage lending after 2000 could have reinforced or exacerbated neighborhood ethno-racial gaps. Residential segregation by income grew after 1970, with greater increases among Black and Hispanic than White families (Bischoff and Reardon 2014). Segregation of both poor and affluent families rose during this period, resulting in “greater polarization of neighborhoods by income” (Bischoff and Reardon 2014:226). More spatial isolation of economically disadvantaged from advantaged families means that a major source of crime differentiates neighborhoods more sharply. Larger increases in Black than White income segregation also intensified Black poverty concentration, which should reinforce higher crime in African American neighborhoods.
In the first decade of the 2000s, housing instability also spiked across the United States as the housing crisis hit. The impact of this upheaval was greatest in Black and Latino neighborhoods because they were disproportionately targeted for predatory and subprime housing loans (Hall, Crowder, and Spring 2015; Hwang, Hankinson, and Brown 2015; Rugh and Massey 2010; Woodstock Institute 2010). As a result, foreclosure rates in Black and Hispanic neighborhoods were, respectively, 2.6 and 2.3 times those in White areas from 2005 to 2012 (Hall et al. 2015). Home lending investments diminished in non-White neighborhoods as a fallout from the foreclosure crisis. For example, during a time of recovery (2014), lenders were more likely to deny conventional loans to Black and Latino compared to White applicants and to steer them toward subprime loans (Faber 2018). These racialized outcomes were most acute for applicants in segregated neighborhoods (Been, Ellen, and Madar 2009). Thus, more extensive housing instabilities in communities of color fueled socioeconomic marginalization in non-White areas, potentially yielding persistent or even greater ethno-racial inequality in neighborhood crime in 2010–2013 than in 2000.
On the other hand, lower crime and slow decreases in segregation may have ameliorated the racial-spatial crime divide. The U.S. violent crime rate fell by 27 percent between 2000 and 2013 (our data’s final year); property crime fell by 24 percent (Federal Bureau of Investigation 2020). If high-crime minority neighborhoods saw larger decreases than other areas, then gaps should narrow (Friedson and Sharkey 2015). Sharkey (2018) argues that communities of color that had particularly high crime in 2000 benefited the most from the crime decline because they had the greatest potential for decreases. Lower rates in White neighborhoods limited possibilities for equally large drops, which could curtail inequality between White and non-White areas. Slow declines in Black-White residential segregation also could reduce crime gaps (Rugh and Massey 2014). If segregation differentially concentrates disadvantage/advantage, decreases should reduce ethno-racial gaps in socioeconomic sources of local crime, and hence in neighborhood crime, at least between African Americans and other groups.
Only three studies examined inequality in neighborhood crime after 2000, and they yielded conflicting results. Friedson and Sharkey (2015) revealed less racial and economic inequality in neighborhood exposure to violence around 2010 than in the prior 10 to 20 years across six U.S. cities; violence rates for poor and non-White neighborhoods became more similar to those for White and nonpoor areas. In contrast, Papachristos, Brazil, and Cheng (2018) found greater socioeconomic inequality in homicides in 2009 than in 1990 across Chicago communities because killings declined most in areas with initially lower levels of homicide and disadvantage. Finally, Krivo et al. (2018) showed that most neighborhoods in 18 studied cities, irrespective of race-ethnic composition, saw declining homicide and burglary rates from 1999 to 2013. However, 11 percent of neighborhoods had stable homicide rates, and 5 percent experienced increasing homicides; almost all neighborhoods in these two groups were predominantly Black. Nearly one-third of neighborhoods in the 18 cities showed increases in burglaries; 45 percent of such areas were predominantly Black. 4 These patterns of increasing crime were unique to Black areas, leading to greater crime inequalities only for Black relative to other neighborhoods.
Wide variation in the social and economic character of cities makes it difficult to extrapolate from just three studies of neighborhoods in a few cities. Guided by a racial structural framework, we assess the scope and sources of ethno-racial inequality in neighborhood crime for a more recent time point across a representative sample of cities. We use newly collected data on reported crime and structural conditions for 8,557 neighborhoods in 71 cities. Thus, we provide the most generalizable evidence available about whether ethno-racial inequality in neighborhood reported crime persisted, increased, or declined after 2000.
Data And Methods
Data and Sample
We use data for census tracts in 71 U.S. cities with populations greater than 100,000 (in 2000) from the second wave of NNCS (NNCS2) (N = 8,557 census tracts). Although census tract boundaries are not always identical to residents’ notions of neighborhoods, they are widely used neighborhood proxies. We obtained data on serious crimes known to police from city police departments. Sociodemographic data come from 2008–2012 American Community Survey (ACS) and 2010 Home Mortgage Disclosure Act data. NNCS2 is the only data set with crime and other data for neighborhoods in a representative set of large U.S. cities. The breadth of coverage, including central cities and suburbs across all regions, makes it uniquely suited for assessing neighborhood inequality in crime after 2000. The 71 cities are a subset of the representative sample of 87 cities that provided comparable violent (homicides and robberies) and property (burglaries, larcenies, and motor vehicle thefts) crime data in wave 1 of NNCS (response rate of 82 percent). The cities analyzed are similar in overall crime, poverty, ethno-racial composition, and regional location to all U.S. cities with populations greater than 100,000 in 2010 but are more racially segregated (statistics available upon request).
Dependent Variables
We collected data on reported homicides, robberies, burglaries, larcenies, and motor vehicle thefts for 2010–2013. For 14 cities, we obtained crime counts for census tracts. In 57 cities, departments provided incident data that we geocoded and aggregated to census tract counts (average geocoding hit rate = 98.2 percent). Four-year average counts of violent (homicides and robberies) and property (burglaries, larcenies, and motor vehicle thefts) crime are the dependent variables (see Table 1 for all operationalizations). 5 We use multiple-year counts to minimize the effect of annual fluctuations for small units.
Operationalization of neighborhood (census tract) variables.
Note: nghd = neighborhood.
Independent Variables
We divide neighborhoods into five ethno-racial types that reflect differential segregation (Peterson and Krivo 2010). White (n = 2,298), African American (n = 1,224), and Latino (n = 824) neighborhoods each have at least 70 percent of the tract population in the respective group. Minority neighborhoods (n = 730) are at least 70 percent African American and Latino, but neither group alone reaches the 70 percent threshold. All other neighborhoods are multiethnic (n = 3,481) and have varying mixes of ethno-racial groups. Since the overall population of the cities examined is only 43 percent White, 21 percent African American, and 26 percent Latino, neighborhoods with 70 percent or more in one group are highly segregated. These neighborhood types that are based on the proportional representation of groups do not necessarily reflect differences in interpersonal and structural relations among neighborhoods’ residents (Mayorga- Gallo 2014). However, we rely on them because they reflect racial stratification in resources, (dis)investments, and institutional responses.
We account for four measures of the racial hierarchy associated with criminal inequality. First, disadvantage is an index of average z-scores for joblessness, professional workers (reverse coded), college graduates (reverse coded), female-headed families, secondary-sector workers, and poverty. Second, residential instability is an index (average z-scores) of the percentage of renters (vs. homeowners) and the percentage who lived in a different residence one year ago. Third, an immigration index averages z-scores for the percentages of foreign born, recent immigrants, and linguistically isolated households. Fourth, we measure residential loans, a primary way to secure wealth, with the dollar amount of home loans issued in 2010. We also control for the percentage of the population that is young and male.
Analytic Strategy
We begin by comparing violent and property crime rates between ethno-racial neighborhood types for circa 2010 to those from 2000 (from Peterson and Krivo 2010). We then describe differences in ethno-racial neighborhood segregation, concentration of disadvantage, residential instability, immigration, and residential loans for 2010. 6 These descriptive portraits provide the backdrop for understanding the sources of recent neighborhood criminal inequality. Finally, we fit negative binomial count models predicting violent and property crime to assess the sources of ethno-racial differences in crime. We estimate nonlinear negative binomial count models because crimes are relatively rare events within census tracts. These models account for overdispersion in which the variance of the dependent variable is considerably larger than the mean, a common concern with count data. We specify that crime counts (rounded to integers) have variable exposure by tract population, making the analysis one of rates per tract population. We use clustered standard errors by city and include city fixed effects to control for all unmeasured city differences.
We estimate a series of models separately for violent and property crime. The baseline model includes the ethno-racial neighborhood types and the percentage of young males. We then add to the baseline model individual neighborhood conditions—disadvantage, residential instability, immigration, and residential loans—one at a time to evaluate how each alone accounts for changes in the neighborhood-type coefficients. Our final full model includes all variables. We assess the statistical significance of changes across models in the coefficients for ethno-racial neighborhood types using a general framework for comparing marginal effects (Mize, Doan, and Long 2019). 7 We conduct analyses with Stata version 15 using the nbreg procedure.
Findings
Inequality in Neighborhood Violent Crime: 2000 and 2010
Figure 1 presents average rates and distributions of reported violent crime (per 1,000 population) by ethno-racial neighborhood type for 2000 and 2010. Violent crime is, on average, lower in 2010 than in 2000 for all neighborhood types. The average White neighborhood had 2.0 violent crimes per 1,000 population in 2000 and 1.4 violent crimes in 2010 (a 30 percent decrease). Average violence in minority neighborhoods also declined by 30 percent (from 7.1 to 5.0), while average rates dropped by about 40 percent in Latino and multiethnic neighborhoods (from about 4.9 to 2.9). In contrast, the relative decline in mean violent crime was only half as much for African American neighborhoods with a decrease of only 15 percent, from 10.0 to 8.5. Of note, absolute declines in average violence rates for African American (1.5 fewer violent crimes per 1,000 population) and the other three non-White neighborhoods (around 2.0 fewer crimes) were greater than for White areas where violence was already much lower in 2000 (0.6 fewer crimes in 2010). These absolute decreases in violence in non-White neighborhoods are clearly substantively important for residents of these communities. Nonetheless, given the smaller percentage decline from very high rates, African American areas continued to have the highest levels of violence with an average rate that is a full six times that for White neighborhoods, rather than five times the mean White rate in 2000. The relative ranking of neighborhood violence by ethno-racial composition also remained the same in 2010 as in 2000. White neighborhoods have the lowest average rates, followed by Latino and multiethnic areas with mean rates about twice those in White areas. Average violence is higher yet in minority neighborhoods but still slightly more than 40 percent less than that in African American areas.

Violent crime rates by ethno-racial neighborhood type, National Neighborhood Crime Study wave 1 (2000) and wave 2 (2010).
Despite large inequalities, the distributions of reported violence indicate important improvements (see Figure 1). All neighborhood types have smaller gaps between the 10th and 90th percentiles in 2010 than a decade earlier, due mainly to notable declines in the 90th percentiles. Very high levels as well as the spread in violent crime decreased for all ethno-racial neighborhoods. Yet marked inequality remains, as seen in the lack of overlap in violence levels across types of areas, particularly between White and African American neighborhoods. About 30 percent of Latino and multiethnic neighborhoods have violence rates greater than those in 90 percent of White communities. Violence is higher in more than half of minority neighborhoods than in almost all White areas. Most striking is that rates of violent crime for African American neighborhoods rarely overlap with levels of violence in White areas. In 2010, 90 percent of African American neighborhoods had higher violent crime rates than nearly all White neighborhoods—a pattern that is unchanged from 2000.
Neighborhood Differences in Ethno-Racial Segregation, Disadvantage, and Other Conditions
Our data highlight the persistence of neighborhood segregation in large U.S. cities (see Figure 2). In 2010, nearly half of White and African American individuals in our sample cities, and slightly more than one third of Latinos, lived in neighborhoods where most other residents (70 percent or more) were of their same race-ethnicity. Only small portions of Whites (4 percent), African Americans (9 percent), and Latinos (8 percent) lived in areas where either of the other two groups alone exceed 70 percent of the population. Nonetheless, many Whites, African Americans, and Latinos also reside in multiethnic neighborhoods (45 percent, 31 percent, and 40 percent, respectively) given the diversity of contemporary U.S. cities.

Whites, African Americans, and Latinos in five ethno-racial neighborhood types in 2010.
A key consequence of racial residential segregation is inequality in neighborhood disadvantage. Figure 3 presents the distributions of neighborhood disadvantage for White compared to African American and Latino neighborhoods. These results show the persistent neighborhood divide in socioeconomic conditions (Peterson and Krivo 2010). Disadvantage is below average for most White neighborhoods and is concentrated at very low levels. Nearly half of White neighborhoods have the lowest (27.5 percent) or the next lowest (22.3 percent) levels of disadvantage across all neighborhoods in the 71 cities. The contrast with African American neighborhoods is glaring. Almost all African American neighborhoods have above average disadvantage (97 percent), and close to one third have the highest levels among more than 8,500 neighborhoods. Most Latino neighborhoods also are highly disadvantaged—98 percent have above average levels. However, disadvantage in Latino neighborhoods is less extreme than in African American areas, with more than half of Latino neighborhoods having high but not the highest disadvantage levels (scores around .64 to .91).

Disadvantage distributions for White, African American, and Latino neighborhoods in 2010.
Comparing disadvantage in White neighborhoods to that in minority and multiethnic areas uncovers how different multiethnic neighborhoods are (see Figure 4). Large shares of multiethnic areas have disadvantage levels at or near the average; a notable share has very low disadvantage, and fewer have very high disadvantage. Minority neighborhoods are more similar to Latino and African American areas; almost 95 percent have above average disadvantage. Thus, the large group of multiethnic neighborhoods have less favorable socioeconomic conditions than White communities but are far better off than neighborhoods where African Americans and Latinos predominate alone or together.

Disadvantage distributions for White, minority, and multiethnic neighborhoods in 2010.
Table 2 shows that disadvantage is not the only crime-producing condition that varies by neighborhood type. White communities also are particularly privileged in having the lowest residential instability (due largely to low levels of renting) and the highest average home loans. The average home loan in White communities is 18 times that in African American areas, 7 times that in Latino neighborhoods, and 6 times that in minority areas, signaling the greater potential for wealth accumulation for White areas. Immigrant concentration, which often protects against crime, is the highest in Latino neighborhoods and lowest in White and African American areas.
Means of residential instability, immigration, and residential loans by ethno-racial neighborhood type, second wave of National Neighborhood Crime Study (N = 8,557).
Note: nghd = neighborhood.
Values are in constant 2000 dollars.
Accounting for Ethno-Racial Inequality in Neighborhood Violent Crime
Do differences in disadvantage and other structural conditions account for ethno-racial inequality in neighborhood violent crime for 2010–2013? Table 3 presents differences in violent crime in African American, Latino, minority, and multiethnic compared to White neighborhoods net of controls and neighborhood structural characteristics. Incidence risk ratios (IRRs) indicate the ratio of average neighborhood violent crime for each ethno-racial neighborhood type compared to that for White neighborhoods; for all other variables, IRRs represent differences in violence for a one-standard-deviation unit difference in the characteristic.
Negative binomial model of neighborhood violent crime with neighborhood characteristics: National Neighborhood Crime Study, wave 2 (2010) (N = 8,557).
Note: All models include dummy variables for city. IRR = incidence risk ratio; nghd = neighborhood.
For the neighborhood-type dummy variables, IRRs represent the ratio of average neighborhood violent crime for the ethno-racial neighborhood type compared to the average for White neighborhoods. For all other variables, IRRs represent differences in violent crime for a one-standard-deviation unit change in the characteristic.
p < .05. **p < .01 (two-tailed tests).
Violent crime in African American neighborhoods is more than five times that in White neighborhoods net of control variables (see Table 3, model 1). The other three neighborhood types also have significantly higher violent crime than White areas; violence ratios range from 2.1 for multiethnic versus White communities to 3.7 for the minority-White gap. These inequalities are not due to the greater concentrations of non-White than White neighborhoods in more segregated, disadvantaged, higher crime, and otherwise dissimilar cities because we adjust for unmeasured differences across urban places with city fixed effects.
How much does each neighborhood condition tapping into the racial structural perspective account for differences in violent crime? Comparing the results for neighborhood types in model 1 to those from models 2 through 5 (which add structural conditions to the baseline one at a time) highlights the prominence of disadvantage and residential loans as individual sources of ethno-racial inequality in neighborhood violence in 2010. After accounting for disadvantage alone, the relative risk of violent crime in African American compared to White neighborhoods is 59 percent less than before this factor was controlled (IRR of 2.2 versus 5.2). Neighborhood disadvantage accounts for a similar percentage of Latino-White and minority-White neighborhood violence differentials (ratios are about 56 percent lower in model 2 than in model 1), but a smaller share of multiethnic-White gaps (a 32 percent decrease). 8 Individually, home lending is second in importance as a source of inequalities in violence (model 5 versus model 1). 9 If African American, Latino, and minority neighborhoods had similar amounts of loan investments as White areas, violent crime gaps would be 30 to 36 percent less than in the baseline model; the multiethnic-White net violence ratio would be 15 percent lower. The individual contributions of residential instability and immigration to ethno-racial violent crime differentials are generally much smaller.
The final model controlling for the structural characteristics simultaneously shows that all are significantly associated with neighborhood violence (model 6). More disadvantaged and residentially unstable areas have more violence. Neighborhoods with more immigration and residential loans have less violence. Disadvantage and residential instability are the strongest predictors of overall variation in violence; they have larger IRRs than other predictors. One-standard-deviation-higher levels of disadvantage and residential instability are associated with 37 percent and 41 percent more violence, respectively. Immigration and residential loan levels that are one standard deviation higher are related to just 7 percent and 9 percent less violence.
To assess which characteristics are most important in accounting for inequality in violent crime, we compare results for neighborhood types when each structural predictor is considered separately (models 2–5) with those from the full model. These comparisons highlight the centrality of neighborhood disadvantage as a source of ethno-racial differentials in violent crime. Recall that disadvantage alone reduces the relative risk of African American–White violence by 59 percent (from 5.24 in model 1 to 2.17 in model 2). The ratio of violence in African American versus White neighborhoods declines by just 9 percent more when residential instability, immigration, and residential loans are added to analyses with just disadvantage (from 2.17 in model 2 to 1.97 in model 6). Minority-White and multiethnic-White violence differentials change little when all factors are considered compared to when only disadvantage is controlled (differences in these two contrasts between models 2 and 6 are not significant). Other aspects of the racialized social structure add little over and above disadvantage to an account of criminal inequality. Latino neighborhoods are different since disadvantage alone fully explains their higher violent crime than crime in White areas (comparing models 1 and 2), but when all factors combine, Latino neighborhoods have more violence than do White areas.
Overall, these results demonstrate that race-based differences in community structural conditions, most importantly socioeconomic disadvantage and residential loans, account for a large portion of the differences in violence across ethno-racial neighborhood types for 2010, similar to a decade earlier (Peterson and Krivo 2010). African American neighborhoods have just less than 2.0 times as much violent crime as White neighborhoods when disadvantage, residential instability, immigration, and residential loans are controlled. Latino and multiethnic neighborhoods have violence rates that are about one third higher than rates in comparable White areas. Relative risks of violent crime in minority neighborhoods are 1.7 times those for White neighborhoods. These inequalities are significantly smaller than in the baseline, providing broad support for a racial structural understanding of the sources of ethno-racial inequality in neighborhood violence in the contemporary United States.
Inequality in Neighborhood Property Crime
Figure 5 presents average rates and distributions of property crime (per 1,000 population) by ethno-racial neighborhood type for 2000 and 2010. Exposure to property crime is much greater than exposure to violent crime. However, in 2000, ethno-racial neighborhood inequality was also far less for property than for violent crime (Peterson and Krivo 2010). Given social changes after 2000, how did property crime differentials change? Mean property crime rates are between 30 and 40 percent lower in 2010 than in 2000 for all neighborhood types except African American areas, where average rates declined by only 21 percent. Therefore, the gap in average property crime between African American and all other neighborhoods increased. This pattern is balanced by substantial decreases in variation within every neighborhood type, including African American areas. The gaps between the 10th and 90th percentiles of property crime rates are notably smaller in 2010 than in 2000, due largely to substantial declines in the 90th percentile for all types of areas. The highest rates of property crime have come down, and variation is markedly less than it was a decade earlier, as is the case for violence. Thus, the large overlap in property crime across ethno-racial neighborhoods is similar in 2010 to 10 years earlier, albeit at lower levels.

Property crime rates by ethno-racial neighborhood type, National Neighborhood Crime Study wave 1 (2000) and wave 2 (2010).
Results from the multivariate models of property crime further highlight the limited inequality in this type of crime (see Table 4). Property crime rates are 65 percent higher in African American than in comparable White neighborhoods when only young males and all city differences are controlled (model 1). This gap is reduced to 19 percent when all structural conditions are accounted for (model 6). Higher property crime in minority and multiethnic than in White neighborhoods is modest in the baseline model (a 32 percent and 24 percent differential, respectively). Differences for minority and multiethnic compared to White areas after all characteristics are controlled are significant but small (18 percent and 16 percent for minority and multiethnic vs. White areas, respectively). Average property crime rates are not significantly different in White and Latino neighborhoods in any model.
Negative binomial model of neighborhood property crime with neighborhood characteristics, National Neighborhood Crime Study wave 2 (2010) (N = 8,557).
Note: All models include dummy variables for city. IRR = incidence risk ratio; nghd = neighborhood.
For the neighborhood-type dummy variables, IRRs represent the ratio of average neighborhood property crime for the ethno-racial neighborhood type compared to the average for White neighborhoods. For all other variables, IRRs represent differences in property crime for a one-standard-deviation unit change in the characteristic.
p < .05. **p < .01 (two-tailed tests).
The relative risks for all four non-White compared to White neighborhoods decline the most when disadvantage and residential loans are added individually to the baseline (models 2 and 5 vs. model 1). However, residential lending, but not disadvantage, is a significant net predictor of property crime in the final analysis (model 6). An additional model in which disadvantage and residential loans are the only characteristics included (not reported) showed that disadvantage became nonsignificant after controlling for home lending. This suggests that residential lending was the most consequential source of property crime inequality after the housing market crash.
Conclusion
In the years since Peterson and Krivo’s (2010) 2000 portrait of ethno-racial inequality in neighborhood crime, the United States has experienced notable social and economic changes: an economic recession and housing crisis, expansion of incarceration and other forms of legal social control, declining racial and increasing income segregation, and a continued decline in serious crime. These shifts necessitate a reexamination of the degree and sources of racial inequality in neighborhood crime for a more recent period. We marshal newly collected evidence to take stock of contemporary levels and ethno-racial inequalities in crime for neighborhoods in 71 large cities across the United States. We find that neighborhoods across the ethno-racial spectrum—White, African American, Latino, minority, and multiethnic—are all safer in 2010–2013 than in 2000. We also find that extreme rates of neighborhood violent and property offending have essentially disappeared across the board. Thus, we show for the first time that the “Great American Crime Drop” extends across the United States to neighborhoods—the sites where residents experience crime most directly.
Reductions in reported crime have advanced safety and quality of life in meaningful ways for residents of all neighborhood types (Sharkey 2018). Most noteworthy are the substantial absolute reductions in crime in African American neighborhoods, the areas with the highest reported crime levels. On average, African American neighborhoods have 1.5 and 16 fewer violent and property crimes, respectively, per 1,000 population than in 2000.
Despite improvements in safety, neighborhood ethno-racial inequality in crime remains a central feature of U.S. cities, particularly for violence. Inequality is most stark for African American neighborhoods because of smaller relative decreases in violent and property crime than for all other neighborhoods. Violence declined by 30 percent in White communities but by only 15 percent in African American neighborhoods. In contrast, violence decreased even more in Latino and multiethnic neighborhoods (by more than 40 percent) than in White communities. As a result, relative crime gaps between African American and White, as well as other ethno-racial neighborhood types, are even greater today than in 2000. This widening of African American versus White and other non-White inequality provides a sobering corrective to the widely heralded crime decline. It also is strikingly consistent with claims that a Black–non-Black divide structures the contemporary United States (e.g., Bean, Lee, and Bachmeier 2013).
We argue that these persistent ethno-racial inequities in serious crime are products of the racial hierarchy that organizes U.S. society. This approach treats the structural underpinnings of the neighborhood criminal divide as there by design, powered by historical and contemporary racist beliefs, policies, and practices. We focus empirically on the geographic concentration of four markers of the racial hierarchy upheld by residential segregation—socioeconomic disadvantage, residential instability, immigration, and residential lending. We find that White areas continue to have much lower levels of disadvantage and instability, and more substantial housing investments, than do African American neighborhoods. Latino, minority, and multiethnic neighborhoods fall in between. These structural inequalities undergird ethno-racial divisions in crime, and adjusting for them dramatically reduces crime gaps between each type of non-White neighborhood and White areas.
However, these four markers of the racial order do not contribute equally to crime differentials. Consistent with a large literature on crime, disadvantage is most central to criminal inequalities (Pratt and Cullen 2005), especially for violence. Nevertheless, home mortgage lending is an increasingly important source of the ethno-racial patterning of crime. Home loans enhance wealth within communities, fortify efforts to stymie disorder and decline, and shore up the ability of neighborhoods to reduce crime (Vélez and Lyons 2014). However, the housing and foreclosure crisis, which hit non-White communities hardest, exacerbated historical racial inequality in neighborhood lending by undermining the willingness of banks to extend mortgage credit to “risky” communities (Faber 2018; Woodstock Institute 2010). Our data reflect this pattern for 2010, with White neighborhoods’ receiving $20 million to $40 million more loans on average than any other neighborhood type. These gaps are more than double those in the 2000 NNCS (Peterson and Krivo 2010) due largely to lesser lending in non-White neighborhoods and large lending increases in White and multiethnic areas. Such changes, along with the greater impact of mortgage lending on neighborhood crime inequality, suggest a need to look beyond the vulnerabilities of non-White communities brought about by economic recessions. We must additionally address practices that exclude minority borrowers and their communities from recovery efforts designed to rebuild neighborhoods and stonewall crime.
Our results also reinforce findings that immigration staves off crime (Lyons et al. 2013), contrary to rhetoric painting immigrants as criminogenic. Given the breadth of our data, there is even more reason to question political efforts to criminalize immigrants, withdraw funds from “immigrant-friendly” cities, utilize the 2020 COVID-19 pandemic to fuel antiimmigrant policies and practices, or build a wall on the U.S. southern border to help make residents safer from crime. It is time for politicians and the public to recognize that immigration does not increase crime but rather shores up communities and keeps crime at bay.
Our analyses still do not tell the full story of the sources of neighborhood ethno-racial gaps in reported crime. Differences in crime across neighborhood types persist when we take into account aspects of the racial social order. The challenge is to continue to elucidate the structural products of discriminatory institutional practices and routine racialized interactions that maintain the U.S. racial hierarchy and reproduce criminal inequality. We focused on disadvantage and housing investments, along with residential instability and immigration. However, the racial-spatial divide also exists in exposure to formal social control (e.g., incarceration and police surveillance), access to political power, social capital, economic (dis)investments beyond mortgage lending, and local services. Neighborhood data on types of legal social control such as incarceration, police response times and demeanor, levels of stop and frisk, and community policing are particularly critical aspects of racialized structures for which information is lacking across cities. While many of these vulnerabilities correlate with socioeconomic disadvantage, data to evaluate their independent roles would enhance insights about how racial structures produce differential community crime rates. In addition, ethno-racial structures that undergird crime are unequal not only within neighborhoods but also by the character of nearby communities. Non-White neighborhoods are much more likely than White areas to be close to criminogenic structural conditions further enhancing differential crime. Finally, specification of the mechanisms that contribute to racial inequality in neighborhood crime will advance a richer racial structural perspective. We also call for additional ethnographic work on neighborhoods, which arguably is better suited than large-scale quantitative studies to uncover the localized adaptions and processes linking structural realities to organization, social control, and crime (e.g., Anderson 1999; Duck 2015; Mayorga-Gallo 2014).
Our discoveries are of broad importance because ethno-racial inequality in crime can exacerbate other inequalities. Exposure to crime exacts direct costs for neighborhood and individual health and well-being (Boggess, Greenbaum, and Tita 2013; Burdick-Will 2016; Sharkey 2018). It also dampens economic prospects for youth who grow up in areas with more violence, thereby reproducing societal inequality (Chetty and Hendren 2018; Sharkey and Torrats-Espinosa 2017). Thus, without sustained commitments to invest in non-White communities, the vast racial-spatial divide in exposure to crime will mete out penalties for individuals and communities and remain a legacy for future generations.
Footnotes
Acknowledgements
We thank Long Doan and Robert Kaufman for their consultation and advice. We also thank Elizabeth Sabbath and Jason Phillips for their research assistance in compiling the data.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by National Science Foundation grants SES-1357207 and SES-1356252.
