Spatial Foundations of Inequality: a Conceptual Model and Empirical Overview.” RSF: the Russell Sage Foundation Journal of the Social Sciences 3(2): 1–33

Spatial Foundations of an em pirical overview Inequality: A Conceptual Model and Empirical Overview George Galster and Patrick S harkey

Inequalities among individuals and house- causal estimates of the impact of space on in- holds in achieved socioeconomic status (individual socioeconomic outcomes. We do not come, wealth, and so on) in the United States advance new empirical research but rather a have reached levels not observed for almost a framing of the issues that will be sufficiently century. We believe that a corresponding evo- robust and comprehensive to permit integra- lution of geographic inequalities in socioeco- tion of all the papers in this issue as illustra- nomic, environmental, institutional, and po- tions and amplifications. Moreover, we hope litical domains both reflect and—more that our model will be useful in generating hy- importantly from our perspective—contribute potheses and thereby stimulating further re- to these inequalities across individuals.1 It is search on this crucial topic. this belief that motivated the Russell Sage In overview, our conceptual model con- Foundation to devote an issue of RSF: The Rus- tends that the variations in geographic context sell Sage Foundation Journal of the Social Sci- across multiple scales (neighborhood, jurisdic- ences to this topic. tion, metropolitan region)—what we call spa- Our primary purpose in this introductory tial opportunity structure—affects the socioeco- article is to develop a holistic, multilevel con- nomic outcomes that individuals can achieve ceptual model for comprehending how space in two ways by altering the payoffs that will be can be considered a foundation of U.S. socio- gained from the attributes individuals have economic inequality. Our secondary aim is to during any given period and the bundle of at- provide a synthetic review of the evidence on tributes that individuals will acquire (both pas- various dimensions of inequality of opportu- sively and actively) during their lifetimes. nity and outcomes in America, and the empir- In the first mechanism, the spatial oppor- ical scholarly literature that provides plausibly tunity structure serves as a mediating factor,

George Galster is Clarence Hilberry Professor of Urban Affairs at Wayne State University.Patrick Sharkey is professor of sociology at New York University.

© 2017 Russell Sage Foundation. Galster, George, and Patrick Sharkey. 2017. “Spatial Foundations of Inequality: A Conceptual Model and Empirical Overview.” RSF: The Russell Sage Foundation Journal of the Social Sciences 3(2): 1–33. DOI: 10.7758/RSF.2017.3.2.01. The authors thank Maren Toft for her comments on an earlier draft and Sylvia Tatman-Burruss for research assistance. Participants at the Russell Sage Foundation conference “Spatial Foundations of Inequality” and two anonymous reviewers provided valuable feedback on early drafts of this article. Direct correspondence to: George Galster at [email protected], Department of Urban Stud- ies and Planning, Room 3198, Faculty-Administration Building, Wayne State University, Detroit, MI 48202; and Patrick Sharkey at [email protected], Department of Sociology, New York University, 295 Lafayette St., Room 4102, New York, NY 10012.

1. We are not the first to make such claims, of course. A similar theme is advanced in numerous seminal works (for example, Wilson 1987; Jencks and Mayer 1990; Brooks-Gunn et al. 1993; Briggs 1995; Brooks-Gunn, Duncan, and Aber 1997). 2 spatial foundations of inequality translating a given bundle of individual attri- tion across space of different segments of the butes into achieved status depending on geog- population as classified by racial-ethnic back- raphy of the individual’s residence, work, and ground and income, economic opportunities, routine activity spaces. In the second mecha- environmental hazards, and violence. nism, the spatial opportunity structure serves as a modifying factor affecting the bundle of Segregation by Economic Status and attributes that individuals develop over time Race-Ethnicity in three ways. First, it directly influences the Trends in household income and wealth in- attributes over which individuals may exercise equality, which have been well documented in little or no volition, such as exposure to envi- academic work and the popular press, are mir- ronmental pollutants or violence. Second, it rored by trends in the degree to which low- and directly influences the attributes over which high-income families live apart from each they exercise considerable volition by shaping other, as measured by economic segregation what they perceive is the most desirable, fea- (Bischoff and Reardon 2014; Jargowsky 1996, sible option. It does so by influencing what in- 2003, 2015; Reardon and Bischoff 2011, 2016; formation about the individual’s options is Watson 2009). No matter how economic segre- provided, what the information objectively in- gation is measured, trends show steady growth dicates about payoffs from these options, and in the degree of segregation by income since how the individual subjectively evaluates the the 1970s. Sean Reardon and Kendra Bischoff information. These decisions early in life lead (2016) use a straightforward measure of the people into various path-dependent trajecto- proportion of families living in neighborhoods ries of achieved socioeconomic status and sub- that have median income at least 50 percent sequent life decisions, in cumulatively rein- above or 50 percent below the metropolitan forcing processes that can stretch across area median to document changes in the de- lifetimes and generations. Third, in the case of gree to which American families have begun children and youth, the spatial opportunity to sort into separate communities stratified by structure indirectly influences their attributes economic status. They find that, in 1970, about through induced changes in the resources, be- 15 percent lived in neighborhoods that were haviors, and attitudes of their caregivers. either extremely affluent or extremely poor. By 2012, that figure had risen to 34 percent. Inequalities in the Spatial Although the growth of affluent neighbor- Opportunity Structure hoods is an important contributor to the rise The first basic claim that motivates this issue of economic segregation (Reardon and Bischoff is that various dimensions of inequality are or- 2011), much of the concern about the issue ganized in space. The spatial organization of stems from the long-term rise of concentrated inequality is, in part, simply a manifestation poverty. Paul Jargowsky (2003, 2015) docu- of inequality occurring at the level of individu- ments trends in the proportion of all Ameri- als, families, and groups that is mapped on to cans and poor Americans living in neighbor- spaces. However, spatial inequality also is due hoods with a poverty rate of 40 percent or to intentional efforts to organize physical greater in a series of reports, showing substan- space in ways that maintain or reinforce in- tial growth in concentrated poverty from 1970 equality (Dreier, Mollenkopf, and Swanstrom to 1990, a decline of high-poverty neighbor- 2001). As a result of both sets of processes, vari- hoods in the 1990s, and a subsequent increase ation is tremendous in economic status, labor in concentrated poverty from 2000 to the most market opportunities, core institutions such recent years in which data were available (from as schools, environmental hazards, and social 2009 through 2013). Since 2000, the number of networks across city blocks, neighborhoods, extreme poverty neighborhoods has risen by cities and towns, metropolitan areas, and re- more than 75 percent and the number of Amer- gions. We begin with a descriptive portrait of icans living in such neighborhoods by more several different dimensions of the spatial op- than 90 percent, from 7.2 million to 13.8 mil- portunity structure, focusing on the distribu- lion (Jargowsky 2015).

rsf: the russell sage foundation journal of the social sciences an em pirical overview 3

Economic inequality in American neighbor- ments than other groups even after account- hoods overlaps with persistent racial and eth- ing for group differences in economic status. nic inequality. Residential segregation of Afri- The gaps are most notable when blacks and can Americans from Anglo Americans is most whites are compared. In all urban areas commonly measured two ways: the dissimilar- across the country, Patrick Sharkey (2014) ity index, which captures the comparative finds that black families with household in- evenness of the overall distribution of two ra- come of $100,000 or more live in and are sur- cial or ethnic groups across the neighborhoods rounded by neighborhoods with higher levels of a city or metropolitan area, and the isolation of disadvantage than white families with in- index, which captures the degree to which come of $30,000 or less (see also Logan 2011; members of a particular racial or ethnic group Reardon, Fox, and Townsend 2015). live in neighborhoods occupied by members of the same group.2 According to both mea- Schools and School Districts sures, the segregation of blacks from whites The research described in the previous section continues to be extremely high in many urban indicates that inequality in residential environ- areas, although black-white segregation has ments (especially in its economic composi- declined steadily since 1970 (Glaeser and Vig- tion) has been growing for all American fami- dor 2012; Logan and Stults 2011; Logan, Stults, lies. These same trends are amplified for and Farley 2014). Measures of evenness in the families with children. Evidence using several distribution of both Hispanics and Asian measures of racial and economic segregation Americans relative to whites show slight in- shows that households with children are dis- creases in the level of segregation over time, tributed less evenly across neighborhoods of whereas measures of their isolation show different racial-ethnic composition and eco- larger increases over time, consistent with the nomic status, respectively, than are house- rapid population growth of both groups (Lo- holds without children (Logan et al. 2001; Jar- gan 2011). gowsky 2015; Owens 2016). The relatively high Beyond the separation of racial and ethnic level of residential segregation among children groups from one another is the question of has important implications for schooling and the average economic status in the neighbor- academic achievement. hoods of each group. Economic segregation Trends in economic segregation in schools within racial and ethnic groups has been ris- are not as clear as trends in residential segre- ing over time, particularly for black and His- gation by income, mainly because of the ab- panic families since 2000 (Bischoff and Rear- sence of precise data on student economic sta- don 2014). This trend has led to more tus in schools across the country (Reardon and inequality within groups; however, between- Owens 2014). Research using student eligibility group inequality continues to be extreme. for free lunch as a proxy for low economic sta- Jargowsky (2015) shows that roughly five mil- tus has documented trends from 1990 to 2010, lion blacks live in neighborhoods with pov- and found growing segregation of low-income erty rates of at least 40 percent, a figure that students between school districts within the is higher than any other racial or ethnic same urban area, and growing segregation of group. Among the poor, 25 percent of blacks low-income students between schools within live in concentrated poverty, versus 7 percent the same district in the 1990s but not the 2000s of poor whites and 17 percent of poor His- (Owens, Reardon, and Jencks 2016). An alterna- panics. These figures reflect a broader set of tive approach focuses on the overall popula- findings in the literature demonstrating that tion of families living within school districts different racial and ethnic groups continue (available from 1970) or on families with chil- to live in highly unequal residential environ- dren in public schools (available since 1990).

2. The exposure index also is used frequently, but this measure is essentially the opposite of the isolation index and is designed to capture the degree to which members of a particular group are exposed to members of another group in their neighborhoods.

rsf: the russell sage foundation journal of the social sciences 4 spatial foundations of inequality

Using this approach, trends in the degree of lessness and economic status (Holzer 1991; segregation between school districts are largely Kain 1992). However, the original focus on the consistent with the trends in economic segre- location of jobs in central city versus suburban gation across neighborhoods. Economic seg- areas applied primarily to large urban centers regation between school districts rose for all in the Northeast and Midwest (Ihlanfeldt and families in the 1970s and 1980s, and continued Sjoquist 1998). Judith Hellerstein, David Neu- to rise for families with children in public mark, and Melissa McInerney (2008) argue for schools in the 1990s and 2000s (Owens, Rear- a more refined perspective that focuses on the don, and Jencks 2016). importance of access to jobs held by black Levels and trends in school racial segrega- Americans, particularly those held by blacks tion are more difficult to summarize. From the with the same level of education. The overall late 1960s through the end of the 1970s, the prevalence of jobs is shown to be less impor- segregation of black and white students within tant than the prevalence of jobs held by other school districts, measured by evenness and ex- blacks, suggesting that discrimination and em- posure, declined substantially (Reardon and ployment networks may be more relevant than Owens 2014). However, segregation between the raw presence of jobs in explaining racial school districts rose, particularly in the north. gaps in employment (see also Hellerstein, Since 1980, exposure of black students to white Kutzbach, and Neumark 2014; Waldinger 1996). students has fallen (Orfield and Lee 2007), a Recent research focuses on spatial inequal- pattern largely explained by the fact that the ity in wages, well-paid jobs, and economic population of students in the United States has growth across urban areas. From 1980 to 2010, grown more diverse over time. Focusing on metropolitan areas that initially had high evenness, trends in black-white school segrega- shares of college-educated workers have expe- tion depend on the exact time frame under rienced greater growth and demand for well- study, but most studies report modest in- paid workers, leading to growing inequality creases since 1980 (Reardon and Owens 2014). across metropolitan areas over time as the returns to college education have grown and ur- Jobs and Economic Opportunities ban economies have become increasingly bi- Early research on spatial inequality in access furcated (Lindley and Machin 2014). Two sets to jobs focused on the shift of employment op- of consequences have arisen. On the one hand, portunities away from central cities and into inequality between cities and metropolitan ar- the suburbs, arguing that changes in urban eas has grown as employment opportunities economies had contributed to growing racial have shrunk absolutely and real wages have gaps in joblessness and welfare receipt (Kain stagnated or fallen in places that are geograph- 1968; Wilson 1987). In putting forth his spatial ically and economically isolated from high- mismatch theory, John Kain (1968) argued that demand global cities. On the other hand, in- the combination of racial segregation, group equality within high-demand urban areas has variation in skills and human capital, discrim- widened as the growing returns to higher edu- ination in the labor and housing markets, and cation have created widening gaps between lack of access to employment networks and highly educated and less-educated workers employment opportunities helped explain the (Florida 2010). relatively high rates of joblessness among An additional strand of evidence docu- black Americans in central cities. Subsequent ments geographical variation in economic mo- research by William Julius Wilson (1996) fo- bility across commuting zones (sets of contig- cused on the growth of joblessness as a pri- uous counties that surround central cities and mary explanation for a set of changes and de- cover the entire nation), using data from tax teriorating conditions in high-poverty, central records for all Americans over time. Raj Chetty city neighborhoods. and colleagues (2014) show that a single, na- Descriptive evidence generally supports the tional measure of economic mobility or persis- argument that spatial proximity to jobs con- tence obscures the tremendous variation tributes to racial and ethnic disparities in job- across regions of the United States and across

rsf: the russell sage foundation journal of the social sciences an em pirical overview 5 specific commuting zones. Subsequent re- 1.7 times greater than in Hispanic neighbor- search exploiting sibling differences in time hoods. The greater exposure of blacks to air spent in low- and high-mobility commuting pollution is particularly pronounced for very zones suggests that places themselves have low-income blacks compared with their white causal effects on the probability of upward mo- and Hispanic counterparts, but is present at bility, although less progress has been made every level of income. Downey and Hawkins in identifying characteristics of places that fa- show that, just as middle- and upper-income cilitate or impede upward mobility (Chetty and black Americans live in neighborhoods with Hendren 2015). levels of disadvantage comparable to poor The idea that places exert independent ef- whites, black households with income equal to fects on the economic outcomes of residents or greater than $50,000 live in neighborhoods receives further support from recent research with more air pollution than white households focusing on the magnitude of the employment making less than $10,000 per year. shock experienced by local areas during the Although racial and economic gaps in expo- Great Recession. Danny Yagan (2016) compares sure to environmental toxins are present at the workers at the same level and in the same retail national level, Downey (2007) documents tre- firm but located in areas of the country that mendous variation in the degree of environ- were hit more or less hard by the economic mental inequality across U.S. metropolitan ar- downturn that began in 2008. Workers in areas eas. In some urban areas, black and Hispanic hit harder by the recession were 1 percentage residents are exposed to much higher levels of point less likely to be employed in 2014, several environmental hazards than whites, while in years after the recession had ended. others there is no disparity. Even as more attention has been paid to pollution and envi- Environmental Hazards ronmental hazards, racial and ethnic gaps In the 1980s, attention began to be focused on have persisted. Kerry Ard (2015) analyzes air the siting of environmental hazards in com- pollution from 1995 to 2004 and finds sharp munities across the country, revealing a pat- declines in average levels of pollution over time tern in which hazardous waste sites were dis- nationally, although declines in central cities proportionately located in communities were less pronounced. However, the gap in ex- occupied primarily by racial and ethnic minor- posure to air pollution between blacks and ity groups (Bryant and Mohai 1992). Since that whites has not changed over time, suggesting time, research has proliferated on spatial vari- the persistent, important role of geography. ation in air pollution, environmental toxins Persistent racial and ethnic gaps in expo- like lead, siting of manufacturing plants, and sure to environmental toxins relate closely to the location of chemical accidents and hazard- a large literature on inequality in health and ous waste. well-being, which is reviewed elsewhere (Diez- This research has found consistent evi- Roux and Mair 2010; Kawachi and Berkman dence that, within most metropolitan areas, 2003; Sampson 2003). Recent research docu- different forms of environmental hazards are ments spatial variation in especially harmful more common in low-income communities toxins, notably lead. At the county level, Paul and in communities of color, though the de- Stretesky (2003) documents an association be- gree of inequality by economic status and race- tween the percentage of black children and air ethnicity varies depending on the specific type lead concentrations. The few studies con- of hazard under study. Liam Downey and Brian ducted at the neighborhood level in specific Hawkins (2008) analyze variation in the con- places document similar links between racial centration of air pollutants released from in- composition and elevated blood lead levels (for dustrial facilities by neighborhood racial- example, Lanphear et al. 1998). Robert Samp- ethnic composition and income; they find that son and Alex Winter (forthcoming) use de- black Americans live in neighborhoods with tailed data from Chicago children measured concentrations of toxic releases 1.45 times from 1995 to 2013 to track changes in elevated greater than those in white neighborhoods and blood lead levels in block groups across the

rsf: the russell sage foundation journal of the social sciences 6 spatial foundations of inequality city, documenting enormous gaps in elevated dren (Sharkey and Sampson 2015). However, blood lead by race and ethnicity in the mid- violent crime is one of the few dimensions of 1990s; in some predominantly black neighbor- spatial inequality that has changed in a posi- hoods, more than 90 percent of tested children tive way over time. Over the past twenty years, had elevated levels of blood lead. These rates, the national rates of violent crime and homi- however, dropped precipitously over the period cide have roughly halved. The cities with the for all groups. Although black and Hispanic highest rates of violence in the 1990s, which neighborhoods continued to have higher rates had disproportionately high prevalence of ra- of elevated blood lead, prevalence was well be- cial and ethnic minority populations and high low 10 percent in all communities by the end levels of poverty, have seen the greatest de- of the period under study. Sampson and Win- clines in violent crime since then (Ellen and ter argue that the patterns reveal both the O’Regan 2009). enormous spatial disparities in exposure to en- Only limited evidence is available on vironmental toxins as well as the power of pub- trends in crime at the level of neighborhoods. lic health intervention to reduce or eliminate Michael Friedson and Sharkey (2015) draw on the consequences of environmental inequality. data from six cities where it is possible to track neighborhood-level trends in crime Violence over at least a decade, and find that the great- Among the developed nations of the world, the est absolute declines in violent crime oc- United States has a relatively high violent curred in the most violent neighborhoods. In crime rate and an extremely high homicide four of the six cities, the relative decline of vi- rate (UNODC 2013). However, the national rate olence was also largest in the cities’ most vio- of homicides obscures the variation in the lent neighborhoods; in the remaining two, prevalence of violence across cities and neigh- the proportional decline of violent crime was borhoods. Whereas many cities are remarkably roughly equivalent in the most violent neigh- safe, cities such as St. Louis, Detroit, and Bal- borhoods and in the rest of the cities. In timore have homicide rates that rival the most these six cities, the trends mean that the level violent, war-torn places in the world (Federal of inequality in community violent crime has Bureau of Investigation 2015). Within cities, declined over time, though it remains severe. violent crime is unevenly distributed across The degree to which poor and nonpoor resi- neighborhoods. For example, in a city like Chi- dents, and white and nonwhite residents, are cago with a high average violent crime rate, exposed to violent crime in their neighbor- some communities in the northern part of the hoods has converged. city are largely untouched. In the neighborhoods of the city’s south and west sides, how- An Interim Summary ever, violence is concentrated in communities Our first claim in this paper is that various di- characterized by poverty, ethnic isolation, and mensions of inequality are organized in space. institutional decay (Papachristos 2013; Samp- We review available evidence and document son 2012). the degree of spatial inequality in economic The spatial concentration of violence is not status, access to well-paid jobs, exposure to en- unique to Chicago. Research from cities with vironmental hazards, and exposure to commu- low or average crime rates, such as Seattle and nity violence. We find that spatial inequality Boston, shows that a disproportionate share of frequently is overlaid with racial and ethnic violent crime takes place within a few city segregation of neighborhoods and schools. blocks and street segments (Braga, Papachris- Where possible, we track changes over time in tos, and Hureau 2010; Braga, Hureau, and Pap- spatial inequality, and show that neighbor- christos 2011; Weisburd et al. 2004). hood economic segregation, concentrated pov- A growing strand of evidence suggests that erty, ethnic residential segregation between violence may be one of the central mecha- whites and both Hispanics and Asians, and nisms by which growing up in a disadvantaged economic and ethnic segregation of schools neighborhood affects the life chances of chil- have risen over time. By contrast, racial resi-

rsf: the russell sage foundation journal of the social sciences an em pirical overview 7 dential segregation between whites and blacks trol (collective norms, role models, peers); and and spatial inequality in community violence local political systems. By achieved socioeco- have declined. nomic status here we mean earnings, wealth, This review suggests that the fault lines for and occupational attainment. spatial inequality may be gradually shifting in Various elements of the spatial opportunity the United States. Whereas the peak of racial structure operate at and vary across spatial segregation, joblessness, and violent crime scales, as demonstrated. This variation occurs may have passed several decades ago, the rise across at least three distinct scales. Across of economic segregation may suggest that eco- neighborhoods, variations in safety, natural nomic status will become an increasingly im- environment, peer groups, social control, in- portant dimension of urban inequality. Fur- stitutions, social networks, and job accessibil- ther, the focus on trends in spatial inequality ity occur. Across local political jurisdictions, should not distract from the finding that U.S. health, education, recreation, and safety pro- neighborhoods continue to feature severe lev- grams vary. Across metropolitan areas, the lo- els of racial, ethnic, and economic segregation cations of employment of various types and and high levels of inequality in violence, envi- associated wages, working conditions, and ronmental hazards, and other features of com- skill requirements vary and housing and other munities thought to be most directly linked to market conditions that affect individuals’ op- the life chances of residents. In the following portunities for advancement differ.4 section, we present a conceptual model de- We view the spatial opportunity structure signed to explain the processes and mecha- as affecting socioeconomic outcomes viastruc - nisms behind the causal effect of such unequal turing opportunity both directly and indirectly. places on individuals’ socioeconomic opportu- It directly affects how, during a given span of nities. time, a given set of personal attributes will pay off in terms of socioeconomic status achieve- A Model of How Variations in the ments. The spatial opportunity structure indi- Spatial Opportunity Structure rectly affects over a longer span the set of at- Generate Inequalities in tributes individuals bring to the opportunity Individuals’ Achieved structure. Some of these indirect effects re- Socioeconomic Status quire little or no individual volition to acquire, We are interested in understanding how the such as aspects of mental and physical health space or spaces in which individuals are em- that may be passively acquired merely by living bedded influences their socioeconomic out- in the natural, built, and social environment comes. We conceptualize this aspect of space and collective norms and local networks that as spatial opportunity structure, the panoply of influence what information people receive and markets, institutions, services, and other nat- how they evaluate it. In the case of children ural and human-made systems that have a and youth, other indirect effects transpire geographic connection and play important through influences on the caregivers that affect roles in people's socioeconomic status achieve- the resources and parenting behaviors brought ments.3 The spatial opportunity structure in- to bear in the household. A final indirect effect cludes labor, housing, and financial markets; occurs by molding individual volition involved criminal justice, education, health, transporta- in decisions related to education, risky behav- tion, and social service systems; the natural iors, marriage, fertility, labor force participa- and built environment; public and private in- tion, illegal activities, and sociopolitical par- stitutional resources and services; social net- ticipation. Decisions regarding these domains works; forces of socialization and social con- are so crucial in determining socioeconomic

3. Similar notions of opportunity structure and geography of opportunity were first introduced by George Galster and Sean Killen (1995).

4. Of course, within any of these three scales contextual variations in all noted domains are entirely possible.

rsf: the russell sage foundation journal of the social sciences 8 spatial foundations of inequality outcomes in our society that we label them life this is portrayed as path C in figure 1. Some decisions. In the following section, we amplify decisions—our life decisions just mentioned— and illustrate these concepts and relationships are especially important in establishing a tra- with the aid of a heuristic visual model. jectory for achieved status outcomes.5 These include actions related to employment, crime, A Heuristic Model of Achieved child-rearing, cognitive and vocational skills, Socioeconomic Status educational credentials, smoking, drinking, A visual model of our conceptual framework substance abuse and other aspects of health, for understanding how space provides a foun- and social networks. Of course, the norms, as- dation for inequalities in achieved status is pirations, information, and resources individ- presented in figure 1. To begin with the most uals bring to bear in a particular life decision- basic and obvious relationship, an individual’s making situation is substantially influenced by attributes will play a fundamental role in pro- multiple inputs supplied by their parents or ducing markers of achieved socioeconomic caregivers, both currently and perhaps previ- status; this is represented by path A in figure ously in their lives, as represented by path D 1. If the individuals in question are adults, we in figure 1. would expect that interpersonal variations in their current bundles of achievement- Spatial Opportunity Structure as Mediator influencing attributes would explain substan- Between Personal Attributes and Achieved Status tial variation in their contemporaneously mea- At this point in our exposition, we take all sured achieved socioeconomic status; in the these fixed and malleable attributes as prede- case of children, current attributes would be termined so we can isolate one crucial role predictive (though less precisely) of future played by the space where the individual is cur- achieved socioeconomic status at some point rently embedded. We posit that the spatial op- or points as adults. Some personal character- portunity structure serves as a mediator be- istics are essentially fixed over the lifetime of tween individuals’ current characteristics and the individual, inasmuch as they are associated their socioeconomic status outcomes (see path with the vagaries of conception and birth. Such A in figure 1). Because the spatial opportunity fixed attributes would include, for example, ge- structure varies dramatically across and within netic signature, place and year of birth, and metropolitan areas in the ways that it evaluates many (though not all) characteristics of the in- personal attributes in the process of translat- dividual’s parents and ancestors. Other per- ing them into achieved status, one’s chances sonal characteristics are (potentially) more for such achievements will be enhanced or malleable over a lifetime. Some may be ac- eroded depending on place of residence, work, quired passively, such as through child-rearing and routine activity space. Several illustrations activities of one’s parents, as is portrayed in make our point. Metro areas with labor market path B in figure 1. Other malleable attributes actors that are more prone to discriminate on will be the product of previous decisions and the bases of gender or ethnicity against those actions by the given individual even though, who have decided to apply for jobs will dimin- once acquired, may no longer be malleable; ish the expected socioeconomic payoffs from

5. We recognize the voluminous literature on human decision-making and considerable debate over the most appropriate model (see review in Galster and Killen, 1995). We think it irrelevant for our model which particular view is taken, so long as one rejects the notion that these choices are purely instinctual or random, having no relationship with the social construction of a current and prospective reality. We think these choices may generally be described as based on bounded rationality: imperfect (perhaps even incorrect) information, subjective assessments, and varying degrees of dispassionate, analytical thought versus impulse and snap judgments contingent on personal context. Though our model has many features in common with the “rational actor” model of Erikson and Jonsson (1996) and Becker (2003) we stress that socio-spatial context is a prime source of information and values related to an individual’s assessments of expected benefits and costs.

rsf: the russell sage foundation journal of the social sciences an em pirical overview 9

Figure 1. Conceptual Framework

Parents’ / Caregivers’ Attributes and Behaviors G I D B H J

Individual’s Spatial Individual’s Life Decisions Individual’s Attributes Opportunity Achieved C A Structure Status

F E

Source: Authors’ compilation.

whatever attributes women and ethnic minor- often-overlooked yet powerful influence is ex- ities bring to the workplace. Even the most at- erted in three distinct ways through the passive tractive attributes from an employer’s perspec- and active acquisition or modification of per- tive may not yield a high income if the potential sonal attributes over time. First, through envi- employee lives far from potential workplaces ronmental exposure, it directly influences and cannot find a suitably fast and reliable some attributes of individuals over which they form of commuter transportation. Under- may exercise little or no volition. Second, it di- resourced, poorly administered schools with rectly influences the attributes of individuals weak teachers and a cadre of disruptive, violent over which they exercise considerable volition peers will be less likely to leverage students’ by shaping what they perceive is the most de- curiosity and native intelligence into literary sirable, feasible option in the process of mak- and numerical competence and, ultimately, ing life decisions. Third, in the case of children marketable educational credentials for those and youth, the spatial opportunity structure who have decided to get a diploma. Those with indirectly influences their attributes through little to no work experience may find that induced changes in the resources, behaviors, neighborhoods dominated by illegal or under- and attitudes of their caregivers. Diagrammat- ground markets will favorably evaluate some ically, we now turn our attention to paths E, F, of their attributes (such as present orientation, and G portrayed in figure 1. predilection for violence) that were discounted Personal attributes are constantly being in mainstream labor markets. Women living in molded by the physical and social environ- neighborhoods dominated by patriarchal ments in which a person lives, even if such norms and collective socialization into rigid molding has not been consciously chosen and gender roles will be less able to convert even may be unobserved by the individual; this is the most productive personality attributes and represented by path E in figure 1. Several ex- educational credentials into socioeconomic amples for the physical and social scientific achievements in the larger society. literature illustrate our point. We know, for example, how variations in air pollution can be Spatial Opportunity Structure as a associated with a range of health outcomes Modifier of Personal Attributes (McConnell et al. 2010; Lovasi et al. 2011). Lead As potent as the effects of the spatial opportu- associated with neighborhoods with older nity structure as mediator may be, we think an housing stock has been shown to cause perma-

rsf: the russell sage foundation journal of the social sciences 10 spatial foundations of inequality nent damage to children’s cognitive functions the medium through which the impacts of the and attention spans (Rau, Reyes, and Urzúa spatial opportunity structure are transmitted 2013). Exposure to violence (both as a victim to those under their care. As illustration, evi- and witness) creates physical, mental, and dence is ample that the health (mental and emotional responses that, among other things, physical) and resources (economic and social) have been shown to interfere with academic of parents have a profound effect on how chil- performance (Sharkey 2010; Sharkey et al. 2012, dren develop in multiple domains (Haveman 2015). Neighborhood or school-based peers, and Wolfe 1994). Thus, should the spatial op- role models, and other collective socialization portunity structure have an impact on any of forces can shape one’s norms, preferences, as- these domains through any of the causal pro- pirations, and behaviors. cesses modeled, the indirect causal link to the As noted, an individual’s attributes are also succeeding generation will be made. A variant modified as a result of the individual’s actions. of this connection is that caregivers have been Of primary importance for achieved status is observed to alter their parenting styles in re- what we termed life decisions. The spatial op- sponse to their perceptions of the spatial con- portunity structure affects such decisions by text in which their children must operate (Gal- shaping individuals’ perceptions of what is the ster and Santiago 2006). most desirable, feasible course of action; this relationship is portrayed in path F in figure 1. Feedback Effects These decision-shaping effects of the spatial To complete our conceptual model we con- opportunity structure are transmitted by influ- sider several feedback effects (see dotted lines encing: what information about the individu- in figure 1). Once a particular life decision has al’s options is provided, what the information been made, the associated attribute becomes objectively indicates about payoffs from these part of the individual’s “résumé” (path H in options, and how the information is subjec- figure 1). This change in the portfolio of attri- tively evaluated by the individual. Local net- butes will affect the individual’s opportunities works can affect the quantity and quality of in the future, perhaps irreversibly, depending information that an individual can access re- on the life decision in question. Certainly the garding the opportunity set. The notion of so- acquisition of educational credentials provides cial isolation associated with minority neigh- a lifelong change in one’s feasible set of op- borhoods of concentrated disadvantage portunities; so does being convicted of a fel- (Wilson 1987) is illustrative. The collective ony. Less obviously, prior life decisions may norms operating within these networks can reshape individuals’ aspirations, preferences, also shape which media of information trans- and evaluative frames. For example, a decision mission are considered more reliable sources to raise children may intensify aversion to risky of data about the opportunity structure. Neigh- entrepreneurial ventures or participation in il- borhood or school-based peers, role models, legal activities. Similarly, if choices to seek and other collective socialization forces can long-term employment have consistently been shape a person’s norms and preferences, frustrated, willingness to invest in human cap- thereby altering the perceived prospective pay- ital development for the future and respect for offs associated with various life decisions. civil authority may wane, leading to a reevalu- Finally, the spatial opportunity structure ination of feasible options in the opportunity set. directly affects the attributes children and A decision to participate in gang activities may youth will exhibit by shaping the resources, at- expose those individuals to different attitudi- titudes, health, and parenting behaviors of nal and aspirational norms that likely alter their adult caregivers; this portrayed as path G their assessments of many options in life deci- in figure 1. In our discussion of paths E and F, sion set. we describe the various mechanisms of how What one has achieved at a given moment spatial context can affect a person’s attributes; in terms of markers of socioeconomic status our point here is simply to note that when such (income, wealth, and occupation) also gener- persons happen to be caregivers they become ates two feedback effects. The first is that the

rsf: the russell sage foundation journal of the social sciences an em pirical overview 11 degree of achieved status shapes the bundle of time. One’s stock of attributes measured at any attributes the person will develop in the future given time will be shaped by the niche or by altering the degree of financial constraint niches of the spatial opportunity structure on obtaining certain attributes (path I in figure they have experienced in the past, both directly 1). For example, greater accumulated wealth by and indirectly through its influence on previ- a certain time in life permits people hence- ous life decisions and actions by caregivers. forth to buy superior training and credentials, Going forward, this set of attributes will con- maintain better health and free themselves strain (to a greater or lesser extent depending from constraints on employment by offloading on the attribute bundle and past socioeco- some child-care responsibilities on to hired nomic achievements) the perceived life deci- caretakers. One will be exposed to different sion options and associated expected payoffs. sources of information, collective norms, peer By way of illustration, a person who has effects, and role models in the workplace de- dropped out of high school and served jail time pending on occupation. for being convicted of a felony will have far Finally, and perhaps most fundamentally, fewer options in the future for socioeconomic achieved status affects what spatial opportu- status achievements than a person who has a nity structure one confronts (path J in figure graduate degree and no brushes with the law; 1). Clearly, for most households in the United and expected financial payoffs associated with States that do not receive subsidies for hous- any similar life decision options they share ing, their residential location and associated (such as working full time) will differ signifi- characteristics of the spatial opportunity struc- cantly. These differences in opportunities will ture will depend on their ability to pay for in turn lead both people down different paths housing. Residential sorting on the bases of of sequential life decisions in the future. Abet- income and wealth is to be expected in an ting this mutually reinforcing sequence of de- economy in which the market performs the cisions over the life course is the financial ef- main resource allocation functions. Other ex- fect on what parts of space one can afford to posures to the spatial opportunity structure access. Those whose paths have resulted in (via interfaces with schools, transportation sys- substantial achievements in status early in life tems, retail shopping, and workplaces) are can afford to occupy more privileged niches similarly molded by ability to pay for products later on, providing themselves and their off- and services. Households with the greatest fi- spring with even better attributes and oppor- nancial means select what they perceive as the tunities, which in turn will spawn even more most desirable niches in which to live and un- productive life decisions. dertake their routine activities, which are ce- teris paribus the most highly priced. The finan- Evolution of the Spatial cial exclusivity of these spaces can be abetted Opportunity Structure by a variety of zoning codes and other develop- This description has taken the housing mar- ment restrictions if the well-heeled can politi- ket, a prime driver of the spatial opportunity cally dominate a local jurisdiction to serve structure, as given. From the perspective of an their interests. At the other extreme, house- individual decision-maker, the assumption is holds with little to no market power are rele- reasonable. From a longer-term, general equi- gated by default to the least expensive, residual librium perspective, the housing market in pockets of the spatial opportunity structure: particular and the spatial opportunity struc- slums, ghettos, and the streets. ture in general is constantly evolving, partly in response to how the population has been sort- Cumulative Causation and ing themselves as housing demanders across Path Dependencies the metropolitan area. As documented earlier, This model should make it obvious that we this sorting process has produced a consider- view the processes involved in achieving socio- able racial and economic segregation and wide economic status as cumulative, path depen- variations in many other contextual indicators. dent, and (typically) mutually reinforcing over It is beyond the scope of this model to consider

rsf: the russell sage foundation journal of the social sciences 12 spatial foundations of inequality all these forces shaping the spatial opportunity their status are relegated to inferior niches in structure; a few illustrative comments are in the spatial opportunity structure, where their order regardless. subsequent choices tend to perpetuate their Some of the alterations in the spatial op- inferior status. When society as a whole views portunity structure may be exogenous to some of these decisions as social problems households, such as a technologically or inter- concentrated in low-status niches, the larger national trade-induced industrial restructur- opportunity structure can be altered in many ing. Other alterations, however, may be influ- ways. Those with financial means move away enced by the aggregate behaviors of households from neighborhoods and schools of concen- within a metropolitan area that have been pro- trated disadvantage, weakening the local retail duced by a previous period’s opportunity sector and the entry-level job opportunities structure. For example, the poor quality of the they provide. The same moves may strain the local public school system serving a neighbor- financial capacity of the local political jurisdic- hood may constrain children’s ability to gain tion, forcing a retrenchment in public services. good skills and credentials. Yet, if many par- Certain locales can thus generate a self- ents decide to participate in a collective politi- reinforcing spiral of spatial decline and indi- cal process, the result may be a reallocation of vidual impoverishment. fiscal resources to improve the local schools. The educational background of the parents of Interim Summary of Heuristic Model students living in the district is also an impor- Within the framework summarized in figure 1, tant constraint on school outcomes. Inasmuch it is easy to comprehend how space plays a vi- as better-educated parents create more intel- tal role not only as a foundation for inequality lectually stimulating home environments, bet- but also for perpetuating intergenerational inter monitor the completion of homework, and equality. Through cumulative causation and demonstrate more interest in what goes on in path dependency, those with the greatest sta- school, the quality of the classroom environ- tus achieved early in life can situate themselves ment will be improved for all students. So if, in a segment of the opportunity structure that in response to inferior public education, enhances their prospects for continued suc- better-educated parents move out of the dis- cess and provides their offspring with im- trict or enroll their children in private schools, proved chances for doing the same. Over time, the constraints on all parents who remain in the spatial opportunity structure in turn the public school system becomes tighter. For evolves in ways that further benefit those with example, housing developers may cater to par- the greatest achieved status. By contrast, those ents with substantial status by building new, who start with little typically are stuck in place, high-quality subdivisions that create exclusive both geographically and socioeconomically, as niches in the spatial opportunity structure. Af- Sharkey and Elwert (2011) document.6 ter incorporation, these niches may provide a wide range of attractive amenities and public Evidence on Effects of Spatial services that encourage the success of the chil- Context on Individual Outcomes dren living there. The second claim motivating this paper is that Thus, those who are successful in one the spatial organization of social and eco- round of spatial status competition are in a nomic inequality maintains or reinforces in- better financial position in the next round in equalities across multiple domains of social the evolving structure, thereby improving their and economic status. Obtaining unbiased, and their children’s odds of perpetuating this meaningful estimates of the independent, success and of generating market forces that causal effect of spatial components of an indi- alter the structure itself over time. Conversely, vidual’s existential context is challenging (for those who early in life make little headway in an extensive discussion, see Galster 2008). Per-

6. The normative underpinnings of the geographic opportunity structure are presented by Dawkins (forthcoming).

rsf: the russell sage foundation journal of the social sciences an em pirical overview 13 haps the most contentious aspect, however, is invariant individual characteristics are the issue of geographic selection bias (Manski eliminated by measuring differences be- 1995, 2000; Duncan, Connell, and Klebanov tween two periods in outcomes and spatial 1997; Ginther, Haveman, and Wolfe 2000; Dietz contexts (Bolster et al. 2007; Galster et al. 2002). The central issue is that individuals be- 2008; Musterd et al. 2008; van Ham and ing studied (or their parents) likely have un- Manley 2009; Galster, Andersson, and Mus- measured motivations, behaviors, and skills terd 2010). related to their own (or their children’s) socio- • Fixed-effect models based on longitudinal economic prospects and move from and to cer- data. Unobserved, time-invariant character- tain types of places as a consequence of these istics of individuals that may lead to both unobserved characteristics. Any observed rela- geographic selection and outcomes are tionship between geographic conditions and measured by individual dummy variables outcomes for adults or their offspring may (Weinberg, Reagan, and Yankow 2004; Mus- 7 therefore be biased. Skeptics may rightly ar- terd, Galster, and Andersson 2012). gue that what is being measured is simply another impact of (unmeasured) individual attri- • Instrumental variables for spatial context butes, not the impact of the space in which the characteristics. Proxy variables for geo- individual resides. graphic characteristics are devised that only vary according to attributes exogenous to Estimating Causal Impacts of the individual and thus are uncorrelated Spatial Context with their unobserved characteristics (Dun- Three general empirical approaches have been can, Connell, and Klebanov 1997; Crowder adopted in response to the challenge of geo- and South 2003; Crowder and Teachman graphic selection bias. The most common ap- 2004; Galster et al. 2007; Kling, Liebman, proach consists of a variety of econometric and Katz 2007; Ludwig et al. 2008; Cutler, techniques applied to observational (nonex- Glaeser, and Vigdor 2008; Sari 2012; Hed- perimentally generated) longitudinal datasets man and Galster 2013; Damm 2014). involving individuals and their spatial con- • Residents of same block. If sorting on indi- texts. The two other, less common approaches vidual unobservables at the census block use natural or experimental designs to gener- level is minimal, then the impacts of net- ate quasi-random or random assignments of works among these very localized neigh- households to neighborhoods. bors should be free of geographic selection bias (Bayer, Ross, and Topa 2008) Econometric Models Based on • Timing of events. Individuals moving into Observational Data certain, well-defined types of places (such Most studies of spatial context effects have as public housing developments) after an used observational data collected from surveys event being investigated (such as a school of individual households in a variety of places achievement test) are likely to share com- as a result of mundane factors associated with mon unobservable characteristics with in- normal market transactions. The subset that dividuals moving into the same places just has tried to overcome geographic selection before the event, so the short-term effect of bias uses one or more of the following ap- the place can be measured by comparing proaches (Galster and Hedman 2013): the two groups’ outcomes (Weinhardt 2014); • Difference models based on longitudinal analogously, Sharkey (2010) and Sharkey et data. The biases from unobserved, time- al. (2012, 2014) address the selection bias

7. The direction of the bias has been the subject of debate, both Christopher Jencks and Susan Mayer (1990) and Marta Tienda (1991) arguing that measured contextual impacts are biased upward, and Jeanne Brooks-Gunn, Greg Duncan, and Lawrence Aber (1997) arguing the opposite. Lisa Gennetian, Lisa Sanbonmatsu, and Jens Ludwig (2011) show that these biases can be substantial enough to seriously distort conclusions about the magnitude and direction of context effects.

rsf: the russell sage foundation journal of the social sciences 14 spatial foundations of inequality

problem by exploiting the variation in the mies adequately capture the bundle of unob- timing of local homicides compared with servables for all times during the panel and interview assessments for a sample of chil- that the effect of this bundle remains constant dren in families that have previously se- during the panel. Instrumental variables must lected the same neighborhood. be both valid and strong. Micro-scale investiga- • Propensity score matching. Individuals who tions are limited to neighborhood effect mech- are closely matched on a wide variety of ob- anisms than operate only at the small geo- servable characteristics that predict their graphic scales and assume no residential similar residential mobility behavior are sorting on unobservables at that scale. Relying likely to be well matched on their unobserv- on the timing of moves immediately before able characteristics as well; comparisons and after and event assumes that context ef- between matches of differences in their spa- fects operate quickly after exposure. Propensity tial contexts and individual outcomes score matching requires assumptions about should thus provide unbiased causal evi- the strong relationship between unobservable dence (Harding 2003). and observable characteristics of individuals. Those who do not move may be exhibiting res- • Inverse probability of treatment weighting idential selection based on unobserved char- (IPTW). Like propensity score matching, acteristics. IPTW uses a model of selection into the treatment status to predict the probability Quasi-Random Assignment that an individual is in the treatment state Natural Experiments in which the individual is observed. A It is sometimes possible to observe nonmarket weighted pseudo-sample is then con- interventions into households’ residential lo- structed in which the treatment and control cations that mimic random assignment. In the groups are balanced on observables. IPTW United States, such experiments typically have models selection into treatment status at been based on court-ordered, public housing multiple time points, allowing for unbiased racial-ethnic desegregation programs (Rosen- estimates of treatment effects over time in baum 1991; Briggs 1997; Fauth, Leventhal, and the presence of observed confounders that Brooks-Gunn 2007), regional fair-share hous- vary over time and may be endogenous to ing requirements (Schwartz 2010; Casciano the treatment. Sharkey and Felix Elwert and Massey 2012) or scattered-site public hous- (2011) use this method in combination with ing assignments (Santiago et al. 2014). In Can- a formal sensitivity analysis to estimate the ada and Europe, they have involved allocation cumulative effect of multigenerational ex- of tenants to social housing (Oreopoulos 2003; posure to neighborhood poverty on cogni- Damm 2009, 2014; Rotger and Galster 2015) or tive development. placement of refugees in particular locales • Nonmovers. Analyzing how exogenous (Edin, Fredricksson, and Åslund 2003; Åslund neighborhood changes induce different and Fredricksson 2009). outcomes for individuals who do not move Although these natural experiments may in- during the analysis period arguably avoids deed provide some exogenous variation in lo- the mobility selection issue (Sharkey 2012; cations, the geographic selection problem is Galster and Hedman 2013; Gibbons, Silva, unlikely to be avoided completely. In most and Weinhardt 2013, 2014). cases, program staff makes assignments and participants have some nontrivial latitude in None of these econometric fixes to observa- which locations they choose, both initially and tional datasets are without challenge. For ex- subsequent to original placement. Moreover, ample, difference models reduce statistical programs that involve rental vouchers (Gau- power by shrinking variation in the outcome treaux, for example) entail selection in who variable and assume that change relationships succeeds in locating rental vacancies in quali- are independent of starting conditions. Fixed- fying locations and signing leases within the effect models assume that the individual dum- requisite period. These various potential selec-

rsf: the russell sage foundation journal of the social sciences an em pirical overview 15 tion processes raise the possibility that low- Massey 2008; Sampson 2008; Burdick-Will et income families who succeed in living persis- al. 2011; Briggs, Popkin, and Goering 2010; tently in low-poverty neighborhoods were Briggs et al. 2008, 2011; Sanbonmatsu et al. especially motivated, resourceful and, perhaps, 2011; Ludwig 2012). The debate focuses on five courageous—traits poorly measured by re- domains. First, although MTO randomly as- searchers but likely ones that would help them signed participants to treatment groups, it ran- and their children succeed irrespective of their domly assigned characteristics neither of spatial contexts. Additional empirical prob- neighborhoods initially occupied by voucher lems can arise if sampled subjects move holders (except maximum poverty rates for the quickly from their quasi-randomly assigned experimental group) nor of neighborhoods in dwellings to another location, thereby mini- which participants in all three groups moved mizing exposure to measured context and po- subsequently. Thus, a question remains about tentially confounding consequences because the degree to which geographic selection on moving itself can be disruptive. As time passes, unobservables persists. Second, MTO may not the randomness of location can erode as selec- have created adequate duration of exposure to tion of who stays in initially assigned places neighborhood conditions by any group at any and who moves away comes into play. Finally, location to observe much treatment effect. limitations are possible in the range of places Third, MTO overlooked the potentially indeli- to which study participants moved or were as- ble developmental effects on adult experimen- signed because of where available private tal group participants who spent their child- rental or subsidized housing was located, hoods in disadvantaged neighborhoods. thereby reducing the power of statistical tests Fourth, it appears that even experimental MTO to discern context effects. movers rarely moved out of predominantly Af- rican American–occupied neighborhoods near Random Assignment Experiments those of concentrated disadvantage and Many researchers advocate a random assign- achieved only modest changes in school qual- ment experimental approach for best avoiding ity and job accessibility. Thus, they may not biases from geographic selection. Data on out- have experienced sizable enhancements in comes that can be produced by an experimen- their geographic opportunity structures. For tal design whereby individuals or households these reasons, MTO may not have provided de- are randomly assigned to different geographic finitive evidence about the potential effects on contexts is indeed, in theory, the preferred low-income families from prolonged residence method. In this regard, the U.S. Moving to Op- in multiply advantaged neighborhoods, de- portunity (MTO) demonstration has been spite its theoretical promise and conventional touted conventionally as the study from which wisdom. to draw conclusions about the magnitude of In summary, none of the three approaches neighborhood effects (Smolensky 2007; San- to measuring effects of spatial context has bonmatsu et al. 2011; Ludwig 2012). The MTO proven limitation-free and unambiguously su- research design randomly assigned public perior. Nevertheless, they as a group offer the housing residents who volunteered to partici- strongest, plausibly causal evidence to date on pate in one of three groups: controls that got the topic at hand. In our review, therefore, we no voucher but stayed in public housing in dis- synthesize findings only from these method- advantaged neighborhoods, recipients of ologically rigorous studies that use one or rental vouchers with no restrictions, and re- more of the approaches. cipients of rental vouchers and relocation assistance who had to move to census tracts with An Overview of the Scientific less than 10 percent poverty rates and remain Literature on Spatial Context Effects for at least a year. We organize our review by six outcome do- Debate over the power of MTO as an unam- mains that clearly are related to socioeconomic biguous test of spatial context effects has been opportunity: cognitive and behavioral develop- considerable (see Clampet-Lundquist and ment, educational performance and attain-

rsf: the russell sage foundation journal of the social sciences 16 spatial foundations of inequality ment, teen fertility, physical and mental sis showed that the effect of multigenerational health, labor force participation and earnings, neighborhood poverty was robust to substan- and crime. We emphasize at the outset that the tial potential bias arising from unobserved se- scope, diversity and complexity of the relevant lection processes. literature is vast in comparison with the length As described, Sharkey and his colleagues ex- imitations of this paper. Thus, we do not at- ploit the timing of incidents of violence to tempt to review findings in any detail, recon- identify the acute impact of exposure to vio- cile conflicting results, nor attempt any formal lence in children’s environments on their per- meta-analysis. Instead, our aim is basic: in formance on cognitive skills assessments. each outcome domain we tally the number of Sharkey (2010) finds that exposure to a recent (methodologically rigorous) studies that find homicide within close proximity to a child’s substantial, statistically significant effects of at home reduced the performance of African least some aspect of spatial context (for at least American children on tests of reading, lan- some set of individuals) and those that do not. guage, and applied problems by more than a third of a standard deviation. In a subsequent Cognitive Skills and study, Sharkey and his colleagues (2012) find Academic Performance similar impacts of recent exposure to nearby Recent meta-analysis of the international lit- homicides on children’s performance on vo- erature (Niewenhuis and Hooimeijer 2014) and cabulary assessments as well as impacts on as- a comprehensive review of the U.S. literature sessments of impulse control and attention. (Sharkey and Faber 2014) find nontrivial neigh- Experimental evidence comes from the borhood effects on the development of cogni- Moving to Opportunity program, and shows tive skills, academic performance, and educa- mixed and complex results. Several years after tional attainment. Our assessment of the the experiment began and ten to fifteen years methodologically sophisticated literature later, no effects of the intervention were found reaches a similar conclusion, though the mag- for the full sample on assessments of cognitive nitude of the neighborhood effect likely varies skills (Sanbonmatsu et al. 2006, 2011). How- across individuals and groups. ever, the experiment generated positive effects Measures of cognitive skills have been used on the reading assessments of African Ameri- to assess evidence for neighborhood effects cans across all cities four to seven years after frequently over the past twenty years (for re- implementation (Sanbonmatsu et al. 2006); views, see Sastry 2012; Sharkey and Faber 2014), positive effects on reading and math scores for but fewer studies have taken steps to address the full sample of boys and girls among fami- the problem of selection bias. Two studies lies that remained in low-poverty neighbor- have modeled selection into poor neighborhoods for longer durations of time (Turner et hoods and then used inverse probability of al. 2012); and strong positive effects for chil- treatment weighting to identify the impact of dren in the Baltimore and Chicago sites, which long-term exposure to neighborhood poverty persisted over ten to fifteen years only for the on cognitive skill development. Using data Chicago sample (Burdick-Will et al. 2011; Lud- from Chicago, Sampson, Sharkey, and Stephen wig, Ladd, and Duncan 2001; Sanbonmatsu et Raudenbush (2008) find that living in neigh- al. 2011). borhoods of concentrated disadvantage leads The research literature on academic out- to substantial declines in reading and lan- comes and educational attainment is also guage skills assessed years later. Sharkey and large. Many studies using one or more of the Elwert (2011) use national data from the Panel described econometric techniques to obtain Study of Income Dynamics and find that fam- plausibly causal estimates from observational ily exposure to neighborhood poverty over con- datasets have been conducted. Methods in- secutive generations reduces children’s perfor- clude propensity score matching (Harding mance on tests of broad reading skills and 2003), sibling comparisons (Aaronson 1998; applied problems skills by more than half of a Plotnick and Hoffman 1999), fixed-effects (Plot- standard deviation. A formal sensitivity analy- nick and Hoffman 1999; Vartanian and Gleason

rsf: the russell sage foundation journal of the social sciences an em pirical overview 17

1999; Jargowsky and El Komi 2011), instrumen- lic housing natural experiment is exploited by tal variables (Duncan, Connell, and Klebanov Anna Maria Santiago and her colleagues (2014) 1997; Crowder and South 2003; Galster et al. to estimate neighborhood effects on the haz- 2007), nonmovers (Gibbons, Silva, and Wein- ard of low-income African American and La- hardt 2014), and timing of events (Sharkey et tino children being diagnosed with a neurode- al. 2014; Weinhardt 2014; Carlson and Cowan velopmental disorder (retardation, learning 2015). All of these find strong residential neigh- disabilities, developmental delays, autism, borhood effects on variously measured educa- ADD-ADHD). Several aspects of neighborhood tional outcomes, with only two exceptions: context (especially safety, prestige, nativity and Plotnick and Hoffman (1999), using U.S. data ethnic mix, neurotoxin pollution) proved and Gibbons, Silva, and Weinhardt (2013) and strongly and robustly predictive. Neurocogni- Weinhardt (2014) using U.K. data. tive developmental disorders were not investi- Numerous studies based on natural experi- gated directly. ments also are relevant in this outcome do- As for risky behaviors, we find only six stud- main. These include data based on Gautreaux ies of context effects involving the described and Yonkers public housing desegregation econometric approaches to overcoming geo- programs (Rosenbaum 1995; Fauth, Leventhal, graphic selection; most identified effects on and Brooks-Gunn 2007; DeLuca et al. 2010), risky behaviors except smoking. Using propen- public housing revitalization programs (Jacob sity score matching, Jennifer Ahern and her 2004; Clampet-Lundquist 2007), assignment to colleagues (2008) find that an individual’s pro- public housing waiting lists (Ludwig et al. pensity to drink was related to the neighbor- 2011), inclusionary zoning mandates (Schwartz hood’s culture of alcohol use; Scott Novak and 2010; Casciano and Massey 2012), combined as- his colleagues (2006), however, find only a sisted housing-education programs (Tach et al. small, barely discernable effect of retail to- 2016) and public housing assignments (Santi- bacco outlet density on youth cigarette smok- ago et al. 2014; Galster et al. 2015, 2016; Galster, ing. Markus Jokela (2014) uses the fixed-effect Santiago, and Stack 2015; Galster and Santiago, modeling approach and finds no impact of forthcoming). These natural quasi-experiments neighborhood disadvantage on the probability provided only one example of no observed con- of smoking. Stephen Gibbons, Olmo Silva, and text effects (Jacob 2004),8 though several of the Felix Weinhardt (2013) also used fixed effects observed effects in other studies were contin- but find that the share of neighbors from lower- gent on gender or ethnicity. status backgrounds increases the chances of Recent evidence from MTO on college at- teen boys engaging in anti-social behaviors like tendance is relevant to the discussion of neigh- graffiti, vandalism, shoplifting, fighting, or borhood effects on educational attainment. public disturbance. Two approaches using in- Chetty, Nathaniel Hendren, and Lawrence verse probability weighting (marginal struc- Katz’s (2015) reanalysis of MTO data found that tural model) methods find that neighborhood moving to a lower-poverty neighborhood sig- poverty is strongly related to the odds of binge nificantly increased college attendance rates drinking (Cerdá et al. 2010) and drug injecting for children who were younger than thirteen (Nandi et al. 2010). when their families moved to the neighbor- Only three examples of either the random hoods, compared with experimental group or quasi-random neighborhood assignment children who moved when they were older or approaches are relevant to risky behaviors. children in the other study groups. Strong context effects on risky behaviors ap- pear in both studies but are contingent on gen- Risky Behaviors and Violence der, ethnicity, and timing. Early MTO findings We now turn to evidence on developmental suggest that substantial reductions on girls’ disorders and risky behaviors. The Denver pub- rates of risky behaviors and boys’ drug use can

8. Little context effect was observed here because the experimental households did not use their housing vouchers to change their neighborhood characteristics significantly.

rsf: the russell sage foundation journal of the social sciences 18 spatial foundations of inequality be attributed to residence in lower-poverty tige and higher percentages of Latinos, though neighborhoods. However, after initial declines strength of effect depended on gender and eth- in risky behavior, boys living in lower-poverty nicity. Results from the MTO demonstration neighborhoods four to seven years after their show that girls in the experimental group first move were more likely to reengage in risky whose parent or parents moved to low-poverty behaviors (Sanbonmatsu et al. 2011). By the end neighborhoods felt safer and less pressured to of the demonstration project, girls assigned to engage in early sexual activity (and thus, by im- low-poverty neighborhoods were less likely to plication, early pregnancy and childbearing) in have serious behavioral problems. No group their new neighborhoods (Popkin, Leventhal, differences in more serious antisocial behav- and Weismann 2010; Sanbonmatsu et al. 2011). iors were significant, however (Sanbonmatsu Chetty, Hendren, and Katz (2015) analyze the et al. 2011). Santiago and her colleagues’ anal- subset of MTO experimental children who ysis of data from a Denver natural experiment moved to low-poverty neighborhoods before reveals that cumulative exposure to multiple they were thirteen and observe that they, in- dimensions of neighborhood context (espe- deed, were less likely to become single parents. cially safety, social status, and ethnic and nativity composition) affected the hazard of ado- Physical and Mental Health lescents running away from home, using Michael Oakes and his colleagues (2015) re- aggressive or violent behavior, or initiating cently completed a comprehensive review of marijuana use, though with substantial ethnic the empirical literature related to neighbor- heterogeneity of relationships. Finally, Magda- hood effects on health. After reviewing 1,369 lena Cerdá and her colleagues (2012) examine articles, using criteria similar criteria to ours, the impact of a new transportation infrastruc- they conclude that only about 1 percent pro- ture intervention in Medellin, Colombia, on duced plausibly causal estimates. A handful violent behavior. and find that investment de- use the described statistical techniques ap- creased violence significantly. plied to observational data and find that the results, though somewhat inconsistent, do not Teen Fertility point to strong context effects on health. Three Very few studies have used any of the described studies based on propensity score methods statistical techniques to account for potential demonstrated no or barely discernable effects geographic selection bias confounding obser- on minority infant mortality using different vational data on the fertility patterns of youth measures of neighborhood (Schootman et al. and their neighborhood contexts. The two ex- 2007; Johnson, Oakes, and Anderton 2008; ceptions are Robert Plotnick and Saul Hoffman Hearst, Oakes, and Johnson 2008). Based on (1999) and David Harding (2003). Plotnick and inverse probability weighting (marginal struc- Hoffman find that neighborhood effects on tural model) methods, researchers have deter- teen childbearing disappeared when using a mined that neighborhood poverty was related model of fixed effects with only observations to mortality in a strong but nonlinear way (Do, of sisters, whereas Harding finds that neigh- Wang, and Elliott 2013) but had mixed effects borhood effects remained significant despite on self-assessed health and disability (Gly- propensity score matching and argues that se- mour et al. 2010). Finally, Jokela (2014) uses the lection bias would need to be unreasonably fixed-effect modeling approach and finds no large to rule out causal effects of neighborhood impact of neighborhood disadvantage on self- socioeconomic conditions on teen childbear- rated health, mental health and physical func- ing. tioning, and amount of physical activity, in- The evidence from natural and random as- stead finding evidence of selection of those signment experiments is more consistent. San- with poorer health into more disadvantaged tiago and her colleagues (2014) find that haz- neighborhoods. ards of teenage childbearing and fathering The random assignment experimental evi- were greater in neighborhoods with higher dence here is also mixed but shows impacts on property crime rates, lower occupational pres- some health outcomes. MTO results show no

rsf: the russell sage foundation journal of the social sciences an em pirical overview 19 significant differences in child asthma rates the discussed econometric techniques on non- among groups assigned to neighborhoods experimental, observational datasets. Several with differing poverty rates, but does show ef- studies using U.S. data (Weinberg, Reagan, and fects on adult obesity and diabetes rates (Lud- Yankow 2004; Dawkins, Shen, and Sanchez wig et al. 2011) and much lower stress levels 2005; Cutler, Glaeser, and Vigdor 2008; Bayer, among adults and children among those as- Ross, and Topa 2008; Sharkey 2012), several us- signed initially to low-poverty neighborhoods ing Swedish data (Galster et al. 2008; Galster, (Sanbonmatsu et al. 2011). Findings for mental Andersson, and Musterd 2010, 2015, 2016; Mus- health suggested neighborhood effects but terd, Galster, and Andersson 2012; Hedman their size and direction were extremely varied, and Galster 2013), one Danish study (Damm depending on lag of measurement after assign- 2014) and one French study (Sari 2012) find ment, gender, and age (see Leventhal and nontrivial neighborhood effects on various Brooks-Gunn 2003; Kessler et al. 2014). Stepha- adult labor market outcomes such as income nie Moulton, Laura Peck, and Keri-Nicole Dill- and employment rates. One U.S.-based study man (2014) analyze the subset of MTO experi- (Plotnick and Hoffman 1999) and three U.K.- mental households who lived for substantial based analyses (Bolster et al. 2007; Propper et periods in low-poverty neighborhoods and al. 2007; van Ham and Manley 2010) find minor, conclude that health benefits may be much if any, neighborhood effects, and instead sug- larger for that group. gest geographic selection dominates. The few natural experiments involving Several researchers have probed the effect health outcomes consistently find neighbor- of spatial context on labor market outcomes hood effects, at least on selected health indica- exploiting the quasi-random assignment oc- tors. Debra Cohen and her colleagues (2006) curring in natural experiments. Studies in the use exogenous shocks in neighborhood alco- United States (Rosenbaum 1991, 1995; Ru- hol outlet density associated with the 1992 Los binowitz and Rosenbaum 2000; DeLuca et al. Angeles riots as a causal identification strategy 2010; Galster, Santiago, and Lucero 2015a, and find strong impacts on neighborhood gon- 2015b; Galster et al. 2015; Chyn 2016), in Swe- orrhea rates. Mark Votruba and Jeffrey Kling den (Edin, Fredricksson, and Åslund 2003; Ås- (2009) analyze data from the Gautreaux public lund and Fredricksson 2009), and in Denmark housing relocation program. They find that (Damm 2009, 2014) all find evidence of strong when young, low-income African American neighborhood effects on several measures of men relocated to higher-education neighbor- adult and teen labor market outcomes in their hoods their all-cause and homicide mortality analyses. The only finding of a trivial neighbor- rates dropped relative to those moving to more hood effect using this approach came from a disadvantaged areas. Finally, Santiago and her Canadian natural experiment study (Oreopou- colleagues (2014) find strong neighborhood ef- los 2003). fects on the diagnoses of several child and ad- Virtually no investigations using MTO data olescent health problems (asthma, obesity) us- uncovered any substantial short- or long-term ing data from the Denver public housing context effects on teen or adult labor market natural experiment, although the relationships outcomes (for example, Ludwig, Duncan, and often depended on gender and ethnicity and Pinkston 2005; Katz, Kling, and Liebman 2001; in some cases manifested nonlinear thresh- Ludwig, Ladd, and Duncan 2001; Ludwig, Dun- olds. Asthma problems, for example, arose can, and Hirschfield 2001; Orr et al. 2003; Kling, sooner for low-income, minority children re- Leibman, and Katz 2007; Ludwig et al. 2008; siding in neighborhoods that had more prop- Sanbonmatsu et al. 2011; Ludwig 2012). Three erty crime, lower occupational prestige, and exceptions are notable. Susan Clampet- higher concentrations of air pollution. Lundquist and Douglas Massey (2008) and Margery Turner and her colleagues (2012) ana- Labor Force Participation and Earnings lyze the subset of MTO experimental house- Most investigators find neighborhood effects holds who lived for extended periods in low- on labor market outcomes when using one of poverty neighborhoods and find that their

rsf: the russell sage foundation journal of the social sciences 20 spatial foundations of inequality adult employment and earnings outcomes are twenty-five who had just moved in committing substantially better than those of the control property and drug crimes over the next two group. Chetty, Hendren, and Katz (2015) ana- years. Stephen Billings, David Deming, and lyze the subset of MTO experimental children Stephen Ross (2016) study with natural experi- who moved to low-poverty neighborhoods be- mental data the determinants of youth crime fore they were thirteen and observed that they of fourteen-year-old students in Mecklenburg had significantly higher earnings as young County, North Carolina. They demonstrate adults than either experimental group children that peer effects on criminal behavior only who moved after age thirteen or children from arise when school peers (of the same race and the other study groups. gender) live less than a half-mile from each other and that the effects are stronger when Crime neighbors are assigned to the same grade. Six recent studies provide consistent evidence Finally, modest evidence indicates context concerning the strong (if heterogeneous by effects on male criminality from MTO. Early gender) causal impact of geographic context impact evaluations indicate fewer arrests on criminality. Mark Livingston and his col- among males from families randomly assigned leagues (2014) use temporal lags and neighbor- to low-poverty neighborhoods (Katz, Kling, hood fixed effects identify the effect of the and Liebman 2001; Ludwig, Duncan, and share of criminal offenders resident in Glasgow Hirshfeld 2001). These effects appeared to di- postal code areas. They find that higher neigh- minish and even reverse over time, however borhood shares of residents committing of- (Kling, Ludwig, and Katz 2005; Sanbonmatsu fenses during a quarter strongly predicted the et al. 2011). probability of first-time violent and property offenses among residents in the subsequent Findings quarter. Table 1 summarizes the findings on the num- Four natural experiments are relevant here. ber of (methodologically rigorous) studies that Exploiting the quasi-randomness of assign- have found substantial, statistically significant ment to public housing in Denver, Santiago effects of spatial context (for at least some set and her colleagues (2014) find that low-income of individuals) and those that have not, by out- Latino and African American youth had greater come domain. The tally makes it clear that the hazards of engaging in violent behaviors in preponderance of evidence in every outcome neighborhoods with lower occupational pres- domain is that multiple aspects of spatial con- tige and higher property crime rates, though text exert important causal influences over a with gendered impacts. Anna Damm and wide range of outcomes related to socioeco- Christian Dustmann (2014) use the Danish dis- nomic opportunity, though which aspects are persed settlement policy for refugees to iden- most powerful depends on the outcome and tify causal impacts of municipal characteristics the gender and ethnicity of the individuals in on youth criminality. They find that the share question. in a municipality of those age fifteen to twenty- five who were convicted of a crime during the Empirical Gaps in the Study of year the family was assigned there strongly Spatial Inequality raised the probability of young male (but not Trends showing the rise of inequality in in- female) refugees being convicted of a crime come and wealth have reached the mainstream (especially for violent crimes and younger as national and world leaders, policymakers, teens) during subsequent years. Gabriel Rotger and the public have become increasingly fo- and Galster (2015) use exogenous assignment cused on inequality in economic status as a to social housing in Copenhagen to identify defining issue of our time. Our goal in this in- strong causal effects of the prior drug offend- troduction, and in the issue as a whole, is to ing characteristics of the housing develop- shed light on the spatial dimensions of in- ment’s residents at time of assignment on the equality. We argue that space is a particularly odds of individuals ages fifteen through severe, and underappreciated, dimension of

rsf: the russell sage foundation journal of the social sciences an em pirical overview 21

Table 1. Conclusions from Causal Analyses of Neighborhood Effects

Significant Effects No Effects

Cognitive and behavioral development Ahern et al. 2008; Cerda et al. 2010; Nandi et al. 2010; Novak et al. 2006; Jokela 2014 Sanbonmatsu et al. 2011; Cerda et al. 2012; Gibbons, Silva, and Weinhardt 2013; Santiago et al. 2014, this volume

Educational performance and attainment Rosenbaum 1995; Duncan, Connell, and Klebanov 1997; Vartanian Plotnick and Hoffman 1999; and Gleason 1999; Crowder and South 2003; Clampet-Lundquist Ludwig, Ladd, and Duncan 2007; Fauth, Leventhal, and Brooks-Gunn 2007; Galster et al. 2001; Jacob 2004; 2007; DeLuca et al. 2010; Schwartz 2010; Sharkey and Sampson Sanbonmatsu et al. 2006, 2010; Jargowsky and El Komi 2011; Sharkey et al. 2012, 2014; 2011; Kling, Liebman, and Casciano and Massey 2012; Gibbons, Silva, and Weinhardt 2014; Katz, 2007; Gibbons, Silva, and Santiago et al. 2014; Carlson and Cowan 2015; Chetty, Hendren, Weinhardt 2013; Weinhardt and Katz 2015; Galster et al. 2015, 2016; Galster, Santiago, and 2014 Stack 2015; Tach et al. 2016; Galster and Santiago, forthcoming

Teen fertility Harding 2003; Popkin, Leventhal and Weismann 2010; Plotnick and Hoffman 1999 Sanbonmatsu et al. 2011; Santiago et al. 2014; Chetty, Hendren and Katz 2015; Galster and Santiago, forthcoming

Physical and mental health Leventhal and Brooks-Gunn 2003; Cohen et al. 2006; Votruba and Schootman et al. 2007; Hearst Kling 2009; Glymour et al. 2010; Ludwig et al. 2011; et al. 2008; Johnson et al. 2008; Sanbonmatsu et al. 2011; Do et al. 2013; Kessler et al. 2014; Jokela 2014 Moulton, Peck, and Dillman 2014; Santiago et al. 2014

Labor force participation and earnings Rosenbaum 1991,1995; Rubinowitz and Rosenbaum 2000; Edin, Plotnick and Hoffman 1999; Fredricksson, and Åslund 2003; Weinberg, Reagan, and Yankow Ludwig, Duncan, and Pinkston 2004; Dawkins, Shen, and Sanchez 2005; Cutler, Glaeser, and 2005; Katz, Kling, and Liebman Vigdor 2008; Bayer, Ross, and Topa 2008, Clampet-Lundquist and 2001; Ludwig, Ladd, and Massey 2008; Galster et al. 2008; Åslund and Fredricksson 2009; Duncan 2001; Ludwig, Duncan, Damm 2009, 2014; DeLuca et al. 2010; Galster, Andersson, and and Hirschfield 2001; Orr et al. Musterd 2010, 2015, 2016; Musterd, Galster, and Andersson 2003; Oreopoulos 2003; 2012; Sari 2012; Sharkey 2012; Turner et al. 2012; Hedman and Bolster et al. 2007; Kling, Galster 2013; Damm 2014; Galster, Santiago, and Lucero 2015a, Leibman, and Katz 2007; 2015b; Chetty, Hendren, and Katz 2015; Galster et al. 2015; Chyn Propper et al. 2007; Ludwig et 2016; Galster and Santiago, forthcoming al. 2008; van Ham and Manley 2010; Sanbonmatsu et al. 2011; Ludwig 2012

Crime Katz, Kling, and Liebman 2001; Ludwig, Duncan, and Hirshfeld Kling, Ludwig, and Katz, 2005; 2001; Livingston et al. 2014; Santiago et al. 2014; Damm and Sanbonmatsu et al. 2011 Dustmann 2014; Rotger and Galster 2015; Billings, Deming, and Ross 2016

Source: Authors’ compilation. Note: Only techniques yielding plausibly causal estimates are summarized.

rsf: the russell sage foundation journal of the social sciences 22 spatial foundations of inequality inequality in the United States. We argue, fur- derstanding geographic variation in the joint ther, that a focus on space is crucial to under- distribution of neighborhood race-ethnic com- standing inequality in social and economic sta- position and economic status. But the article tus because, as summarized in figure 1, space also contains substantive conclusions that plays both mediating and modifying roles in stand on their own as meaningful contribu- the relationship between individual attributes tions to our understanding of the way that race and achieved status. and income interact in neighborhoods across The articles in this issue fill in important America. It reinforces findings from other re- gaps in what we know about spatial inequality search showing that black and Hispanic house- and its relationship to other dimensions of the holds live in lower-income communities after U.S. stratification system. Specifically, the con- considering their own household income, and tributions provide theory and evidence on it reveals that black and Hispanic neighbors three questions. tend to have lower incomes than white or Asian First, what are the scale and dimensions of neighbors for all racial groups, regardless of spatial inequality in the United States, and how income. The authors note how this simple does the stratification of space emerge and finding may shed light on the way that within- change? The first article in this issue, from Sean neighborhood racial gaps in income could Reardon, Joseph Townsend, and Lindsay Fox, “play a role in shaping racial stereotypes.” may well become a seminal contribution to The article from Ann Owens moves beyond this question by developing what the authors much of the existing literature on residential refer to as “a general approach to describing segregation by pointing to the unique segrega- the joint distribution of race and income tion of households with children and to the among neighborhoods.” Over time, research- role of schools in contributing to the uneven ers have published research that describes the distribution of racial and ethnic groups. Ow- average neighborhood characteristics of differ- ens links the literature on neighborhood and ent racial and ethnic groups at different levels school segregation by analyzing segregation of income, but these studies have offered within and between school districts, providing piecemeal evidence on the way that race and “evidence on the degree to which school op- ethnicity and economic status overlap across tions, operationalized here as residence in a the nation’s neighborhoods. This article devel- particular school district boundary, contrib- ops a systematic approach to characterizing utes to racial residential segregation.” She the joint distribution of race and income in demonstrates that children live in more un- American communities, and opens the way for equal residential environments than adults, a wide range of different analyses that can offer meaning any consequences of racial segrega- more refined insights into the nature of spatial tion will be more pronounced among young inequality. people. At the same time, the growing diversity The approach Reardon, Townsend, and Fox of the United States population is most pro- develop can be used as a flexible tool to analyze nounced among young people, which means how, for instance, the average racial composi- all children will continue to be exposed to a tion of neighborhoods changes for white, rising share of neighbors from different racial black, Hispanic, or Asian families at the bot- and ethnic backgrounds, particularly Hispan- tom and the top of the income distribution. ics. The authors suggest that the results presented In a particularly novel contribution to the in the article could be used to characterize the literature, Owens provides evidence suggesting way that neighborhood income changes as that school boundaries play an important role household income rises, and how this varies in explaining why households with children across racial and ethnic groups. Applying this are more segregated than those without. Seg- set of methods to counties, metropolitan ar- regation between school districts is higher eas, or commuting zones across the United among children than among adults, indicating States, the results presented in this article can that the school boundary takes on added im- become an extremely valuable resource for un- portance for families with children. To under-

rsf: the russell sage foundation journal of the social sciences an em pirical overview 23 stand neighborhood segregation, Owens ar- neighborhood poverty is based on the image gues, it is crucial to consider the way that of poor, crime-ridden neighborhoods of the parents use the boundaries of school districts deindustrialized cities of the Northeast and when they make decisions about where to live. Midwest, Sampson, Schachner, and Mare draw Rory Kramer’s article shifts from the nation on a unique dataset that has followed families as a whole to a single city, Philadelphia. He fo- and neighborhoods in Los Angeles from 2000 cuses on the way that physical attributes of to 2013, and focus their attention on processes geographic spaces can be used to form clear of change in a city that looks nothing like the boundaries in order to maintain racial segrega- places that are the settings for most of the lit- tion in this highly stratified city that has expe- erature on urban inequality. By following fam- rienced substantial demographic change, ar- ilies over the course of a severe economic guing for a theoretical focus on boundaries as downturn, they shed light on how change a natural means to measure and understand arises, both through residential mobility and the spatial scale of neighborhoods. Kramer as an external shock to stationary families puts forth an innovative method to define and across neighborhoods. measure spatial boundaries, using spatial ana- Although the stability of neighborhood in- lytic tools to observe salient geographic bound- equality in Los Angeles is similar to that in Chi- aries in Philadelphia that serve as markers sep- cago, the traditional laboratory for the study arating communities from each other. of urban poverty, the dynamics leading to ur- Although this method is effective in identi- ban inequality are very different in Los Ange- fying the boundaries that maintain racial seg- les. The most disadvantaged neighborhoods regation, it is less effective in assessing why are occupied by both black and Latino fami- and how those boundaries change (or do not lies, and the rigid boundaries between central change) over time. For this task, local history city and suburbs that characterize cities like is crucial. Kramer goes on to present a histori- Chicago are much less salient in Los Angeles. cal analysis outlining the factors that shape the Residential mobility does not disrupt the rigid rigidity and salience of geographic boundaries racial and ethnic hierarchy of neighborhoods in specific neighborhoods in Philadelphia. He in this city, nor does the shock of the Great argues that the collective response of whites to Recession. Despite differences in the position- the potential for neighborhood racial and eth- ing of groups within this hierarchy, the persis- nic change helps explain why some boundaries tence of urban inequality in Los Angeles bears persist but others fade away. The conclusion to striking resemblance to that in older industrial the article provides a powerful theory linking cities like Chicago. the features of physical space, demographic Continuing the California-based studies, change, local history, and collective action in John Hipp and Charis Kubrin ask how levels an attempt to understand how the stratifica- and changes in the racial, ethnic, and eco- tion of a city’s neighborhoods emerges and nomic composition of the area that surrounds changes over time. a neighborhood—what they call an egohood— Like most of the papers in this volume, both is associated with levels and changes in crime the Kramer and the Owens articles move from within the focal community. The idea and op- descriptive science toward explanation; they erationalization of egohoods is itself an impor- thus make a contribution not only to questions tant contribution, an improvement on the de- of scale and dimensions of spatial inequality fault conception of neighborhoods as distinct but also to our second major question addressed areas separated by an administrative bound- in this issue: what are the processes that generate ary. Hipp and Kubrin consider the unique area and reproduce spatial inequality? Robert Samp- surrounding every block within the city of Los son, Jared Schachner, and Robert Mare address Angeles. It is not just the level of disadvantage this question with an analysis of the mecha- that influences criminal activity, the authors nisms linking individual residential trajecto- argue, but also the mix of people within the ries with aggregate patterns of neighborhood surrounding area, the potential for interac- inequality. Although much of the literature on tions between low-income and high-income

rsf: the russell sage foundation journal of the social sciences 24 spatial foundations of inequality

Angelenos, that predicts the probability of a away from home, engaging in violent or aggres- crime. And it is not simply the mix of people sive behavior, and using marijuana. They find at a given time but also the changes that are that neighborhood racial and ethnic composi- unfolding within and around a community tion, social and economic composition, and that matter when considering the likelihood safety are strongly connected with elevated lev- that crime will begin to rise or fall in a particu- els of these risk behaviors, all of which have lar area. Hipp and Kubrin make several meth- the potential to generate long-term and to have odological advancements in the study of urban severe impacts on young people’s developmen- dynamics and crime, and in doing so generate tal trajectories. new substantive insights into the social pro- The study takes seriously the idea that processes that make criminal activity more or less cesses of sorting into neighborhoods across likely. Their article is likely to become a major Denver work differently for black and Latino contribution to the study of spatial inequality families, even if they are all receiving housing and crime. assistance from the city. By estimating the im- A third question addressed in this issue is how pact of neighborhood conditions separately for space can serve as a mechanism to maintain, re- black and Latino adolescents, the research not inforce, or reproduce inequality. Again, it is a only allows for stronger causal inferences but question addressed, in many ways, by the ar- also provides unique insight into the impact ticles already described, but three additional of neighborhoods for Latino youths, a group studies address it directly. Lincoln Quillian de- not well represented in the literature. The velops a formal model that allows for a more study allows for a more refined look at the spe- refined understanding of the conditions under cific aspects of neighborhoods that are most which the separation of different segments of salient for adolescent risky behaviors, and any two groups, classified by any dimension of does so in a way that generates important in- status or advantage, is likely to result in an am- sights that can be used to guide housing pol- plification or reduction of inequality. One nat- icy. ural application is to understand how changes In the final article, Christopher Browning, in the joint distribution of race and income in Catherine Calder, Lauren Krivo, Anna Smith, the United States has influenced overall racial and Bethany Boettner take a novel approach to inequality in terms of spatial exposure to dis- analyzing neighborhood effect mechanisms by advantaged neighbors. Quillian’s contribution examining the extent to which individuals thus provides a clear theoretical model of how from different social and economic back- racial segregation is related to racial inequality, grounds share space when they carry out rou- and his application to urban areas in the tine activities such as going shopping or going United States is an excellent complement to to work. Using data from Los Angeles, the au- the contribution from Reardon, Townsend, thors find that families with higher social and and Fox. economic status are less likely to share com- Anna Maria Santiago, Eun Lye Lee, Jessica mon spaces with a diverse set of neighbors, Lucero, and Rebecca Wiersma exploit a natural and particularly in highly unequal neighbor- experiment whereby families receiving hous- hoods. The analysis allows one to see beyond ing assistance in Denver were assigned to the socioeconomic composition of neighbor- apartment units through a system that appears hoods to the set of interactions that residents to be close to random. The authors document have with each other. In this way, this article patterns of movement for families in the pro- presents evidence on interaction-based mech- gram, and show persuasively that assignment anisms of neighborhood effects that have been to apartments in different neighborhoods can posited for several decades, yet rarely tested be considered exogenous if it is analyzed empirically. within racial and ethnic groups. They examine Our hope is that this collection of articles how various dimensions of the neighborhood advances the literature on the spatial founda- environment are linked with three risky behav- tions of inequality and generates additional iors during adolescence, including running research taking seriously the idea of space as

rsf: the russell sage foundation journal of the social sciences an em pirical overview 25 a core dimension of stratification in the United sis of Robbery Incidents at Street Corners and States and beyond. We are grateful for the Block Faces in Boston.” Journal of Research in chance to edit such high-quality articles and Crime and Delinquency 48(1): 7–32. to be able to work with this group of impres- Braga, Anthony, Andrew Papachristos, and David sive scholars. Hureau. 2010. “The Concentration and Stability of Gun Violence at Micro Places in Boston, 1980– References 2008.” Journal of Quantitative Criminology 26(1): Aaronson, Daniel. 1998. “Using Sibling Data to Esti- 33–53. mate the Impact of Neighborhoods on Children’s Briggs, Xavier, ed. 1995. The Geography of Opportu- Educational Outcomes.” Journal of Human Re- nity. Washington, D.C.: Brookings Institution sources 33(4): 915–46. Press. Ahern, Jennifer, Sandro Galea, Alan Hubbard, Lor- Briggs, Xavier. 1997. “Moving Up Versus Moving Out: raine Midanik, and S. Leonard Syme. 2008. “‘Cul- Researching and Interpreting Neighborhood Ef- ture of Drinking’ and Individual Problems with fects in Housing Mobility Programs.” Housing Alcohol Use.” American Journal of Epidemiology Policy Debate 8(1): 195–234. 167(9): 1041–49. Briggs, Xavier, Elizabeth Cove, Cynthia Duarte, and Ard, Kerry 2015. “Trends in Exposure to Industrial Air Margery Austin Turner. 2011. “How Does Leaving Toxins for Different Racial and Socioeconomic High-Poverty Neighborhoods Affect the Employ- Groups: A Spatial and Temporal Examination of ment Prospects of Low-Income Mothers and Environmental Inequality in the U.S. from 1995 to Youth?” In Neighborhood and Life Chances: How 2004.” Social Science Research 53: 375–90. Place Matters in Modern America, edited by Har- Åslund, Olaf, and Peter Fredriksson. 2009. “Peer Ef- riet Newburger, Eugenie Birch, and Susan fects in Welfare Dependence: Quasi- Wachter. Philadelphia: University of Pennsylvania Experimental Evidence.” Journal of Human Re- Press. sources 44(3): 798–825. Briggs, Xavier, Kadija Ferryman, Susan Popkin, and Bayer, Patrick, Steven Ross, and George Topa. 2008. Maria Rendon. 2008. “Why Did the Moving to “Place of Work and Place of Residence: Informal Opportunity Experiment Not Get Young People Hiring Networks and Labor Market Outcomes.” into Better Schools?” Housing Policy Debate Journal of Political Economy 116(6): 1150–96. 19(1): 53–91. Becker, Rolf. 2003. “Educational Expansion and Per- Briggs, Xavier, Susan Popkin, and John Goering. sistent Inequalities of Education.” European Soci- 2010. Moving to Opportunity. New York: Oxford ological Review 19(1): 1–24. University Press. Billings, Stephen, David Deming, and Stephen Ross. Brooks-Gunn, Jeanne, Greg J. Duncan, and J. Law- 2016. “Partners in Crime: Schools, Neighbor- rence Aber, eds. 1997. Neighborhood Poverty, vol. hoods and the Formation of Criminal Networks.” 1: Context and Consequences for Children. New NBER working paper no. 21962. Cambridge, York: Russell Sage Foundation. Mass.: National Bureau of Economic Research. Brooks-Gunn, Jeanne, Greg J. Duncan, Pamela Kle- Bischoff, Kendra, and Sean Reardon. 2014. “Residen- banov, and Naomi Sealand. 1993. “Do Neighbor- tial Segregation by Income, 1970–2009.” In Di- hoods Influence Child and Adolescent Develop- versity and Disparities: America Enters a New ment?” American Journal of Sociology 99(2): Century, edited by John R. Logan. New York: 353–95. Russell Sage Foundation. Browning, Christopher, Catherine Calder, Lauren Bolster, Anne, Simon Burgess, Ron Johnston, Kelvyn Krivo, Anna Smith, and Bethany Boettner. 2017. Jones, Carol Propper, and Rebecca Sarker. 2007. “Socioeconomic Segregation of Activity Spaces “Neighbourhoods, Households and Income Dy- in Urban Neighborhoods: Does Shared Residence namics: a Semi-Parametric Investigation of Mean Shared Routines?” RSF: Russell Sage Foun- Neighbourhood Effects.” Journal of Economic Ge- dation Journal of the Social Sciences 3(2): 210–31. ography 7(1): 1–38. doi: 10.7758/RSF.2017.3.2.09. Braga, Anthony, David Hureau, and Andrew Papa- Bryant, Bunyan, and Paul Mohai, eds. 1992. Race and christos. 2011. “The Relevance of Micro Places in the Incidence of Environmental Hazards. Boulder, Citywide Robbery Trends: A Longitudinal Analy- Colo.: Westview Press.

rsf: the russell sage foundation journal of the social sciences 26 spatial foundations of inequality

Burdick-Will, Julia, Jens Ludwig, Stephen Rauden- Clampet-Lundquist, Susan. 2007. “No More ‘Bois bush, Robert Sampson, Lisa Sanbonmatsu, and Ball: The Impact of Relocation from Public Hous- Patrick Sharkey. 2011. “Converging Evidence for ing on Adolescents.” Journal of Adolescent Re- Neighborhood Effects on Children’s Test Scores: search 22(3): 298–323. An Experimental, Quasi-Experimental, and Ob- Clampet-Lundquist, Susan, and Douglas Massey. servational Comparison.” In Whither Opportunity: 2008. “Neighborhood Effects on Economic Self- Rising Inequality, Schools, and Children’s Life Sufficiency: A Reconsideration of the Moving to Chances, edited by Greg J. Duncan and Richard Opportunity Experiment.” American Journal of J. Murnane. New York: Russell Sage Foundation Sociology 114(1): 107–43. / Chicago: Spencer Foundation. Cohen, Deborah, Bonnie Ghosh-Dastidar, Richard Carlson, Deven, and Joshua Cowen. 2015. “Student Scribner, Angela Miu, Molly Scott, Paul Robinson, Neighborhoods, Schools, and Test Score Growth: Thomas Farley, Ricky Bluthenthal, and Didra Evidence from Milwaukee, Wisconsin.” Sociology Brown-Taylor. 2006. “Alcohol Outlets, Gonorrhea, of Education 88(1): 38–55. and the Los Angeles Civil Unrest: a Longitudinal Casciano, Rebecca, and Douglas Massey. 2012. Analysis.” Social Science and Medicine 62(12): “School Context and Educational Outcomes: Re- 3062–71. sults from a Quasi-Experimental Study.” Urban Crowder, Kyle, and Scott South. 2003. “Neighbor- Affairs Review 48(2): 180–204. hood Distress and School Dropout: The Variable Cerdá, Magdalena, Ana Diez-Roux, Eric Tchetgen Significance of Community Context.” Social Sci- Tchetgen, Penny Gordon-Larsen, and Catarina ence Research 32(4): 659–98. Kiefe. 2010. “The Relationship Between Neigh- Crowder, Kyle, and Jay Teachman. 2004. “Do Resi- borhood Poverty and Alcohol Use: Estimation by dential Conditions Explain the Relationship Be- Marginal Structural Models.” Epidemiology 21(4): tween Living Arrangements and Adolescent Be- 482–89. havior?” Journal of Marriage and the Family 66(3): Cerdá, Magdalena, Jeffrey Morenoff, Ben Hansen, 721–38. Kimberly Tessari Hicks, Luis Duque, Alexandra Cutler, David, Edward Glaeser, and Jacob Vigdor. Restrepo, and Ana Diez-Roux. 2012. “Reducing 2008. “When Are Ghettos Bad? Lessons from Violence by Transforming Neighborhoods: A Nat- Immigrant Segregation in the United States.” ural Experiment in Medellin, Colombia.” Ameri- Journal of Urban Economics 63(3): 759–74. can Journal of Epidemiology 175(10): 1045–53. Damm, Anna Piil. 2009. “Ethnic Enclaves and Immi- Chetty, Raj, and Nathaniel Hendren. 2015. “The Ef- grant Labor Market Outcomes: Quasi- fects of Neighborhoods on Children’s Long-Term Experimental Evidence.” Journal of Labor Eco- Outcomes: Quasi-Experimental Estimates for the nomics 27(2): 281–314. United States.” Unpublished paper. Stanford Uni- ———. 2014. “Neighborhood Quality and Labor Mar- versity, Palo Alto, Calif. ket Outcomes: Evidence from a Quasi-Random Chetty, Raj, Nathaniel Hendren, and Lawrence Katz. Neighborhood Assignment of Immigrants.” Jour- 2015. “The Effects of Exposure to Better Neigh- nal of Urban Economics 79(January): 139–66. borhoods on Children: New Evidence from the Damm, Anna Piil, and Christian Dustmann. 2014. Moving to Opportunity Experiment.” NBER work- “Does Growing Up in a High Crime Neighbor- ing paper no. 21156. Cambridge, Mass.: National hood Affect Youth Criminal Behavior?” American Bureau of Economic Research. Economic Review 104(6): 1806–32. Chetty, Raj, Nathaniel Hendren, Patrick Kline, and Dawkins, Casey. Forthcoming. “Putting Equality in Emmanuel Saez. 2014. “Where Is the Land of Place: The Normative Foundations of Geographic Opportunity? The Geography of Intergenera- Equality of Opportunity.” Housing Policy Debate. tional Mobility in the United States.” NBER work- Dawkins, Casey, Qing Shen, and Thomas Sanchez. ing paper no. 19843. Cambridge, Mass.: National 2005. “Race, Space and Unemployment Dura- Bureau of Economic Research. tion.” Journal of Urban Economics 58(1): 91–113. Chyn, Eric. 2016. “Moved to Opportunity: The Long- DeLuca, Stefanie, Greg Duncan, Ruby Mendenhall, Run Effect of Public Housing Demolition on La- and Michere Keels. 2010. “Gautreaux Mothers bor Market Outcomes of Children.” Unpublished and Their Children.” Housing Policy Debate 20(1): paper. University of Michigan, Ann Arbor. 7–25.

rsf: the russell sage foundation journal of the social sciences an em pirical overview 27

Dietz, Robert. 2002. “The Estimation of Neighbor- Federal Bureau of Investigation. 2015. “Uniform hood Effects in the Social Sciences.” Social Sci- Crime Reports, 1960–2012.” Uniform Crime Re- ence Research 31(4): 539–75. porting Data Tool. Washington: U.S. Department Diez-Roux, Ana, and Christina Mair. 2010. “Neigh- of Justice. Accessed June 21, 2016. https://www borhoods and Health.” Annals of the New York .fbi.gov/about-us/cjis/ucr/. Academy of Sciences 1186(1): 125–45. Florida, Richard. 2010. The Great Reset: How New Do, Phuong, Lu Wang, and Michael Elliott. 2013. “In- Ways of Living and Working Drive Post-Crash vestigating the Relationship Between Neighbor- Prosperity. Toronto: Random House Canada. hood Poverty and Mortality Risk: A Marginal Friedson, Michael, and Patrick Sharkey. 2015. “Vio- Structural Modeling Approach.” Social Science lence and Neighborhood Disadvantage after the and Medicine 91(August): 58–66. Crime Decline.” The Annals of the American Downey, Liam. 2007. “US Metropolitan-Area Varia- Academy of Political and Social Science 660(1): tion in Environmental Inequality Outcomes.” Ur- 341–58. ban Studies 44(5–6): 953–77. Galster, George. 2008. “Quantifying the Effect of Downey, Liam, and Brian Dawkins. 2008. “Race In- Neighbourhood on Individuals: Challenges, Alter- come, and Environmental Inequality in the native Approaches and Promising Directions.” United States.” Sociological Perspectives 51(4): Journal of Applied Social Science Studies 759–81. [Schmollers Jahrbuch/ Zeitscrift fur Wirtschafts- Dreier, Peter, John Mollenkopf, and Todd Swanstrom. und Sozialwissenschaften] 128(1): 7–48. 2001. Place Matters: Metropolitics for the Twenty- Galster, George, Roger Andersson, and Sako Mus- First Century. Lawrence: University Press of Kan- terd. 2010. “Who Is Affected by Neighbourhood sas. Income Mix? Gender, Age, Family, Employment, Duncan, Greg J., James Connell, and Patricia Kle- and Income Differences.” Urban Studies 48(13): banov. 1997. “Conceptual and Methodological Is- 2915–44. sues In Estimating Causal Effects of Neighbor- ———. 2015. “Are Males’ Incomes Influenced by the hoods and Family Conditions on Individual Income Mix of Their Male Neighbors? Explora- Development.” In Neighborhood Poverty, vol. 1, tions into Nonlinear and Threshold Effects in Context and Consequences for Children, edited by Stockholm.” Housing Studies 30(2): 315–43. Jeanne Brooks-Gunn, Greg J. Duncan, and J. Law- ———. 2016. “Neighborhood Social Mix and Adults’ rence Aber. New York: Russell Sage Foundation. Income Trajectories: Longitudinal Evidence from Edin, Per-Anders, Peter Fredricksson, and Olof Ås- Stockholm.” Geografisker Annaler, Series B, Hu- lund. 2003. “Ethnic Enclaves and the Economic man Geography. Success of Immigrants: Evidence from a Natural Galster, George, Roger Andersson, Sako Musterd, Experiment.” Quarterly Journal of Economics and Timo Kauppinen. 2008. “Does Neighborhood 118(1): 329–57. Income Mix Affect Earnings of Adults?” Journal Ellen, Ingrid, and Katherine O’Regan. 2009. “Crime of Urban Economics 63(3): 858–70. and U.S. Cities: Recent Patterns and Implica- Galster, George, and Lina Hedman. 2013. “Measur- tions.” Annals of the American Academy of Politi- ing Neighborhood Effects Non-experimentally: cal and Social Science 626(1): 22–38. How Much Do Alternative Methods Matter? Erikson, Robert, and Jan Jonsson. 1996. “Explaining Housing Studies 28(3): 473–98. Class Inequality in Education: The Swedish Test Galster, George, and Sean Killen. 1995. “The Geog- Case.” In Can Education Be Equalized? The raphy of Metropolitan Opportunity: A Reconnais- Swedish Case in Comparative Perspective, edited sance and Conceptual Framework.” Housing Pol- by Robert Erikson and Jan Jonsson. Boulder, icy Debate 6(1): 7–44. Colo.: Westview Press. Galster, George, David Marcotte, Marvin Mandell, Fauth, Rebecca, Tama Leventhal, and Jeanne Hal Wolman, and Nancy Augustine. 2007. “The Brooks-Gunn. 2007. “Welcome to the Neighbor- Impact of Childhood Neighborhood Poverty on hood? Long-Term Impacts of Moving to Low- Young Adult Outcomes.” Housing Studies 22(5): Poverty Neighborhoods on Poor Children’s and 723–52. Adolescents’ Outcomes.” Journal of Research on Galster, George, and Anna Santiago. 2006. “What’s Adolescence 17(2): 249–84. the ‘Hood Got to Do with It? Parental Percep-

rsf: the russell sage foundation journal of the social sciences 28 spatial foundations of inequality

tions About How Neighborhood Mechanisms Af- the Segregated Century: Racial Separation in fect Their Children.” Journal of Urban Affairs America’s Neighborhoods, 1890–2010. New York: 28(3): 201–26. Manhattan Institute for Policy Research. ———. Forthcoming. “Neighborhood Ethnic Compo- Glymour, M. Maria, Mahasin Mujahid, Qiong Wu, sition and Outcomes for Low-Income Latino and Kellee White, and Eric J. Tchetgen Tchetgen. African American Children.” Urban Studies. 2010. “Neighborhood Disadvantage and Self- Galster, George, Anna Santiago, Jackie Cutsinger, Assessed Health, Disability, and Depressive and Lisa Stack. 2016. “Neighborhood Effects on Symptoms: Longitudinal Results from the Health Secondary School Performance of Latino and Af- and Retirement Study.” American Journal of Epi- rican American Youth: Evidence from a Natural demiology 20(11): 856–61. Experiment in Denver.” Journal of Urban Econom- Harding, David. 2003. “Counterfactual Models of ics 93(May): 30–48. Neighborhood Effects: The Effect of Neighbor- Galster, George, Anna Santiago, and Jessica Lucero. hood Poverty on Dropping Out and Teenage 2015a. “Adrift at the Margins of Urban Society: Pregnancy.” American Journal of Sociology 109(3): What Role Does Neighborhood Play?” Urban Af- 676–719. fairs Review 51(1): 10–45. Haveman, Robert, and Barbara Wolfe. 1994. Suc- ———. 2015b. “Employment of Low-Income African ceeding Generations: On the Effects of Invest- American and Latino Teens: Does Neighborhood ments in Children. New York: Russell Sage Foun- Social Mix Matter?” Housing Studies 30(2): 192– dation. 227. Hearst, Mary, Michael Oakes, Pamela Johnson. 2008. Galster, George, Anna Santiago, Jessica Lucero, and “The Effect of Racial Residential Segregation on Jackie Cutsinger. 2015. “Adolescent Neighbor- Black Infant Mortality.” American Journal of Epi- hood Context and Young Adult Economic Out- demiology 168(11): 1247–54. comes for Low-Income African Americans and Hedman, Lina, and George Galster. 2013. “Neigh- Latinos.” Journal of Economic Geography 16(2): borhood Income Sorting and the Effects of 471–503. doi: 10.1093/jeg/lbv004. Neighborhood Income Mix on Income: A Holis- Galster, George, Anna Santiago, and Lisa Stack. tic Empirical Exploration.” Urban Studies 50(1): 2015. “Elementary School Difficulties of Low- 107–27. Income Latino and African American Youth: The Hellerstein, Judith, Mark. Kutzbach, and David Neu- Role of Geographic Context.” Journal of Urban Af- mark. 2014. “Do Labor Market Networks Have an fairs 38(4): 477–502. doi: 10.1111/juaf.12266. Important Spatial Dimension?” Journal of Urban Gennetian, Lisa A., Lisa Sanbonmatsu, and Jens Economics 79(1): 39–58. Ludwig. 2011. “An Overview of Moving to Oppor- Hellerstein, Judith, David Neumark, and Melissa Mc- tunity: A Random Assignment Housing Mobility Inerney. 2008. “Spatial Mismatch or Racial Mis- Study in Five U.S. Cities.” In Neighborhood and match?” Journal of Urban Economics 64(2): 464– Life Chances: How Place Matters in Modern 79. America, edited by Harriet B. Newburger, Euge- Hipp, John, and Charles Kubrin. 2017. “From Bad to nie L. Birch, and Susan M. Wachter. Philadelphia: Worse: How Changing Inequality in Nearby Ar- University of Pennsylvania Press. eas Impacts Local Crime” RSF: Russell Sage Gibbons, Stephen, Olmo Silva, and Felix Weinhardt. Foundation Journal of the Social Sciences 3(2): 2013. “Everybody Needs Good Neighbours? Evi- 129–51. doi: 10.7758/RSF.2017.3.2.06. dence from Students’ Outcomes in England.” The Holzer, Harry J. 1991. “The Spatial Mismatch Hy- Economic Journal 123(571): 831–74. pothesis: What Has the Evidence Shown?” Urban ———. 2014. “Neighbourhood Turnover and Teenage Studies 28(1): 105–22. Achievement.” IZA discussion paper no. 8381. Ihlanfeldt, Keith R., and David L. Sjoquist. 1998. Bonn: Institute for the Study of Labor. “The Spatial Mismatch Hypothesis: A Review of Ginther, Donna, Robert Haveman, and Barbara Recent Studies and Their Implications for Wel- Wolfe. 2000. “Neighborhood Attributes as Deter- fare Reform.” Housing Policy Debate 9(4): 849– minants of Children’s Outcomes.” Journal of Hu- 92. man Resources 35(4): 603–42. Jacob, Brian. 2004. “Public Housing, Housing Glaeser, Edward, and Jacob Vigdor. 2012. The End of Vouchers, and Student Achievement: Evidence

rsf: the russell sage foundation journal of the social sciences an em pirical overview 29

from Public Housing Demolitions in Chicago.” 2014. “Associations of Housing Mobility Interven- American Economic Review 94(1): 233–58. tions for Children in High-Poverty Neighbor- Jargowsky, Paul. 1996. “Take the Money and Run: hoods with Subsequent Mental Disorders During Economic Segregation in U.S. Metropolitan Ar- Adolescence.” Journal of the American Planning eas.” American Sociological Review 61(6): 984– Association 311(9): 937–48. 98. Kling, Jeffrey, Jeffrey Liebman, and Lawrence Katz. ———. 2003. “Stunning Progress, Hidden Problems: 2007. “Experimental Analysis of Neighborhood The Dramatic Decline of Concentrated Poverty in Effects.” Econometrica 75(1): 83–119. the 1990s.” Living Cities Census Series no. 24. Kling, Jeffrey, Jens Ludwig, and Lawrence Katz. Washington, DC: Brookings Institution Press. 2005. “Neighborhood Effects on Crime for Fe- ———. 2015. “The Architecture of Segregation.” New male and Male Youth: Evidence from a Random- York: The Century Foundation. ized Housing Voucher Experiment.” Quarterly Jargowsky, Paul, and Mohamed El Komi. 2011. “Be- Journal of Economics 120(1): 87–131. fore or After the Bell? School Context and Neigh- Kramer, Rory. 2017. “Defensible Spaces in Philadel- borhood Effects on Student Achievement.” In phia: Exploring Neighborhood Boundaries Neighborhood and Life Chances: How Place Mat- Through Spatial Analysis.” RSF: Russell Sage ters in Modern America, edited by Harriet B. Foundation Journal of the Social Sciences 3(2): Newburger, Eugenie L. Birch, and Susan M. 81–101. doi: 10.7758/RSF.2017.3.2.04. Wachter. Philadelphia: University of Pennsylvania Lanphear, Bruce P., Robert S. Byrd, Peggy Auinger, Press. and Stanley J. Schaffer. 1998. “Community Char- Jencks, Cristopher, and Susan Mayer. 1990. “The So- acteristics Associated with Elevated Blood Lead cial Consequences of Growing Up in a Poor Levels in Children.” Pediatrics 101(2): 264–71. Neighborhood.” In Inner-City Poverty in the Leventhal, Tama, and Jeanne Brooks-Gunn. 2003. United States, edited by Lawrence Lynn and Mi- “Moving to Opportunity: An Experimental Study chael MacGeary. Washington, D.C.: National of Neighborhood Effects on Mental Health.” Academy Press. American Journal of Public Health 93(9): 1576– Johnson, Pamela Jo, J. Michael Oakes, and Douglas 82. L. Anderton. 2008. “Neighborhood Poverty and Lindley, Joanne, and Stephen Machin. 2014. “Spatial American Indian Infant Death: Are the Effects Changes in Labour Market Inequality.” Journal of Identifiable?” American Journal of Epidemiology Urban Economics 79: 121–138. 18(7): 552–59. Livingston, Mark, George Galster, Ade Kearns, and Jokela, Markus. 2014. “Are Neighborhood Health As- Jon Bannister. 2014. “Criminal Neighborhoods: sociations Causal? A 10-Year Prospective Cohort Does the Density of Prior Offenders in an Area Study with Repeated Measurements.” American Encourage Others to Commit Crime?” Environ- Journal of Epidemiology 180(8): 776–84. ment and Planning A 46(10): 2469–88. Kain, John F. 1968. “Housing Segregation, Negro Logan, John. 2011. “Separate and Unequal: The Employment, and Metropolitan Decentralization.” Neighborhood Gap for Blacks, Hispanics and Quarterly Journal of Economics 83(2): 175–97. Asians in Metropolitan America.” Project US2010 ———. 1992. “The Spatial Mismatch Hypothesis: report. Providence, R.I.: Brown University. Ac- Three Decades Later.” Housing Policy Debate cessed June 20, 2016. http://www.s4.brown.edu 3(2): 371–460. /us2010/Data/Report/report0727.pdf. Katz, Lawrence, Jeffrey Kling, and Jeffrey Leibman. Logan, John, Deirdre Oakley, Polly Smith, Jacob 2001. “Moving to Opportunity in Boston: Early Stowell, and Brian Stults. 2001. “Separating the Results of a Randomized Mobility Experiment.” Children.” Lewis Mumford Center working paper. Quarterly Journal of Economics 116(2): 607–54. Albany: State University of New York. Kawachi, Ichiro, and Lisa Berkman, eds. 2003. Logan, John, and Brian Stults. 2011. “The Persis- Neighborhoods and Health. New York: Oxford tence of Segregation in the Metropolis: New University Press. Findings from the 2010 Census.” Project US2010 Kessler, Ronald, Greg Duncan, Lisa Gennetian, Law- census brief. Providence, R.I.: Brown University. rence Katz, Jeffrey Kling, Nancy Sampson, Lisa Accessed June 20, 2016. http://www.s4.brown Sanbonmatsu, Alan Zaslavsky, and Jens Ludwig. .edu/us2010/Data/Report/report2.pdf.

rsf: the russell sage foundation journal of the social sciences 30 spatial foundations of inequality

Logan, John, Brian Stults, and Reynolds Farley. 2014. Moulton, Stephanie, Laura Peck, and Keri-Nicole “Segregation of Minorities in the Metropolis: Two Dillman. 2014. “Moving to Opportunity’s Impact Decades of Change.” Demography 41(1): 1–22. on Health and Well-Being Among High-Dosage Lovasi, Gina, James Quinn, Virginia Rauh, Frederica Participants.” Housing Policy Debate 24(2): 415– Perera, Howard Andrews, Robin Garfinkel, Lori 45. Hoepner, Robin Whyatt, and Andrew Rundle. Musterd, Sako, Roger Andersson, George Galster, 2011. “Chlorpyrifos Exposure and Urban Residen- and Timo Kauppinen. 2008. “Are Immigrants’ tial Environment Characteristics as Determinants Earnings Influenced by the Characteristics of of Early Childhood Neurodevelopment.” Ameri- Their Neighbours?” Environment and Planning A can Journal of Public Health 101(1): 63–70. 40(4): 785–805. Ludwig, Jens. 2012. “Moving to Opportunity: Guest Musterd, Sako, George Galster, and Roger Anders- Editor’s Introduction.” Cityscape 14(2): 1–28. son. 2012. “Temporal Dimensions and the Mea- Ludwig, Jens, Greg J. Duncan, and Paul Hirschfield. surement of Neighbourhood Effects.” Environ- 2001. “Urban Poverty and Juvenile Crime: Evi- ment and Planning A 44(3): 605–27. dence from a Randomized Experiment.” Quar- Nandi, Arijit, Thomas Glass, Stephen Cole, Haitao terly Journal of Economics 116(2): 655–79. Chu, Sandro Galea, David Celentano, Gregory Ludwig, Jens, Greg J. Duncan, and Joshua Pinkston. Kirk, David Vlahov, William Latimer, and Shruti 2005. “Neighborhood Effects on Economic Self- Mehta. 2010. “Neighborhood Poverty and Injec- Sufficiency: Evidence from a Randomized tion Cessation in a Sample of Injection Drug Us- Housing-Mobility Experiment.” Journal of Public ers.” American Journal of Epidemiology 171(4): Economics 89(1): 131–56. 391–98. Ludwig, Jens, Helen F. Ladd, and Greg J. Duncan. Nieuwenhuis, Jaap, and Pieter Hooimeijer. 2014. 2001. “Urban Poverty on Educational Outcomes.” “The Association between Neighbourhoods and Brookings-Wharton Papers on Urban Affairs 2: Educational Achievement: A Systematic Review 147–201. Accessed June 20, 2016. http://home and Meta-Analysis.” In Neighbourhood Effects on .uchicago.edu/ludwigj/papers/Ludwig_BWPUA Youth’s Achievements: The Moderating Role of _2001.pdf. Personality, edited by Jaap Nieuwenhuis. PhD Ludwig, Jens, Jeffrey Liebman, Jeffrey Kling, Greg J. diss. Utrecht University. Duncan, Lawrence Katz, Ronald Kessler, and Lisa Novak, Scott, Sean Reardon, Stephen Raudenbush, Sanbonmatsu. 2008. “What Can We Learn and Stephen Buka. 2006. “Retail Tobacco Outlet About Neighborhood Effects from the Moving to Density and Youth Cigarette Smoking: A Opportunity Experiment?” American Journal of Propensity-Modeling Approach.” American Jour- Sociology 114(1): 144–88. nal of Public Health 96(4): 670–76. Ludwig, Jens, Lisa Sanbonmatsu, Lisa Gennetian, Oakes, J. Michael, Kate Andrade, Ifrah Biyoow, and Adam Emma, Greg J. Duncan, and Lawrence Logan Cowan. 2015. “Twenty Years of Neighbor- Katz. 2011. “Neighborhoods, Obesity, and Diabe- hood Effect Research: An Assessment.” Current tes—A Randomized Social Experiment.” New En- Epidemiology Reports 2(1): 1–8. gland Journal of Medicine 365(16): 1509–19. Oreopoulos, Philip. 2003. “The Long-Run Conse- Manski, Charles. 1995. Identification Problems in the quences of Living in a Poor Neighborhood.” Social Sciences Cambridge, Mass.: Harvard Uni- Quarterly Journal of Economics 118(4): 1533–75. versity Press. Orfield, Gary, and Chungmei Lee. 2007. “Historic ———. 2000. “Economic Analysis of Social Interac- Reversals, Accelerating Resegregation, and the tions.” Journal of Economic Perspectives 14(3): Need for New Integration Strategies.” Los Ange- 115–36. les: University of California, Civil Rights Project. McConnell, Rob, Talat Islam, Ketan Shankardass, Orr, Larry, Judith Feins, Robin Jacob, Eric Beecroft, Michael Jerrett, Fred Lurmann, Frank Gilliland, Lisa Sanbonmatsu, Lawrence Katz, Jeffrey Lieb- Jim Gauderman, Ed Avol, Nino Künzli, Ling Yao, man, and Jeffrey Kling. 2003. Moving to Opportu- John Peters, and Kiros Berhane. 2010. “Childhood nity: Interim Impacts Evaluation. Washington: U.S. Incident Asthma and Traffic-Related Air Pollution Department of Housing and Urban Development. at Home and School.” Environmental Health Per- Owens, Ann. 2016. “Inequality in Children’s Con- spectives 118(7): 1021–26. texts: The Economic Segregation of Households

rsf: the russell sage foundation journal of the social sciences an em pirical overview 31

with and Without Children.” American Sociologi- 2015. “Neighborhood Income Composition by cal Review 81(3): 549–74. Race and Income, 1990–2009.” Annals of the ———. 2017. “Racial Residential Segregation of American Academy of Political and Social Science School-Age Children and Adults: The Role of 68(1): 78–97. Schooling as a Segregating Force.” RSF: Russell Reardon, Sean, and Ann Owens. 2014. “60 Years Af- Sage Foundation Journal of the Social Sciences ter Brown: Trends and Consequences of School 3(2): 63–80. doi: 10.7758/RSF.2017.3.2.03. Segregation.” Annual Review of Sociology 40: Owens, Ann, Sean Reardon, and Christopher Jencks. 199–218. 2016. “Income Segregation between Schools and Reardon, Sean, Joseph Townsend, and Lindsay Fox. Districts, 1990 to 2010.” Stanford Center for Edu- 2017. “Characteristics of the Joint Distribution of cation Policy working paper no. 16–04. Palo Alto, Race and Income Among Neighborhoods.” RSF: Calif.: Stanford University. Russell Sage Foundation Journal of the Social Sci- Papachristos, Andrew. 2013. “48 Years of Crime in ences 3(2): 34–62. doi: 10.7758/RSF.2017.3.2.02. Chicago: A Descriptive Analysis of Serious Crime Rosenbaum, James. 1991. “Black Pioneers: Do Moves Trends from 1965 to 2013.” ISPS working paper to the Suburbs Increase Economic Opportunity 13–023. Chicago: University of Chicago. for Mothers and Children?” Housing Policy De- Plotnick, Robert, and Saul Hoffman. 1999. “The Ef- bate 2(4): 1179–213. fect of Neighborhood Characteristics on Young ———. 1995. “Changing the Geography of Opportu- Adult Outcomes: Alternative Estimates.” Social nity by Expanding Residential Choice: Lessons Science Quarterly 80(1): 1–18. from the Gautreaux Program.” Housing Policy De- Popkin, Susan, Tama Leventhal, and Gretchen Weis- bate 6(1): 231–69. mann. 2010. “Girls in the ‘Hood: How Safety Af- Rotger, Gabriel Pons, and George Galster. 2015. fects the Life Chances of Low-Income Girls.” Ur- “Neighborhood Context and Criminal Behaviors ban Affairs Review 45(6): 715–44. of the Disadvantaged: Evidence from a Copenha- Propper, Carol, Simon Burgess, Anne Bolster, George gen Natural Experiment.” Presented at the Euro- Leckie, Kelvyn Jones, and Ron Johnston. 2007. pean Network for Housing Research meetings. “The Impact of Neighbourhood on the Income Lisbon. (June 28, 2015). and Mental Health of British Social Renters.” Ur- Rubinowitz, Leonard, and James Rosenbaum. 2000. ban Studies 44(2): 393–415. Crossing the Class and Color Lines: From Public Quillian, Lincoln. 2017. “Segregation as a Source of Housing to White Suburbia. Chicago: University Contextual Advantage: A Formal Theory with Ap- of Chicago Press. plication to American Cities.” RSF: Russell Sage Sampson, Robert. 2003. “The Neighborhood Con- Foundation Journal of the Social Sciences 3(2): text of Well Being.” Perspectives in Biology and 152–69. doi: 10.7758/RSF.2017.3.2.07. Medicine 46(3): S53–73. Rau, Tomás, Loreto Reyes, and Sergio Urzúa. 2013. ———. 2008. “Moving to Inequality: Neighborhood Ef- “The Long-Term Effects of Early Lead Exposure: fects and Experiments Meet Social Structure.” Evidence from a Case of Environmental Negli- American Journal of Sociology 114(11): 189–231. gence.” NBER working paper no. 18915. Cam- ———. 2012. Great American City: Chicago and the bridge, Mass.: National Bureau of Economic Re- Enduring Neighborhood Effect. Chicago: Univer- search. Accessed June 20, 2016. http://www sity of Chicago Press. .nber.org/papers/w18915. Sampson, Robert, Jared Schachner, and Robert Reardon, Sean, and Kendra Bischoff. 2011. “Income Mare. 2017. “Urban Income Inequality and the Inequality and Income Segregation.” American Great Recession in Sunbelt Form: Disentangling Journal of Sociology 116(4): 1092–153. Individual and Neighborhood-Level Change in ———. 2016. “The Continuing Increase in Income Los Angeles.” RSF: Russell Sage Foundation Jour- Segregation, 2007–2012.” Palo Alto, Calif.: Stan- nal of the Social Sciences 3(2): 102–28. doi: ford Center for Education Policy Analysis. Ac- 10.7758/RSF.2017.3.2.05. cessed June 20, 2016. http://cepa.stanford.edu Sampson, Robert, Patrick Sharkey, and Stephen /content/continuing-increase-income-segrega Raudenbush. 2008. “Durable Effects of Con- tion-2007-2012. centrated Disadvantage on Verbal Ability Reardon, Sean, Lindsay Fox, and Joseph Townsend. Among African-American Children.” Proceed-

rsf: the russell sage foundation journal of the social sciences 32 spatial foundations of inequality

ings of the National Academy of Sciences Homicides on Children’s Cognitive Performance.” 105(3): 845–52. Proceedings of the National Academy of Sciences Sampson, Robert, and Alex Winter. Forthcoming. 107(26): 11733–38. “The Racial Ecology of Lead Poisoning: Toxic In- ———. 2012. “An Alternative Approach to Addressing equality in Chicago, 1995–2013.” Du Bois Review: Selection into and out of Social Settings: Neigh- Social Science Research on Race. borhood Change and African American Chil- Sanbonmatsu, Lisa, Lawrence Katz, Jens Ludwig, dren’s Economic Outcomes.” Sociological Meth- Lisa Gennetian, Greg J. Duncan, Ronald Kessler, ods and Research 41(2): 251–93. Emma Adam, Thomas McDade, and Stacy Tes- ———. 2014. “Spatial Segmentation and the Black sler Lindau. 2011. Moving to Opportunity for Fair Middle Class.” American Journal of Sociology Housing Demonstration Program—Final Impacts 119(4): 903–54. Evaluation. Washington: U.S. Department of Sharkey, Patrick, and Felix Elwert. 2011. “The Legacy Housing and Urban Development. of Disadvantage: Multigenerational Neighbor- Sanbonmatsu, Lisa, Jeffrey Kling, Greg J. Duncan, hood Effects on Cognitive Ability.” American and Jeanne Brooks-Gunn. 2006. “Neighborhoods Journal of Sociology 116(6): 1934–81. and Academic Achievement: Evidence from the Sharkey, Patrick, and Jacob Faber. 2014. “Where, Moving to Opportunity Experiment.” Journal of When, Why, and for Whom Do Residential Con- Human Resources 41(4): 649–91. texts Matter? Moving Away from the Dichoto- Santiago, Anna Maria, George Galster, Jessica Lu- mous Understanding of Neighborhood Effects.” cero, Karen Ishler, Eun Lye Lee, Georgios Kyprio- Annual Review of Sociology 40(1): 559–79. takis, and Lisa Stack. 2014. Opportunity Neigh- Sharkey, Patrick, and Robert Sampson. 2010. “The borhoods for Latino and African American Acute Effect of Local Homicides on Children’s Children. Washington: U.S. Department of Hous- Cognitive Performance.” Proceedings of the ing and Urban Development. National Academy of Sciences 107(26): 11733– Santiago, Anna Maria, Eun Lye Lee, Jessica Lucero, 38. and Rebecca Wiersma. 2017. “How Living in the ———. 2015. “Neighborhood Violence and Cognitive ‘Hood Affects Risky Behaviors Among Latino and Functioning.” In Social Neuroscience: Brain, Mind, African American Youth.” RSF: Russell Sage and Society, edited by Russell Schutt, Larry Seid- Foundation Journal of the Social Sciences 3(2): man, and Matcheri Keshavan. Cambridge, Mass.: 170–209. doi: 10.7758/RSF.2017.3.2.08. Harvard University Press. Sari, Florent. 2012. “Analysis of Neighbourhood Ef- Sharkey, Patrick, Amy Schwartz, Ingrid Ellen, and fects and Work Behaviour: Evidence from Paris.” Johanna Lacoe. 2014. “High Stakes in the Class- Housing Studies 27(1): 45–76. room, High Stakes on the Street: The Effects of Sastry, Narayan. 2012. “Neighborhood Effects on Community Violence on Students’ Standardized Children’s Achievement: A Review of Recent Re- Test Performance.” Sociological Science 1(May): search.” In Oxford Handbook on Child Develop- 199–220. ment and Poverty, edited by Rosalind B. King and Sharkey, Patrick, Nicole Tirado-Strayer, Andrew Pa- Valerie Maholmes. New York: Oxford University pachristos, and C. Cybele Raver. 2012. “The Ef- Press. fect of Local Violence on Children’s Attention Schootman, Mario, Elena Andresen, Fredric Wolin- and Impulse Control.” American Journal of Public sky, Theodore Malmstrom, Philip Miller, and Health 102(12): 2287. Douglas Miller. 2007. “Neighbourhood Environ- Smolensky, Eugene. 2007. “Children in the Vanguard ment and the Incidence of Depressive Symptoms of the U.S. Welfare State.” Journal of Economic Among Middle-Aged African Americans.” Journal Literature 45(4): 1011–23. of Epidemiology and Community Health 61(6): Stretesky, Paul B. 2003. “The Distribution of Air 527–32. Lead Levels Across U.S. Counties: Implications Schwartz, Heather. 2010. Housing Policy Is School for the Production of Racial Inequality.” Sociologi- Policy: Economically Integrative Housing Pro- cal Spectrum 23(1): 91–118. motes Academic Achievement in Montgomery Tach, Laura, Sara Jacoby, Douglas Wiebe, Terry County, MD. New York: Century Foundation. Guerra, and Therese Richmond. 2016. “The Ef- Sharkey, Patrick. 2010. “The Acute Effect of Local fect of Microneighborhood Conditions on Adult

rsf: the russell sage foundation journal of the social sciences an em pirical overview 33

Educational Attainment in a Subsidized Housing fects of Neighborhood Characteristics on the Intervention.” Housing Policy Debate 26(2): 380– Mortality of Black Male Youth: Evidence from 97. Gautreaux, Chicago.” Social Science and Medi- Tienda, Marta. 1991. “Poor People and Poor Places: cine 68(5): 814–23. Deciphering Neighborhood Effects on Poverty Waldinger, Roger. 1996. Still the Promised City? Afri- Outcomes.” In Macro-micro Linkages in Sociol- can Americans and New Immigrants in Postindus- ogy, edited by Joan Huber. Newbury Park, Calif.: trial New York. Cambridge Mass.: Harvard Uni- Sage. versity Press. Turner, Margery Austin, Jennifer Comey, Daniel Watson, Tara. 2009. “Inequality and the Measure- Kuehn, and Austin Nichols. 2012. Residential Mo- ment of Residential Segregation by Income.” Re- bility, High-Opportunity Neighborhoods, and Out- view of Income and Wealth 55(3): 820–44. comes for Low-Income Families: Insights from the Weinberg, Bruce, Patricia Reagan, and Jeffrey Moving to Opportunity Demonstration. Washing- Yankow. 2004. “Do Neighborhoods Affect Work ton: U.S. Department of Housing and Urban De- Behavior? Evidence from the NLSY79.” Journal of velopment. Labor Economics 22(4): 891–924. United Nations Office on Drugs and Crime Weinhardt, Felix. 2014. “Social Housing, Neighbor- (UNODC). 2013. “Global Study on Homicide hood Quality and Student Performance.” Journal 2013.” Vienna: UNODC. of Urban Economics 82(1): 12–31. van Ham, Maarten, and David Manley. 2009. “The Ef- Weisburd, David, Shawn Bushway, Cynthia Lum, fect of Neighbourhood Housing Tenure Mix on and SueMing Yang. 2004. “Trajectories of Crime Labor Market Outcomes: A Longitudinal Per- at Places: A Longitudinal Study of Street Seg- spective.” IZA discussion paper no. 4094. Bonn: ments in the City of Seattle.” Criminology 42(2): Institute for the Study of Labor. 283–322. ———. 2010. “The Effect of Neighborhood Housing Wilson, William Julius. 1987. The Truly Disadvan- Tenure Mix on Labour Market Outcomes: A Lon- taged: The Inner City, the Underclass, and Public gitudinal Investigation of Neighborhood Effects.” Policy. Chicago: University of Chicago Press. Journal of Economic Geography 10(2): 257–82. ———. 1996. When Work Disappears: The World of Vartanian, Thomas, and Philip Gleason. 1999. “Do the New Urban Poor. New York: Alfred A. Knopf. Neighborhood Conditions Affect High School Yagan, Danny. 2016. “The Enduring Employment Im- Dropout and College Graduation Rates?” Journal pact of Your Great Recession Location.” Unpub- of Socio-Economics 28(1): 21–24. lished working paper. Department of Economics, Votruba, Mark Edward, and Jeffrey Kling. 2009. “Ef- University of California, Berkeley.

rsf: the russell sage foundation journal of the social sciences