How Place Matters
The workshop on equality of opportunity featured three papers that analyzed recently collected data on spatial mismatch and neighborhood effects in metropolitan areas. This research sheds light on specific factors and mechanisms that create barriers to opportunity for inner-city residents. From this research, policy makers and coordinators of intervention programming can begin to think more specifically about how to capitalize on this knowledge to overcome spatial and place-based barriers. The papers explored three important areas affecting many poor inner-city residents: the effects of neighborhood on child development, transitions from welfare to work, and health outcomes. Presenters sought to capture current trends in inner-city neighborhoods; identify the particular mechanisms of observed neighborhood effects; and discuss how to overcome barriers to child and adolescent development, employment, and health and well-being. The content of these papers and the discussion the papers generated are summarized in this chapter.
NEIGHBORHOOD EFFECTS ON CHILD AND ADOLESCENT DEVELOPMENT
In the past two decades, scholars and policy makers have become increasingly interested in the effects that neighborhood may have on children and adolescents. Significant demographic shifts, changes in labor force participation, and new theoretical perspectives about social disorganization
and ecological models of human development all contributed to the growing concern that neighborhood could be an important influence on whether young people transition into adulthood with the skills needed to succeed. At the workshop, Tama Leventhal presented a comprehensive analysis of the relationship between neighborhood and child development. Her paper included a review of research published from 1990 to 1998, an analysis of data from recent studies, and a discussion of theoretical mechanisms that may be most salient for children and adolescents. The following section highlights the key issues Leventhal identified in each of these areas.
Neighborhood Effects and Development: A Review of Eight Years of Research
In her review of research undertaken in the 1990s on neighborhood effects, Leventhal examined databases from a wide variety of disciplines, including psychology, sociology, demography, economics, and epidemiology. She included only those studies and datasets that controlled for family and background characteristics in order to show effects above and beyond family socioeconomic factors. Three important issues set the framework for Leventhal’s approach to these data. First, the review was framed in terms of the three structural dimensions most frequently used to classify neighborhoods in order to understand the direction and power of neighborhood effects: socioeconomic status (SES), racial and ethnic diversity, and residential instability. Each factor was measured through census tract data. As a result, these neighborhood descriptors are just that—descriptions of structural factors in a neighborhood—and do not illuminate underlying social processes.
Definitions of these dimensions differed somewhat across studies, but the following descriptions are applicable to most. Low SES is typically defined as a given percentage of the population with household incomes below the poverty level, the percentage of residents on public assistance, the unemployment rate of a neighborhood, or sometimes the percentage of single parents. Often definitions of low SES will use more than one of these descriptive measures. High SES usually includes mean or median income, percentage of working professionals, and percentage of residents with a college education.
Measures of racial and ethnic diversity were even more consistent across the studies Leventhal examined and were generally characterized by the percentage of African American, Latino, and foreign-born residents in a
neighborhood. Definitions of residential instability often included the proportion of residents who moved within the past 5 years, the proportion of households whose residents lived in their current homes for less than 10 years, and the proportion of owner-occupied homes.
With these structural definitions framing her review of the research, Leventhal chose to stratify the data into four age periods in order to determine if there were critical ages at which young people were more vulnerable to neighborhood influences. The age periods were early childhood (birth to age 6), late childhood (ages 7 to 10), early adolescence (ages 11 to 15), and late adolescence (ages 16 to 19). Finally, Leventhal paid particular attention to three developmental outcomes as measures of well-being in these age groups: school readiness and achievement; manifestation of behavioral, social, and emotional problems; and sexuality and fertility outcomes.
Leventhal’s comprehensive review of data revealed several consistent relationships between neighborhood structural factors and measures of well-being, each suggesting that neighborhood effects are an active influence on child and adolescent development. High neighborhood SES was positively associated with young children’s school readiness outcomes as well as adolescents’ educational achievement. Low neighborhood SES was associated with an increase in the number of younger children exhibiting behavioral problems. Similarly, adolescents with problem behaviors and delinquency increased in frequency with low SES and residential instability. Additionally, some evidence suggested an association between employment indicators and adolescents’ sexual activity and fertility outcomes, but these associations were less consistent.
Leventhal’s review suggests that some neighborhood effects are more important at certain ages, with some outcomes likely to be more strongly affected than others. Furthermore, although there is a strong body of knowledge on structural neighborhood factors affecting younger children and older adolescents, data on children in late childhood and early adolescence are lacking, and this poses a significant barrier to understanding how neighborhood effects influence child development. Leventhal also noted that research has yet to be thoroughly developed that specifically explores whether neighborhood effects are more powerful during a critical period in a child’s development and, if so, which factors are the most important and when.
New Data on Neighborhoods and Child Development
Despite an overall lack of information on the timing of neighborhood effects, there are a few recent studies in which the topic has been explored. This section reviews the results of Leventhal’s analysis of data from the Infant Health and Development Program and the Moving to Opportunity (MTO) experiment.1
Leventhal took an exploratory look at developmental timing and the magnitude of effects by analyzing data from the Infant Health and Development Program—a randomized trial of an early childhood educational intervention program for low-birthweight babies in eight cities. Specifically, she evaluated whether the effects of neighborhood low income during early childhood appear to be more significantly detrimental than if neighborhood low income is experienced later. She found that living in a low-income neighborhood at age 5 was the most detrimental compared to other ages. This negative effect was manifested in a nine-point decrement in children’s IQ scores. By comparison, neighborhood poverty at birth or age 8 was not associated with a decrease in IQ. Although Leventhal emphasized that her findings represented exploratory research, the results corresponded to what she found in her review of the literature: developmental timing and the magnitude of neighborhood effects are potentially important and fruitful areas for future research.
Leventhal’s analysis of the MTO study centered on observing the effect that neighborhood had on the developmental outcomes of young people. The MTO study included 4,600 families from five different cities (Baltimore, Boston, Chicago, Los Angeles, and New York), all of whom lived in public housing developments in neighborhoods classified as high poverty (i.e., 40 percent or more of residents earned wages below the poverty level or were on public assistance). MTO was a voluntary housing experiment, and most families (75 percent) that elected to participate cited getting away
Brooks-Gunn, Duncan, Klebanov, & Sealand (1993) and Chase-Lansdale, Gordon, Brooks-Gunn, & Klebanov (1997) offer useful background on the Infant Health and Development Project. A list of readings about the Moving to Opportunity study can be found on the MTO quick document access website, http://www.wws.princeton.edu/~kling/mto/quick.html [viewed online May 13, 2002].
from drugs and gangs as the primary reason for joining the study. Families were placed randomly into three groups: (1) the experimental group, which received Section 8 vouchers, counseling, and assistance from local nonprofits to move into private housing in low-poverty neighborhoods (i.e., neighborhoods with 10 percent or less poverty); (2) the Section 8 comparison group, which received vouchers under the regular Section 8 program but no special assistance in finding new housing; and (3) the in-place control group, which did not receive vouchers to allow them to relocate or for support but which continued to receive normal subsidies to remain in public housing.
Approximately half of the experimental group in the MTO study relocated to housing in low-poverty areas. Of the group that moved, 90 percent relocated to neighborhoods with less than 10 percent of the population living in poverty, and 9.5 percent relocated to neighborhoods in which 10 to 39 percent of the residents were poor. While this was a most encouraging outcome, only 12 percent of the Section 8 group moved to low-poverty neighborhoods. Seventy percent of this group moved to a neighborhood with a moderate rate of poverty (10 to 39 percent), and 18 percent remained in high-poverty neighborhoods. This suggests that, if housing programs seek to reduce the number of public assistance recipients living in high- and moderate-poverty neighborhoods, the programs will need to provide more support than just vouchers to facilitate a move to a low-poverty neighborhood.
That the experimental group did successfully move to low-poverty neighborhoods meant that Leventhal and her colleagues could compare individual developmental outcomes for families living in a variety of neighborhoods. Across several site-specific evaluations, differences were observed for a number of measures of well-being, including academic attainment, behavioral and emotional problems, and health outcomes. The results suggested that moving to low-poverty neighborhoods may be more beneficial for younger children than for adolescents. This may be due to the fact that disrupting the peer network of an adolescent is more problematic because of the important social role that peers play in adolescents’ lives. Younger children tend to have less established social groups and are less affected by a move—they simply form new friendships at their new locations, and those new relationships are what influence them later in adolescence. As a result, younger children may benefit from new opportunities and positive neighborhood effects that are available in low-poverty neighborhoods, whereas adolescents experience such a move as primarily disruptive to their social
networks, a negative effect that seems to override some of the benefits offered by the new neighborhood. A summary of the specific results of the MTO study follows.2
In Baltimore, Ludwig, Duncan, and Ladd (2001) found that younger children ages 5 to 11 at random assignment in the experimental group were more likely to pass state-required reading exams than those in an in-place control group that remained in high-poverty neighborhoods. This complements Leventhal’s results derived from Infant Health and Development Program data in that both studies suggest that neighborhood might affect educational outcomes. In Leventhal’s New York MTO evaluation, younger children (ages 8 to 13) who moved to low-poverty neighborhoods experienced fewer problems with anxiety and depression. There were no significant differences for older children (ages 14 to 18). The New York evaluation also found an effect for gender: boys in both the experimental and the Section 8 groups reported significantly fewer dependency problems (e.g., clinging to adults, crying too much, excessive demands for attention) compared to in-place controls. Furthermore, these effects were quite powerful, with experimental and Section 8 boys experiencing about a one-third reduction in such symptoms compared to controls. The Boston evaluation by Katz, Kling, and Liebman (2001) also observed that boys in both the experimental and the Section 8 groups had fewer maternal-reported behavioral problems than the in-place controls.
Leventhal speculated that boys may benefit more than girls from a move because parents tend to let boys have more access than girls to neighborhoods (and thus neighborhood influences). Concerns about safety may encourage parents to keep their girls “closer to home,” and as a result girls may be less affected by neighborhood influences. In comparison, boys who were exposed to gangs and violence prior to a move should, in theory, reap greater benefits from living in a better neighborhood.
Children in Boston’s experimental group experienced another benefit of moving to a low-poverty neighborhood—namely, improved health outcomes compared to in-place controls. Children ages 6 to 15 were less likely to have an injury, accident, or asthma attack requiring medical assistance. This neighborhood effect could be due to safer and improved home envi
ronments as a result of moving out of public housing. The reduction in asthma attacks could also be attributed to reduced chronic stress—an issue often associated with living in high-poverty and high-crime neighborhoods.
Mechanisms of Neighborhood Effects
Although the MTO study that Leventhal discussed provided powerful evidence of the link between neighborhood effects and particular outcomes, such studies do not illuminate the underlying mechanisms that may transmit neighborhood effects. This situation is problematic because intervention programs are likely to be much more effective if a specific mechanism can be targeted, rather than treating a broad array of correlated factors that appear to be linked to certain outcomes. One straightforward example comes from the Boston MTO evaluation, which found that asthma attacks decreased when children moved to low-poverty neighborhoods. A suspected mechanism by which this occurs is a reduction in chronic stress caused by fear of crime, violence, and harassment that young people might experience regularly in a high-poverty, high-crime neighborhood. If this is the case, comprehensive programs that seek to reduce crime in high-poverty neighborhoods would be an appropriate use of public funds. However, it is also possible that children’s asthma attacks could be caused by an environmental stimulus, such as dust and pollution from demolition, construction, landfills, or other industrial waste. If many high-poverty neighborhoods in Boston were in close proximity—closer than low-poverty neighborhoods—to such environmental hazards, it could be that this environmental contaminant was the underlying cause of increased asthma attacks. This scenario would require a dramatically different use of public funds, and only research that identifies specific mechanisms can best direct these types of expenditure.
Leventhal discussed three possible mechanisms by which neighborhood effects could be explained—namely, institutional resources, relationships, and norms and collective efficacy. A more lengthy discussion of these mechanisms is presented in Chapter 2 in the section on neighborhood effects research. Very briefly, the institutional resource mechanism posits that neighborhood influences are transmitted through the quality, quantity, and diversity of community resources. The relationship mechanism focuses on the potential for neighborhood effects to be translated to young people through their relationships with other residents, including parents, peers, and adult neighbors. Finally, the norms and collective efficacy mecha
nism suggests that neighborhood effects can be explained by the extent of formal and informal community institutions that are present to monitor residents’ behavior and physical threats to them.
Three points should be kept in mind during this discussion. First, it is generally believed that neighborhood effects influence young people indirectly through family, peers, and institutions such as schools. Second, the relationship mechanism is thought to be the most important for understanding how neighborhoods affect young children who, in theory, should be more isolated from and therefore least affected by community influences. Parental attributes, social networks, behavior characteristics, and quality of the home environment can serve as a buffer to negative neighborhood effects or can act to transmit this influence to young people who might not otherwise be impacted. Finally, the utility of these models depends on the particular outcome under investigation as well as the age of the individual for whom the outcome is expected. In other words, no singular mechanism will explain all outcomes, and the mechanism that most powerfully transmits a neighborhood effect will depend on the child’s age.
Leventhal’s careful examination of the way in which neighborhood could affect child and adolescent development touched on a number of important points. Her review of data and research conducted during the 1990s clearly suggested that neighborhoods do influence development, that the outcomes influenced vary depending on the particular resources of the community, and that there may be critical ages when neighborhood effects may be especially detrimental or nurturing to children’s development. Her analysis of data from the Infant Health and Development Program and the MTO study provided preliminary empirical results that support previous findings that neighborhoods do matter to young people. Neighborhood effects may touch a broad range of areas of young people’s well-being, including educational performance, social and behavioral well-being, and health outcomes. Finally, Leventhal presented three important mechanisms through which neighborhood effects are transmitted to young people. Many of the mechanisms and their outcomes for children of various age groups need to be explored in future research. Such scholarship has the potential to assist in the development of effective intervention programs and the direction of public expenditures.
Workshop participants who commented on Leventhal’s paper noted
several important caveats to remember when considering these results. First, it is significant that only half of the participants in the experimental group of the MTO study moved. This raises questions about whether those individuals shared some common trait that explains positive outcomes of moving rather than actual neighborhood effects due to living in a low-poverty neighborhood. Second, Xavier Briggs noted in submitted comments that research studies such as Leventhal’s make it difficult to conceptualize neighborhoods as dynamic, fluid, churning spaces rather than static concepts with clear boundaries. Future research would do well to find ways to address this issue. Finally, results from the MTO study are preliminary, so initially observed phenomena may not be as significant at the end of the five-year project period.
SPATIAL BARRIERS TO TRANSITIONS FROM WELFARE TO WORK
In 1996 Congress passed the Personal Responsibility and Work Opportunity Reconciliation Act. This act places a 60-month lifetime limit on federal cash welfare benefits and mandates that participants meet work requirements after receiving two years of support. Failure to meet this requirement could result in the loss of cash benefits as well as services (e.g., subsidized child care) and the right to participate in parallel programs such as Medicaid. The first families came to their five-year lifetime limit on cash subsidies in 2001.3
Research on welfare-to-work programs has been encouraging in that many programs have improved employment rates and earnings. However, there are still opportunities to further improve transitions to work in terms of earnings and stable employment. Spatial barriers can present important challenges to these programs and, if not addressed, may reduce the effectiveness of the programs. Research on spatial mismatch can be used to identify instances in which geographic isolation may be a factor in the success or failure of a welfare-to-work program in a specific city, and this knowledge can be used to develop strategies that could reduce the impact of mismatch on welfare-to-work transitions.
Claudia Coulton engaged in this type of research, examining how place
Information from the Children’s Defense Fund, http://libertynet.org/~edcivic/welfcdf.html [viewed online January 3, 2002].
influences the success or failure of welfare-to-work programs by exploring newly collected data from the Project on Devolution and Urban Change conducted by the Manpower Demonstration Research Corporation. This study began in 1997, prior to key changes in welfare legislation, and includes neighborhoods from Cleveland, Los Angeles, Miami, and Philadelphia. Waves of data collection are planned through 2003 in order to observe the results of welfare reform with respect to changes in welfare caseloads and employment. The study was designed to reflect concerns about place-based inequalities as they affect welfare dependency and, as Coulton (2001) stated, “is unique among studies of welfare reform because of its focus on urban communities with high concentrations of welfare, its expansive data collection, and its integrated and multidisciplinary approach” (p. 46).
Survey data about employment and earnings in 1998 were gathered from female-headed families that had been on welfare during 1995. An additional step of geocoding the residence of each participant (i.e., linking a participant’s data to the zip code in which he or she resided) in 1995 and 1998 was taken, and social and economic indicators were analyzed from census tract data for all neighborhoods in which participants lived. This process was used to facilitate an analysis of place-based barriers and neighborhood effects on participants.
Coulton centered her analysis on four metropolitan inequalities: concentration and isolation of welfare recipients, neighborhood effects on work, barriers to job access for low-skilled and inner-city workers, and residential mobility and neighborhood change. What follows is a summary of her analysis.
Concentration of Poverty and Isolation of Welfare Recipients
Concentrated poverty and isolation of welfare recipients were Coulton’s points of departure for examining place-based inequities. As established in the overview of previous work on inner-city neighborhoods, concentrated poverty is an important condition—perhaps a foundational one—for the decline of neighborhoods and fodder for the proliferation of neighborhood conditions that have negative effects on residents. When poor neighborhoods are geographically isolated in regions of a city that lack public transportation and are distant from areas of job growth, the neighborhood effects on residents can be particularly troubling.
Previous research has certainly documented the increasing propensity
of poor residents to cluster in specific areas in inner cities and has correlated a concentration and isolation of the urban poor with negative neighborhood effects. Cities, however, have undergone significant changes in the past decade. Because of this shift, Coulton began her analysis by examining data from the first two years of the Project on Devolution and Urban Change study to determine the extent to which welfare recipients were concentrated in certain inner-city neighborhoods. She found that in 1995 welfare recipients in Cleveland and Philadelphia were highly concentrated in certain neighborhoods as measured on three different indices. In comparison, Miami and Los Angeles were less highly concentrated, with poor and affluent families spread somewhat more evenly throughout the city and with fewer neighborhoods in which welfare recipients exceeded 20 percent of the population.
By 1998 the concentration of welfare recipients in poor areas had fallen in all four cities. This change was due to the drop that occurred by 1998 in the overall number of welfare cases. As a result, many neighborhoods that had exceeded a 20 percent threshold of welfare families living in the area now fell below this mark. Coulton explained that to the extent welfare dependency influences residents and neighborhoods through a threshold effect—in this case the threshold was a 20 percent or more concentration of welfare recipients in a neighborhood—fewer families in 1998 were exposed to this threshold. In other words, if neighborhood effects that perpetuate joblessness or other negative social traits are a significant influence only if a large number of residents are on welfare, the fact that fewer neighborhoods actually reached this 20 percent threshold suggests an overall improvement in the neighborhoods in which welfare recipients were living. Despite this encouraging trend of deconcentration, Coulton found that families that remained welfare dependent were still more likely to live in poverty-concentrated neighborhoods.
In addition to tracking the extent to which welfare recipients were concentrated in certain neighborhoods, Coulton explored whether welfare recipients were also living in “disparate” neighborhoods. Coulton defined disparate neighborhoods as those that not only were poor but also had high rates of child maltreatment, births to unmarried females, births to girls ages 10 to 17, high rates of violent crime, and low median values of single-family homes. Using instruments to measure the rates of each of these five categories, Coulton labeled a neighborhood as disparate if the rates for these indicators measured more than twice the region’s median.
Disparate census tracts were found in all four cities and represented the
following percentages of all neighborhoods in each city: 19 percent of Cleveland neighborhoods were disparate, 2 percent in Los Angeles, 10 percent in Miami, and 4 percent in Philadelphia. Coulton noted that the rate in Philadelphia was artificially low because the city’s regional median value of disparity was quite high. This was one drawback to the methodology Coulton used to identify disparate neighborhoods: because each neighborhood was compared to a median rate of disparity for the region, cities like Philadelphia in which so many neighborhoods were poor and had high levels of social disorder did not calculate as disparate because their rates were not twice as high as the regional median. As a result, neighborhoods in Philadelphia that would have registered as disparate in another city did not score as such because they were being compared to an area in which poverty and disparity were widespread.
One encouraging trend for all cities was that the number of tracts disparate on all five indicators fell from the medians calculated for the 1992-1995 period versus the 1996-1999 period, suggesting an improvement in the quality of many neighborhoods during this time. The number of welfare recipients living in disparate neighborhoods also decreased during this time: Cleveland dropped to 34.5 percent, Miami to 10.7, and Los Angeles to 0.2 (Philadelphia could not be characterized in this way).
Finally, Coulton examined the extent to which high concentrations of welfare recipients in a census tract overlapped with tracts that were disparate. Her argument was that high overlap suggests that a concentration of welfare recipients may drive other disparities in inner-city neighborhoods. The results in Cleveland were particularly telling: although the numbers of high-welfare tracts and disparate tracts decreased, the number of tracts that were both high welfare and disparate increased. This finding suggests that, even though welfare caseloads are declining, the people remaining on public assistance may be living in worse circumstances now than before welfare reform.
It should be noted that this pattern was not seen in every city: Miami’s remaining welfare tracts were less likely to also be disparate, suggesting that neighborhood conditions have improved for welfare recipients there. Sue Popkin noted that there is an alternative interpretation: Florida initiated time limits on welfare recipiency before other states. As a result, the number of welfare recipients may be artificially low not because people have made successful transitions to work but because they are simply no longer on the welfare rolls.
Coulton’s assessment of the overall status of inner-city neighborhoods
is one of positive trends: disparate neighborhoods in all four cities—despite the broad geographic, historical, and economic diversity that characterizes each—are improving. This general positive trend is not, however, without a disquieting caveat. As Coulton stated, in cities such as Cleveland “in which welfare families were highly concentrated and segregated before welfare reform, the remaining high-welfare neighborhoods have become even worse off” (p. 54). Should the economy remain in its recent downward trajectory, one concern is that the concentration of welfare recipients in disparate neighborhoods may be foreshadowing a problematic future trend.
Residential Movement and Neighborhood Churning
One straightforward manner in which concentrated poverty can be reduced is to create opportunities for welfare recipients or the working poor to move to better neighborhoods. For individuals receiving housing subsidies and other supplemental benefits, public policy and programming such as Section 8 mobility programs can help facilitate a move. Programs like Section 8 are tenant based in that they provide a rent subsidy to low-income families. This subsidy allows low-income families to afford private-market housing in a location determined by the families to be the most beneficial to them. This is very different from government-operated public housing.
Unfortunately, local neighborhood politics, zoning restrictions, and the perception that subsidized and public housing is damaging to a neighborhood often prevent these types of public policy and housing programs from being implemented. It is not uncommon for residents in “good” neighborhoods to resist the placement of subsidized housing in their communities, thus closing down one avenue for desegregating neighborhoods. Furthermore, a strategy of relocating welfare recipients to less disparate or less welfare-concentrated neighborhoods has a very important limit: it will never be feasible, nor perhaps desirable, to relocate entire neighborhoods of people in order to disperse a disparate census tract. Thus, approaches that facilitate welfare recipients moving to a new neighborhood can be helpful to a limited number of people, and although it is one path to ease concentrated neighborhood poverty or disparity, it may have a relatively limited scope. Finally, previous research suggests the demographic profile of women on welfare is that of a less mobile population. African Americans are also less likely to be able to move out of poor neighborhoods.
The second piece of Coulton’s research explored the extent to which
residential mobility is used by families as a strategy to improve their situation and the number of opportunities available to them. The presence of residential mobility or churning in a neighborhood may indicate that residents move to better neighborhoods or closer to jobs when they have the opportunity to do so. On the other hand, frequent residential changes—especially if residents are not moving closer to tangible opportunities—may make it more difficult for community networks to develop and for individuals to tap into neighborhood resources.
Coulton found considerable residential mobility among participants in the Urban Change study, with 61 percent of Cleveland residents, 61 percent of Los Angeles residents, 39 percent of Miami residents, and 52 percent of Philadelphia residents having made at least one move to a new census tract between 1995 and 1998-1999. Most of the moves were in response to a crisis or to seek less expensive housing. Only 10 percent moved because they wished to live in a better neighborhood and less than 1 percent moved to be closer to jobs or child care. The majority of movers were not consciously using residential mobility as a strategy to improve job or neighborhood-related opportunities.
Two other insights developed from these data. The first was that movers and nonmovers differed on a number of characteristics. Not only were movers younger, were more likely to be married, had younger children, were less likely to live in public housing, and were more likely to have a housing subsidy, they also were more likely to perceive their neighborhoods favorably and to view destination neighborhoods as an improvement over their current location. Movers seemed to be a somewhat more resilient group who began in and later moved to better neighborhoods than their nonmoving counterparts.
The second insight was that, even though movers seemed to fare somewhat better than nonmovers, changing city conditions were a more powerful factor in predicting whether welfare recipients actually lived and moved into areas that were not distressed. Coulton offered the following example and explanation: “In Miami, both movers and non-movers experienced positive neighborhood change. In Philadelphia, it appears that the trends were negative for both movers and non-movers. Thus, it appears that without special mobility policies, most welfare families do not get out of distressed neighborhoods on their own even though those who do move begin in somewhat more advantaged places and experience greater neighborhood satisfaction” (p. 56).
Another study of Cleveland residents transitioning from welfare to
work found similar results. This study found that while almost one-quarter of individuals exiting welfare moved within the first six months of getting off welfare, most moved between inner-city neighborhoods rather than to high-job-growth suburban locations. As many increased as decreased their commute distances by moving. This also suggests that residential mobility was not deliberately used to improve job access.
Although residential mobility has the potential to assist welfare recipients in relocating to neighborhoods with greater opportunities, on the whole this does not seem to be occurring. Rather, residents often move reactively to deal with rising housing costs or a crisis and not to improve access to jobs or other services that could promote a better quality of life.
Neighborhood Effects on Welfare-to-Work Success
Many scholars have speculated that neighborhood effects could be influential in successful transitions from welfare to work. Because data from the Project on Devolution and Urban Change study indicated that welfare recipients were concentrated in poor neighborhoods and that many of those neighborhoods displayed significant social and economic disparities, it was possible to test whether neighborhood exerted an independent (in this case negative) effect on employment. Through participant interviews, data on two measures of employment—the number of months employed during the previous 12 months and the participant’s earnings in the month prior to the interview—were collected to measure the extent to which families were finding work, maintaining employment, and earning a wage that would enable them to be financially self-sufficient.
Coulton attempted to test whether violent crime, economic disinvestments, and welfare dependency were factors through which a neighborhood might exert an influence on individual employment through modeling techniques. Rates of violent crime were chosen because earlier studies of women on welfare suggested that fear of victimization could be a barrier to work. Declining housing values represented whether a neighborhood was developing economically and, by extension, the extent to which socioeconomic status would be influential on residents. It also represented the extent to which residents of these neighborhoods might be perceived negatively by employers, who would then be discouraged from hiring them as workers. The concentration of families on welfare represented the extent to which neighborhoods lacked social networks (e.g., informal job information) that could support and facilitate work.
After controlling for individual and family differences to the extent possible, significant effects were found for several of the factors tested. Violent crime and disinvestments had a significant effect on the number of months worked, and welfare concentration and disinvestments had significant effects on earnings. Neighborhood effects had a significant influence and appeared to create barriers to finding and maintaining well-paying positions. Despite this outcome, Coulton cautioned that in studies like the Project on Devolution and Urban Change, neighborhood effects could be inflated by individual characteristics that were not controlled. This fact is generally what is meant when researchers talk about neighborhoods as endogenous: residents may choose a neighborhood because they share similar individual attributes unbeknownst to the researcher. As a result, observed neighborhood effects may actually be the result of these shared and unmeasured characteristics.
Researchers have developed a strategy to deal with the endogeneity of neighborhoods and that is to create social experiments in which families are relocated to neighborhoods they would not have had the means to choose themselves so that the results can be observed. Essentially, a treatment group that received a housing subsidy and assistance in relocating to a middle-class suburban neighborhood and a control group that received a subsidy but remained in low-income inner-city neighborhoods would be created. Studies such as the MTO experiment and naturally occurring quasi-experimental programs such as the Gautreaux Program have yielded varying results regarding the extent to which neighborhoods exert significant influences on residents.
The Gautreaux Program began in Chicago in the 1970s and through a lottery process moved some African American families on public assistance to public housing in white suburban neighborhoods. A significant difference in employment rates was found for suburban residents compared to those living in the city—suburban movers had a 16 percentage point higher rate of employment than city dwellers. Children of suburban movers also appeared to benefit from the move. However, suburban movers did not appear to fare better with respect to hours worked or wages.
Interestingly, the MTO study did not replicate these results. Employment and welfare outcomes for the treatment group that moved to a low-poverty suburban neighborhood were not significantly different from those of a randomized Section 8 comparison group. A Baltimore treatment group did show higher rates of moving from welfare to work, although Baltimore was the only city in which this effect was significant. Differences between
the Gautreax Program and MTO could be due to the longer follow-up period of Gautreaux and the fact that data from the MTO employment outcomes were being collected at a time of high labor demand and economic growth, thus improving the employment prospects of all participants.4
The power of neighborhood effects remains an area of exploration, as do the particular variables (such as a booming economy) that may intervene. Future studies that involve the kind of social experiment produced in the Gautreaux Program and MTO would be an invaluable resource for researchers and potentially participants.
Spatial Constraints to Job Access
In addition to the Project on Devolution and Urban Change study, Coulton analyzed data from a longitudinal study of families leaving welfare in the Cleveland metropolitan area. This study offered the opportunity to evaluate the extent to which spatial mismatch created significant barriers for families leaving welfare. Two data sources were available in this study. The first was a survey of adults whose welfare cases had been closed for at least two consecutive months. The survey documented their employment experiences, the job and residential locations of participants six months after leaving welfare, and the racial composition of the neighborhoods in which participants lived and worked. The second dataset was a database of entry-level job openings in the greater Cleveland metropolitan area used to estimate the number of entry-level positions available in various neighborhoods. In addition, a method was developed to estimate travel time by public transportation or car between various census tracts as a way to gauge the accessibility of high-growth job areas to the residential locations of welfare leavers.
Coulton found that, although there were pockets of job opportunities in the inner city, 83 percent of low-skilled job openings occurred in the outlying suburbs of Cleveland. In contrast, 75 percent of welfare leavers lived in the inner city and the rest lived in inner-ring suburbs. Clearly, there was a geographic mismatch between the location of jobs and the residences
Additional information on the MTO study and the Gautreaux Program can be found at http://www.mtoresearch.org [viewed online May 31, 2002].
of this pool of potential workers. In Cleveland the spatial mismatch was compounded by fairly substantial challenges in getting workers to outlying suburbs: inner-city welfare leavers who relied on public transportation would be able to reach less than one-quarter of the available jobs within 30 minutes and one-half of the jobs within 90 minutes.
Given these conditions, it is perhaps not surprising that Coulton found that owning an automobile was the most important factor in gaining access to jobs. For example, African American and white welfare leavers who were able to drive a car to work had similar rates of job access. However, when comparing the entire population of welfare leavers, race is a more salient issue in two ways. Regarding car ownership, only 39 percent of African American welfare leavers had access to a car, compared to 51 percent of their white counterparts.
Race also appeared to influence the location in which welfare leavers found employment. African American welfare leavers—including those who lived in the suburbs—found jobs in the inner city. Furthermore, these individuals worked in census tracts that had a higher proportion of African American residents. In contrast, white welfare leavers found employment in a pattern that mirrored the city-suburb ratio of available jobs. Certainly, barriers to traveling to jobs in the suburbs are an important factor in explaining these results, but the explanation might not end there. For example, employers located in census tracts with a low percentage of minorities may hire whites preferentially. It may also suggest that African American welfare leavers seek jobs in areas in which they believe they will be less likely to experience discriminatory attitudes.
The end result of the spatial mismatch between African American welfare leavers and jobs and limited access to automobiles is that blacks found jobs in an area 65 percent as large as that of white welfare leavers. African American welfare leavers living in the suburbs were more likely to commute to the city for work than were their white counterparts (30 percent compared to 13 percent). Finally, black welfare leavers had the least spatially dispersed employment patterns and whites the most dispersed.
Coulton’s paper was broad in scope, considering in depth the effects on transitions from welfare to work of concentrated poverty and a high percentage of welfare recipients in a neighborhood, a broad array of factors that could lead to neighborhood effects, the outcomes of residential mobil
ity and churning, and spatial-geographic barriers to work. Research in these areas is promising, suggesting possible mechanisms by which neighborhoods might create barriers or opportunities for people transitioning from welfare to work, and yet a number of studies (e.g., MTO versus Gautreaux) have produced mixed results on the extent to which neighborhood is a significant influence on employment. This is especially true, as Gordon Berlin noted, when neighborhood effects are compared to the influence of family and individual characteristics, SES, and the impact of racial differences and racial segregation in neighborhoods. Furthermore, separating the independent effects of these influences is difficult.
Harry Holzer complicated the issue further by pointing out that, when it comes to welfare to work, not only do studies of place yield mixed results, but the particular city in which a study is undertaken makes an enormous difference in the extent to which place-based factors represent a significant influence on residents. The time period in which a study is conducted, as well as the tightness of the labor market then, can overwhelm any observed effects of place. For instance, MTO may have not detected effects because even in control neighborhoods as much as 50 percent of the population was employed. Prior to the booming economy of the late 1990s, these control neighborhoods consistently had employment rates of 20 to 30 percent. Significant changes in the economy will have an impact on studies that are social experiments. Finally, neighborhood effects and spatial barriers to work will likely interact in complicated ways with the time period, overall labor market conditions, structural characteristics of the neighborhood, and individual differences (e.g., race, gender, family size and stability) of the participants.
Conducting research on place-based influences on welfare-to-work transitions is clearly a challenging task. In general, discussants at the workshop thought it was fair to say that research suggests place does matter in facilitating successful transitions and creating barriers but that more research is needed to identify the particular mechanisms by which these effects occur in specific city contexts.
“GETTING UNDER ONE’S SKIN”: NEIGHBORHOOD EFFECTS ON HEALTH
In addition to influencing transitions from welfare to work and child development, neighborhood environments have been theorized to have effects on the health outcomes of residents. While it may seem straightfor
ward that factors such as geographic isolation could create important barriers to finding employment and that social environments affect children’s social and emotional development, the way in which neighborhood conditions might, as one discussant put it, “get under one’s skin” to yield negative physical outcomes is perhaps less obvious.
Jeffrey Morenoff centered his discussion on two neighborhood conditions that stand to affect health—exposure to chronic environmental stress, such as violent crime, and the presence or absence of supportive social relationships. He also analyzed data that identified the underlying mechanisms by which these conditions are transferred to individuals in a manner that produces poor health outcomes. This section reviews the results of two of Morenoff’s studies. The first paper examined neighborhood effects that can provide conditions in which violent crime can proliferate. This research found that collective efficacy—defined as the presence of social cohesion, trust, and a willingness on the part of residents to take action to enforce social norms—was important in explaining the relationship between neighborhood kin/friendship ties and homicide rates and that spatial dependence of homicide is quite strong.
The second paper focused on the underlying mechanisms by which neighborhood effects are translated into health outcomes. Homicide rates proved to be an important predictor of low birth weights, suggesting that regular exposure to the chronic stress of a violent neighborhood produces a “weathering effect” that results in negative health outcomes. In addition to the effects of chronic stress, Morenoff examined reciprocal relationships as a second mechanism by which neighborhoods could affect health. Finally, Morenoff highlighted new perspectives—life course and event course approaches—that are useful in more accurately conceptualizing the ways in which neighborhoods impact residents.
A life course focus conceptualizes the effects of neighborhood on individuals as something that happens over the course of one’s development and life cycle. The weathering effect is a good example of the type of neighborhood effect that fits into a life course approach because a researcher must attend to the way in which experiences can have a cumulative effect on individuals over time. In contrast, an event course approach would focus on the way in which spatial relationships can create the circumstances for certain events to occur. In addition, the approach considers the impact of specific events on individuals. Both approaches are useful in understanding how neighborhoods can affect residents, but they emphasize different aspects of place-based influences.
Data for these studies were drawn in part from the community survey completed as part of the Project on Human Development in Chicago Neighborhoods. This survey assessed the status of Chicago’s neighborhoods in 1995 and 2001 in terms of their social, economic, organizational, political, and cultural structures. By using these community surveys in conjunction with other data sources on crime and health outcomes, Morenoff was able to explore the relationships between place and individual outcomes in important new ways.
Neighborhood Effects and Crime
Neighborhood crime can exert important influences on residents and affect a wide range of issues. Neighbors who are constantly concerned about their safety may be less willing to make use of public spaces and resources and may fear using public transportation to travel to employment, especially if they work a nonstandard shift. In addition, the psychological reactions residents may experience from being exposed to violent crime can reduce their sense of well-being and increase stress in problematic ways.
Morenoff’s study of homicide rates in Chicago neighborhoods sought to answer why some neighborhoods are more violent than others by identifying the underlying mechanisms. Were there neighborhood characteristics beyond SES that accounted for the uneven distribution of violent crimes? To answer these questions, Morenoff examined rates of collective efficacy and social ties in neighborhoods, two mechanisms that have received the attention of scholars for the potential they are thought to have in creating an environment less conducive to crime. Essentially, if neighbors are willing to informally supervise and intervene in the activities occurring in neighborhood space, it may be more difficult for individuals bent on illicit activities to accomplish their goals.
In more specific terms, collective efficacy is a representation of the extent to which neighbors share ideas about acceptable social behavior and the expectation that they would take appropriate action if one observed someone behaving in a manner that violated normative standards. It combines social control and community cohesion and trust in a manner that emphasizes residents’ willingness to take action and make use of social resources for specific purposes. In this respect, collective efficacy moves beyond traditional notions of social capital that center on the potential resources stemming from close social ties in a neighborhood because collective efficacy captures the extent to which residents are willing and expect each
other to take action and mobilize public resources. For example, in a neighborhood with high collective efficacy, neighbors would state that they expected other residents would take action if they saw children skipping school or spray painting graffiti, if someone observed a fight in front of their house, or if it was discovered that the city planned to close a nearby library or fire station. Residents would also describe their neighborhood as close knit, meaning that neighbors share the same values and trust each other, share in the supervision of young people, and are willing to assist other residents.
The strength of social ties (measured by the number of kin and friends that residents reported living in the neighborhood) and collective efficacy were measured for Chicago neighborhoods, and each neighborhood was classified into one of four categories: (1) weak social ties and low collective efficacy; (2) weak social ties and high collective efficacy; (3) strong social ties and low collective efficacy; and (4) strong social ties and high collective efficacy. Using Geographic Information Systems software, a typological map was created to represent neighborhoods matching these categories (see Figure 3-1). This map was overlaid with markers of homicide “hot spots” (i.e., areas with high homicide rates) and “cold spots” (low homicide rates).
The results showed that areas with low collective efficacy, regardless of the strength or weakness of social ties, had high rates of homicide, whereas areas high in collective efficacy were homicide cold spots. This suggests that collective efficacy is an extremely important mechanism through which a neighborhood climate that discourages crime can be created. It also suggests that the potential of social ties to improve the climate of a neighborhood is mediated through collective efficacy. In other words, the potential for close relationships among neighbors is not enough to affect the climate of a neighborhood. Rather, neighbors must be willing to act as social agents by exerting a regulatory force over inappropriate behavior they observe.
Morenoff’s analysis also suggested that spatial dynamics may be an important factor in the rates of violent crime. Neighborhoods that were close to a community with high collective efficacy had lower homicide rates than those that were not. Collective efficacy appeared to have a spillover effect into other neighborhoods regardless of whether the other neighborhoods had high or low collective efficacy. Morenoff noted that this finding suggests researchers need to be careful not to think of neighborhoods as islands unto themselves but rather as spatially interconnected locations that will affect and be influenced by the conditions that surround them.
Although SES remains the strongest and most consistent predictor of
neighborhood homicide rates, collective efficacy and spatial dependence exert powerful independent effects on violent crime. In Morenoff’s study, social ties and institutional resources did not have independent effects, suggesting that the potential of relationship and institutional resources is mediated through collective efficacy: if neighbors are not willing to use these resources, they cannot be of help in creating an environment that reduces opportunities for crime to occur.
Neighborhood Effects and Health
Many health outcomes have long been thought to be influenced by one’s environmental conditions. Low birth weights may well be one such outcome. Large and persistent racial and ethnic disparities have been documented, and many researchers have asked to what extent these disparities can actually be explained by the neighborhood in which one lives. At the workshop, Morenoff presented a conceptual model of factors that may represent important neighborhood influences on a mother’s health (see Figure 3-2). Neighborhood effects included structural features of the environment such as SES and racial segregation, institutional resources such as health care and social services, and ecological sources of stress such as crime and violence. Morenoff’s presentation focused on the effects of ecological sources of stress (e.g., homicide as a marker of exposure to violent crime) and social resources that offer insight into racial disparities in neighborhoods. The next two sections discuss the mechanisms underlying these two particular neighborhood effects and summarize the results of Morenoff’s data.
Ecological Influences: Homicide Hot Spots and Low Birth Weight
Researchers have theorized that chronic stress stemming from one’s environment and lifestyle can create an “allostatic load,” meaning a cumulative physiological effect stemming from prolonged exposure to stress. The effects of chronic stress have been linked to a number of poor health outcomes, including heart disease, asthma, and low birth weight—an outcome that may be particularly problematic because of the potential to predispose a child to developmental problems. Because of the link between stress and these health conditions, outcomes such as low-birthweight babies offer a good representation of the extent to which neighborhoods were “getting under the skin” of residents. It is also the demographic factor that is regularly tracked by a number of sources that can be linked to specific spatial locations.
One explanation of how environmental stress can influence individuals is captured in the notion of a “weathering effect.” This term represents the impact that stress can have over time, suggesting that prolonged exposure to chronic ecological stress—such as violent crime—may have a cumulative effect on individuals that can lead to poor health outcomes. In other words, the longer individuals are exposed to environmental stressors, the more likely it is that this allostatic load will translate into low-birth-weight babies, asthma attacks, heart disease, and other health problems.
Morenoff began this part of the study by mapping clusters of homicides and low-birth-weight babies. As visible in Figure 3-3, there was significant overlap in the proximity of these occurrences. A multivariate analysis confirmed what the maps suggested: after controlling for a wide range of individual-level risk factors linked to low-birth-weight babies, neighborhood homicide rates were an important predictor of birth weight.
In terms of understanding the way in which neighborhoods can affect residents, it is worth pointing out that crime is an adverse event to which individuals can be exposed. This exposure can be in many forms—for example, witnessing an assault can trigger long-lasting fears of being attacked. Exposure to a crime has a different type of effect compared to other neighborhood effects that are related to the compositional characteristics of the population in residence. For example, high unemployment rates in an area can weaken information networks about potential employment. This structural characteristic—unemployment—can affect social networks. In contrast, violent crimes are events, and, although such events may be influenced in part by social mechanisms like neighborhood collective efficacy,
treating event-based influences in the same way as a demographic characteristic may provide fewer insights into how to remedy the situation.
In addition to finding a significant spatial overlap between high rates of homicide and low birth weights, Morenoff discovered an important relationship between the age of mothers and their babies’ birth weights. In low-crime neighborhoods there is relatively little relationship between maternal age and birth weights. However, in high-crime neighborhoods, African Americans appear to experience a weathering effect, whereby as the age of
the mother increases, birth weight drops. This suggests that chronic exposure to this type of ecological stress creates an allostatic load that affects these women over time. In contrast, non-Hispanic whites and Mexican Americans appear to be more adversely affected by living in a high-crime neighborhood if they give birth at a younger age than at an older age, suggesting that a mechanism other than weathering may be needed to understand this phenomenon.
Reciprocal Relationships and Racial Disparities in Birth Weight
Racial disparities in Chicago’s birth weights have been significant and persistent. For example, from 1989 to 1996 non-Hispanic whites had the highest average birth weights, with Hispanics of Mexican descent having only slightly lower averages. African Americans had the lowest birth weights, and these were substantially lower than those for non-Hispanic whites and Hispanics (see Figure 3-4). These racial disparities raise important questions about the role of neighborhoods. For example, it might make sense to assume that because ethnic minorities living in inner cities often reside in the poorest neighborhoods and have substantially lower average incomes than whites, SES must account for these observed differences. Certainly, this could be part of the picture, yet it cannot explain what some have called the Hispanic paradox: low SES does not appear to affect the birth weights of the babies of Mexican American women. Another mechanism must have a mediating influence.
In his presentation, Morenoff again turned to collective efficacy but this time focused his measurement on reciprocal exchange rather than social control and cohesion. Again, collective efficacy can be understood as the extent to which residents are willing to take action and use the social and institutional resources in their neighborhoods. Because of this a variety of measurements can be used to capture this “will to act,” and scales that are specific to the particular outcome will more accurately determine whether collective efficacy is the mediating mechanism in a situation. Reciprocal exchange measured the extent to which neighbors were willing to care for one another and lend assistance. The scale inquired about such matters as the frequency with which neighbors exchanged favors for each other, watched over each other’s property, socialized and visited with one another, and asked each other for advice.
The results indicated that reciprocal exchange has a significant protective effect on birth weights, especially for whites and Mexican Americans.
Nonsignificant effects were found for African Americans, and Morenoff speculated that may be because family support has historically been primary for African Americans, with friends and neighbors representing secondary sources of support. Therefore, the scale of reciprocal exchange that focused on neighbors may have masked the importance of familial social support in mediating birth weights for African Americans.
Another interesting result of this part of the study was the relationship between neighborhood composition and birth weights. Mexican American babies’ birth weights were highest in minority neighborhoods and lowest for Mexican families living in predominantly white neighborhoods. Scholars have speculated that close-knit Mexican communities may provide a buffer against acculturation. This pattern could be of great benefit if part of the U.S. culture not assimilated into Mexican communities includes less healthy lifestyle choices (e.g., eating at fast food resturants, smoking). In contrast, racial segregation posed a significant health risk for African American mothers above and beyond other measured neighborhood risk factors.
Developmental Perspective and Event-Oriented Approaches
In this part of the discussion, Morenoff emphasized the need to examine neighborhood phenomena in two ways. A person-based view would
emphasize developmental effects of neighborhood conditions that may become attached to a resident and yield long-term consequences. For instance, chronic stress from exposure to violence and crime that leads to low birth weight may place a child at risk for learning disabilities that can make school experiences more daunting. In this case, neighborhood conditions may stay with a mother and her child for years. In contrast, an event-based perspective would focus on the situational effects of the likelihood of events occurring in a neighborhood. This approach may be particularly helpful in understanding how to prevent crime and other event-based issues that may have shorter-term, more episodic effects. For example, several situational factors must come together to provide a venue in which crime can occur: a motivated offender and suitable target must meet in the same place, and this place must lack capable guardians who can or would intervene effectively. When these events occur together, a crime can occur. An event-based approach would be helpful in developing strategies to reduce crime. In contrast, addressing the effects of crime on individuals would be better served with a people-based approach.
Morenoff’s analysis of health outcomes and their relationship to place and neigborhood effects covered a broad range of material and identified a number of important relationships. Collective efficacy emerged as an important mechanism that has the potential to reduce neighborhood crime and improve health outcomes such as birth weights. Morenoff stressed that the concept of collective efficacy reflects not just shared values or a sense of community in a neighborhood but also a belief in the willingness of other neighbors to act and intervene in situations.
One point stressed by Morenoff and workshop discussant Patricia O’Campo was that the specific nature of collective efficacy will vary depending on the phenomenon being explored. As a result, researchers must be careful to tailor the scale of collective efficacy to the phenomenon in order to accurately evaluate its presence and effect. For example, with regard to reducing juvenile crime, instruments of collective efficacy might measure the extent to which neighbors would intervene if they observed young people engaged in vandalism or other illicit activities. In contrast, in determining whether collective efficacy is important to birth weights, social ties and reciprocity among neighbors (such as doing favors and caring for one another) would be the salient issues to measure.
Attending to individual life course patterns, critical ages, and developmental trajectories as they are influenced by race and ethnicity was identified by workshop discussants as an important part of conducting rigorous research on neighborhood effects. This was most evident in Morenoff’s discussion of the weathering effect as a mechanism that is important to African American birth weights but not to women of other ethnicities. Finally, while a developmental or people-based approach to neighborhood effects may be important in understanding how negative effects become attached to individuals, solving some of the situational-based causes of these effects may require an event-based analysis.
The workshop discussions summarized in this chapter reflect the growing knowledge of neighborhood effects and advances in conducting research to identify key mechanisms responsible for translating neighborhood conditions to individual outcomes. While the data are preliminary—the Moving to Opportunity study involves five years of data collection, while the Project on Human Development in Chicago Neighborhoods will continue for eight years—results thus far indicate that neighborhood has important implications for residents with regard to employment and welfare, child and adolescent development, and health outcomes. In addition, it is clear that how a neighborhood affects its residents is complicated and that developing appropriate policy and programmatic responses will be challenging. Despite this challenge, the careful measures and critical analyses displayed in research such as that conducted by Coulton, Leventhal, and Morenoff demonstrate that identification of factors that can be targeted for intervention is possible and may have the potential to yield better neighborhoods and better outcomes for residents in the future.