Residential Mobility and Opportunities: Early Impacts of the Moving to Opportunity Demonstration Program in Chicago

Housing Policy Debate Volume 12, Issue 2 321 Fannie Mae Foundation 2001. All Rights Reserved. 321 Residential Mobility and Opportunities: Early Impacts of the Moving to Opportunity Demonstration Program in Chicago Emily Rosenbaum Fordham University Laura E. Harris The Urban Institute Abstract This article uses survey data from the Moving to Opportunity demonstration program in Chicago to explore changes for households moving from public housing. The focus is on two key areas: housing and neighborhood conditions, and labor force participation and employment of householders. The experimental design of the program allows the differences between comparison households, which moved with a regular Section 8 voucher, and experimental households, which moved to low-poverty neighborhoods with housing counseling assistance, to be examined. The findings, based on interviews an average of 18 months after families moved, reveal dramatic improvements in neighborhood and housing conditions for all participating families; experimental families experienced even greater gains in terms of housing and especially neighborhood conditions. Labor force participation and employment increased for householders in both groups, likely fueled by the robust economy throughout much of the country and supporting similar findings for program participants in New York and Boston. Keywords: Housing; Mobility; Neighborhoods Introduction The concentration of economic and social disadvantage in the nation s inner cities presents a formidable challenge to policy makers. Extremely disadvantaged areas offer their residents far fewer opportunities for social and economic mobility than more affluent areas and thus can adversely affect residents life chances. Empirical research has demonstrated that residence in disadvantaged neighborhoods increases the risk of dropping out of high school, teen childbearing, and adolescent delinquency, among other negative outcomes (Brooks-Gunn, Duncan, and Aber 1997; Ellen and Turner 1997; Leventhal and Brooks-Gunn 2000). While such neighborhood effects tend to be smaller than the effects of family and background characteristics, their emergence, net of controls for individual-level attributes, lends support to the idea that people s life chances can be improved if they move to neighborhoods offering more and better opportunities (Rosenbaum 1995). Such an idea

322 Emily Rosenbaum and Laura E. Harris is not lost on residents of highly disadvantaged neighborhoods, who view such mobility as a key family-management strategy, but one that is too often unavailable because of inadequate financial resources (Furstenberg et al. 1999). Tenant-based housing assistance (i.e., vouchers and certificates) can help diminish the financial constraints preventing low-income families from relocating to better neighborhoods (Goering, Stebbins, and Siewert 1995; Turner 1998). Recent research demonstrates that low-income households receiving tenant-based assistance are less likely than their project-based counterparts to live in highly segregated and extremely poor neighborhoods (Newman and Schnare 1997; see also Goering, Stebbins, and Siewert 1995; Hartung and Henig 1997; Varady and Walker 2000). However, evidence also suggests that the ability of assisted households to relocate to better neighborhoods is affected by their race and by their relocation to suburban, versus urban, areas (Goering, Stebbins, and Siewert 1995; Turner 1998). For example, among assisted households, blacks are more likely than whites to relocate to areas with higher concentrations of poverty and black residents (Goering, Stebbins, and Siewert 1995; Turner 1998), a differential also found in the broader housing market. This racial differential in locational outcomes reflects the persistence of racial discrimination by housing market actors (Yinger 1995), including landlords, which may also limit the ability of minority families to acquire residences in suburban locations. Thus, while housing vouchers can alleviate financial constraints on the housing choices of low-income households, they are not effective against the racial barriers that persist in the housing market. Evidence from housing mobility programs, however, suggests that the racial barriers constraining the housing choices of nonwhite assisted households can be overcome through the provision of additional services, such as housing counseling and landlord outreach (Goering, Stebbins, and Siewert 1995; Turner 1998). As a result, supplemental services can improve the ability of assisted households to gain access to the resources and opportunities available in better neighborhoods, and thus the potential to improve their social and economic standing. For example, longterm evaluations of the Gautreaux program in Chicago suggest that children moving to the suburbs were more likely than children who moved elsewhere in Chicago to have enrolled in college-bound tracks while in high school, to have graduated from high school, and to have enrolled in college (Rosenbaum 1991, 1995). A number of methodological problems with these evaluations, including the absence of a true experimental design, have raised questions about their generalizability. Because housing vouchers constitute the main thrust of low-income housing policy in the United States, there is a critical need for additional information on whether housing mobility programs improve the life chances of poor families.

Early Impacts of the Moving to Opportunity Program 323 Information from the Moving to Opportunity (MTO) demonstration program can help address this need. Beginning in 1994, the MTO program offered housing vouchers to low-income families living in publicly assisted housing in high-poverty neighborhoods (tracts with population poverty rates of at least 40 percent) in five cities (Baltimore, Boston, Chicago, Los Angeles, and New York). The program is characterized by a controlled experimental design. Participating families were randomly assigned to one of three groups: (1) the experimental, or MTO, group, which received housing assistance and mobility counseling 1 (in addition to the basic counseling received by any assisted family) and was required to move to low-poverty neighborhoods (i.e., tracts with a population poverty rate not exceeding 10 percent); (2) the comparison, or Section 8, group, which received housing assistance and could move anywhere; and (3) the control group, which received no vouchers or certificates (but continued to receive project-based assistance) and could move on their own. Among the long-term, policy-relevant research questions, two concern the question of whether assisted mobility can improve the lives of poor families (Goering et al. 1999). The first relates to the impact of housing counseling on families housing choices and their housing and neighborhood conditions. The second focuses on the ultimate impact of neighborhood conditions on the social and economic well-being of participating families and children, including educational and employmentrelated outcomes (U.S. Department of Housing and Urban Development [HUD] 1996). In this article, we describe the short-term effects of moving from public housing on families participating in Chicago s MTO program. With the overall goals of the program in mind, we focus on housing and neighborhood conditions and on respondents employment. The article addresses 1 As described in a later section, the housing assistance and mobility counseling portion of the MTO program was ultimately handled by CHAC, a private, nonprofit organization. At CHAC, there was an outreach coordinator whose job was to find available units and to compile a list each week of such available units that experimental households and their counselors could use in their housing searches. The outreach coordinator used a variety of techniques to locate available units: searching the major daily papers and gathering local daily or weekly papers, visiting community centers and neighborhood hangouts to look for postings about available units, and driving around neighborhoods to look for For Rent signs. Households in the experimental group were assigned to a housing counselor who helped them go over these lists and negotiate the process of renting an apartment. Counselors would call MTO participants when they learned of an available apartment and often drive the families to visit prospective units and facilitate interactions with landlords. While these services were offered to MTO experimental households in general, there were only two counselors for the entire MTO program in Chicago. Discussions with both the counselors and the participating families revealed that there was a wide variation in how much interaction each household had with its counselor. We will address the participating families reports of their experiences with counseling later.

324 Emily Rosenbaum and Laura E. Harris two basic questions: First, to what degree does participation in the MTO program change families experiences of these aspects of well-being? And second, how do families experiences differ, depending on program group assignment? While by design we are limited to examining the short-term effects of moving and do so mainly in a descriptive format, the analysis addresses important issues relating to MTO s effectiveness in improving the well-being of participating families and the particular effect of the program requirement of moving to low-poverty neighborhoods. In the next section, we describe the sources of data used in the analysis. We then describe the results, beginning with an overview of the characteristics of the families in the analytical sample. Finally, we conclude with a discussion of the implications of the analysis for housing mobility programs in general and for long-term evaluations of MTO in particular. Sources of data The analysis relies on three sources of data. The first is the Urban Institute s Underclass Database, which provides 1990 census information for the census tracts from which the participating families moved (the origin neighborhoods) and for the tracts to which they moved (the destination neighborhoods). The second is HUD s baseline survey, which participating families completed before moving. 2 The baseline survey collected basic sociodemographic information, attitudinal measures, measures of housing and neighborhood conditions, and information on a variety of other topics. The respondent was the householder the person applying for the program. The third source of data is a telephone survey conducted with mover families in Chicago after they moved from public housing. 3 Administrative problems at the Chicago site significantly delayed the receipt of the contact information (telephone numbers and addresses). In 1994, soon after the MTO program began, the Chicago Housing Authority (CHA) was taken over by HUD, and a private organization, CHAC, assumed responsibility for the Section 8 program (where MTO was housed). In 1996, the not-for-profit organization that had provided the counseling and search assistance to experimental families the Leadership Council for Metropolitan Opportunities decided to not continue 2 Among the 120 families in the analytical sample, almost two-thirds (63.4 percent) completed the baseline survey in 1995, 18.3 percent in 1996, and 18.3 percent in 1997. Twenty percent of the 120 families in the sample moved in 1995, 36.7 percent in 1996, 36.6 percent in 1997, and 6.7 percent in 1998. 3 This survey was funded by a small grant from HUD as part of a program funding eight teams of researchers to do short-term evaluations of the five sites. Our team was the only one to examine the Chicago site.

Early Impacts of the Moving to Opportunity Program 325 in this role, resulting in CHAC s assuming these tasks as well. These changes in the administration of CHA and MTO slowed the initial pace of participant enrollment and, as a consequence, our receipt of contact information. Thus, whereas 20 percent of the 120 families we ultimately surveyed moved during 1995, our first survey did not occur until late 1996. 4 As important as the delay in receiving information was its quality. In particular, telephone numbers were available for only a minority (less than a third) of mover families, and many phones had been disconnected by the time we received the information. We relied on four distinct strategies for locating hard-to-contact mover families. The first was contacting the friend or family member indicated by the participating family on the baseline survey as someone likely to know how to contact them. The second was using the local telephone directory. The third involved writing to participating families and asking them to return to us in a self-addressed stamped envelope their phone number and a convenient time we could contact them. 5 Finally, a member of the research team traveled to Chicago to locate more families in person and either interviewed them at that time (in person) or called them back at a mutually convenient hour. While each of these strategies added to our list of successful contacts, they did not completely overcome the problems we encountered. As a result, our overall response rate was quite modest. The goal for the number of households to lease up in Chicago was 285 (143 in the experimental, or MTO, group and 142 in the comparison, or Section 8, group), but the research team ultimately received contact information for 234 mover families, of whom 120 (or 51.3 percent) were contacted and surveyed. 6 Of these 120 families, 67 were in the MTO group, and the remaining 53 were in the Section 8 group. These 120 families, then, comprised the basic analytical sample. 4 The postmove survey was conducted between November 1996 and November 1998, 13 months, on average, after the families moved and 24.6 months after the baseline survey. 5 Although this method was fairly successful, many letters were returned by the post office, suggesting that the households were no longer at these addresses. Others simply went unanswered, implying either that the respondents received the letters and chose not to respond, or that the addresses were so out of date that the post office no longer returned them. 6 The modest response rate reflects in part the administrative problems at the Chicago site, but also the problems involved in conducting a telephone survey with a very poor and mobile population. An evaluation of the New York City site (Leventhal and Brooks- Gunn 2001) achieved only a 68 percent response rate for an in-person survey. The New York sample, moreover, included in-place (i.e., nonmoving) control families as well as mover families that had not yet moved; these kinds of families should have been more easily located than those who had actually moved.

326 Emily Rosenbaum and Laura E. Harris Given the modest response rate, a key issue to consider is the extent to which respondents to the postmove survey may have differed from noninterviewed mover families in the Chicago MTO program. To address this issue, simple comparisons of interviewed families with noninterviewed families, specific to program group, were performed using baseline data. The variables chosen were attitudinal items tapping into feelings of social distance and outlook for the future, since the racial and socioeconomic homogeneity of the program s families in Chicago limited the utility of background variables in this kind of effort. The results of these comparisons are found in table A.1. Simple two-tailed t-tests found no significant differences for MTO householders in terms of feelings of social distance or the householder s outlook about the family s prospects in the new neighborhood. However, for the Section 8 householders, the comparisons indicate that interviewed householders were significantly more likely than noninterviewed householders to feel good or very good about sending their children to schools where their new classmates were either half or almost all white (p 0.01, two-tailed t-test), to feel good or very good about living somewhere where more than half of their neighbors earned more money than they did (p 0.10, two-tailed t-test), and to be sure or very sure that they would like living in a neighborhood they had never lived in before (p 0.05, two-tailed t-test). The results of these simple comparisons suggest that the results reported here for MTO families are representative of those of all MTO mover families in Chicago. The findings for the Section 8 families suggest that our results might be biased in favor of those families with the most optimistic outlook on their potential new neighbors and neighborhoods. This bias suggests that any positive findings for families in the Section 8 group might be exaggerated because our sample is biased toward the families that might be the most successful at finding homes in better neighborhoods. In addition, because the findings for the Section 8 group might be overstated, the differences uncovered between MTO and Section 8 mover families might be understated relative to the situation that would have held if all mover families had been interviewed. 7 7 To take full advantage of the experimental design of MTO, it would be necessary to include the unsuccessful movers in the research, i.e., those who joined the program and were randomly assigned to a program group, but who did not move (Katz, Kling, and Leibman 2001; Leventhal and Brooks-Gunn 2001). Including both movers and nonmovers would have allowed us to examine the intent-to-treat effects of program-group assignment. The main focus of our research, however, was to examine outcomes for the successful movers, investigating differences between households from the experimental group and those from the comparison group that actually moved from public housing. On average, at the five sites, experimental families were less likely than comparison families to lease up (Goering et al. 1999). In addition, because the original goal of our project was to examine how mover families had adjusted to their new neighborhoods in the short term, our original research design did not include families assigned to the control group.

Early Impacts of the Moving to Opportunity Program 327 Results Overview of sample characteristics With random assignment, the characteristics of all MTO and Section 8 families should be statistically indistinguishable. Statistical tests confirmed this for virtually all background variables from the baseline survey for our sample of interviewed movers. As a result, this section describes the characteristics of the 120 families in the sample as a group, and the few instances where statistically significant differences emerge between MTO and Section 8 families are indicated. Descriptive statistics for the entire sample are shown in table 1. At the Chicago site, virtually all household heads were women and (non- Hispanic) African Americans. Respondents averaged slightly less than 33 years of age. Most almost two-thirds had never been married, and slightly fewer than half reported having lived with both parents until age 16. In general, the householder tended to be the only adult in the household. Sample households also contained an average of slightly fewer than two children aged 6 to 17 and just under one child aged 5 or younger (for MTO families, the mean number of younger children was 0.63; for Section 8 families, it was 1.10, p 0.01), for an average household size of just under four people. 8 Given the nature of the population involved in this demonstration program, relatively low levels of socioeconomic status and human capital were to be expected. Slightly fewer than two-thirds of respondents had received either their high school diplomas or their General Education Diploma (GED). Only 27.8 percent reported working full- or part-time at the time of the baseline survey, but the vast majority reported some lifetime work experience. Also, most households received a form of needs-tested public assistance: 79.3 percent received Aid to Families with Dependent Children (AFDC), 82.8 percent received food stamps, and 67.9 percent received Medicaid. In addition, the data indicate that for most respondents, reliance on public assistance was a long-term experience. The average length of time since the first application for AFDC was just under 11 years, and 68 percent reported that their mothers also received AFDC. On the baseline survey, respondents answered questions about their previous residential experiences. These experiences may have influenced their search for housing, since they affected the respondents familiarity with different kinds of neighborhoods, as well as their knowledge of housing opportunities outside of public housing. Relatively few 8 Every household but one had children. The number of children ranged from 0 to 7 for those between the ages of 6 and 17, and from 0 to 4 for those under 5 years of age. The mean number of children regardless of age was 2.53.

328 Emily Rosenbaum and Laura E. Harris Table 1. Descriptive Baseline Statistics of Interviewed Householders and Their Households Moving as Part of the Moving to Opportunity Demonstration Program in Chicago (Maximum N = 120) Characteristic Percent Female 96.6 Black 97.4 Age (mean) 32.7 Marital status Never married 65.2 Currently married 7.0 Separated 19.1 Widowed/divorced 8.7 Lived with both parents until age 16 44.0 Graduated from high school a 63.0 Currently working full-time 14.8 Currently working part-time 13.0 Ever worked for pay b 85.6 Receives AFDC 79.3 Receives food stamps 82.8 Receives Medicaid 67.9 Mother received AFDC 68.1 Years since first received AFDC (mean) 10.6 Household size Total (mean) 3.78 Number of adults (age 18 and older; mean) 1.25 Number of children ages 6 17 (mean) 1.69 Number of children ages 0 5 (mean) 0.84 Residential experiences c Has ever lived outside of Chicago 21.9 Has ever lived in a neighborhood that was A mix of African American and white 51.5 A mix of African American and Hispanic 27.8 A mix of Hispanic and white 14.9 A mix of African American, Hispanic, and white 30.3 Mostly white 8.5 Residential preferences Where would like to move Somewhere else in my neighborhood 1.8 Different neighborhood in Chicago 66.7 Neighborhood in the suburbs 21.9 Different city outside of Chicago 7.9 Other 1.8 Kind of neighborhood would like most to live in Mostly African American 8.8 Mostly white 2.6 A mix of African American and white 35.1 A mix of Hispanic and white 0.9 A mix of African American, Hispanic, and white 49.1 Other 3.5 a Includes General Education Diploma. b Includes those currently working. c Because respondents could choose all categories that applied to their lifetime experiences, percentages will not sum to 100 percent. AFDC = Aid to Families with Dependent Children.

Early Impacts of the Moving to Opportunity Program 329 respondents (21.9 percent) reported ever having lived outside the city of Chicago, yet a substantial proportion appeared to have lived at some point in their lives in various kinds of racially and ethnically mixed neighborhoods. That is, just over half reported having lived in neighborhoods with a mix of African Americans and whites, while 30.3 percent reported having lived in areas with a mix of African Americans, Hispanics, and whites, and 27.8 percent reported having lived in areas with a mix of Hispanics and African Americans. 9 Far smaller proportions reported living in neighborhoods with few or no African Americans, namely, Hispanic-white neighborhoods (14.9 percent) and mostly white neighborhoods (8.5 percent). These experiences could also have influenced respondents preferences for the kinds of areas to which they wanted to move. On the baseline survey, just over 20 percent expressed a preference for moving to the suburbs. In terms of racial/ethnic composition, almost half (49.1 percent) said that they would like to move to an area composed of a mix of African Americans, Hispanics, and whites. Slightly more than a third expressed a preference for racially mixed (African American white) neighborhoods. Relatively few (8.8 percent) said that they would like to move to a mostly African-American neighborhood, and even fewer chose mostly white (2.6 percent) and mixed Hispanic-white (0.9 percent) as the neighborhood compositions they would prefer. The finding that most respondents in the sample preferred to move to racially and ethnically mixed neighborhoods concurs with findings of similar kinds of preferences expressed by African-American respondents in general-population samples (e.g., Farley, Fielding, and Krysan 1997; Zubrinsky and Bobo 1996). Changes in neighborhood conditions The goal of the MTO program is to help poor families move from disadvantaged neighborhoods to neighborhoods with richer opportunities for social and economic mobility. Thus, key questions to address are, To what degree has the program achieved this goal in the short run, and to what degree have the gains in neighborhood conditions varied depending on program-group status? Because of the program requirement that MTO experimental families move to low-poverty neighborhoods, 9 It should be noted that the question on the baseline survey about neighborhood racial composition did not specify the share of each group present in a given neighborhood type, but instead used phrases such as a mix of. Thus, when respondents answered that they had lived in a neighborhood with a mix of African Americans and whites, they may not have meant a neighborhood with equal parts African Americans and whites. In addition, because respondents could choose more than one response category (the question was phrased to tap into lifetime experience in different kinds of neighborhoods), the percentages reported here will not sum to 100 percent.

330 Emily Rosenbaum and Laura E. Harris it was expected that families in this group would have moved to qualitatively better neighborhoods than Section 8 families did. Two kinds of neighborhood indicators were used to answer these questions. The first consisted of objective indicators, namely, tract-level characteristics from the 1990 census. Thus, when examining objective neighborhood characteristics, we defined neighborhoods as census tracts. 10 The second kind of indicator was more subjective, consisting of respondents reports of problems in their neighborhoods. When examining both the objective and subjective indicators of neighborhood conditions, we make comparisons between origin and destination neighborhoods, as well as between the destination neighborhoods chosen by MTO and Section 8 families. Changes in objective characteristics. Indicators relating to four aspects of neighborhood conditions suburban location, racial/ethnic composition, economic status, and social context for origin and destination tracts are presented in table 2. 11 All origin tracts were located in the central city. The data indicate very clearly that these areas were essentially racially homogeneous; thus, while the families were living in public housing, they had very little, if any, contact in their neighborhoods with persons of other racial or ethnic backgrounds. Taken as a group, the indicators of economic status tell a story of extreme poverty and economic isolation: Almost three-fourths of the population was officially poor in 1990, and more than half 58.6 percent relied on public assistance. Furthermore, the extremely low median household income (just over $6,000), combined with the virtual absence of homeownership, indicates that the origin neighborhoods were not just lacking income but wealth as well. With respect to the social context prevailing in origin neighborhoods, the data suggest that these areas offered relatively few role models of postsecondary education or active participation in the legitimate labor force (Wilson 1987, 1996). Moreover, the vast majority of families and subfamilies 12 in the area were headed by single women. Thus, the areas from which the families in Chicago moved were extremely isolated, racially, economically, and socially, and appeared to have offered the families few if any opportunities for, and examples of, social and economic achievement. 10 Naturally, the families in the sample are unlikely to define their neighborhoods in precisely the same way. 11 This article addresses the characteristics of new neighborhoods chosen by mover households, rather than the location of their moves. Goering et al. (1999) have created maps to document the origin and destination neighborhoods of MTO families throughout the entire metropolitan area in another report. 12 A subfamily is a family unit (i.e., a married couple with or without children under 18, or a single parent with children under 18) residing in a housing unit headed by someone else.

Early Impacts of the Moving to Opportunity Program 331 Table 2. Summary Statistics Based on Selected 1990 Census Tract Characteristics, for All Origin Tracts and for Destination Tracts by Program-Group Status Destination Tracts Origin Tract Characteristic Tracts Section 8 MTO Suburban location 0.00 1.90 31.30 (0.00) (14.00) (47.00) Racial/ethnic composition Percent non-hispanic black 99.29 90.00 57.22 (0.79) (20.51) (38.32) Percent non-hispanic white 0.11 5.99 34.10 (0.20) (14.25) (32.05) Percent Hispanic 0.60 3.21 7.17 (0.70) (8.53) (11.82) Economic status Percent of tract population in poverty 74.99 36.61 10.60 (13.70) (15.76) (8.84) Percent of households receiving 58.61 31.15 10.15 public assistance (16.06) (13.34) (8.66) Median household income $6,211.92 $17,041.19 $33,716.12 ($1,678.38) ($8,736.85) ($8,702.38) Percent of owner-occupied homes 2.82 26.22 66.17 (4.81) (21.61) (19.88) Social context Percent of female-headed families 84.72 65.54 36.99 and subfamilies (5.97) (16.94) (16.89) Percent of adolescents ages 16 19 17.03 18.11 9.49 neither enrolled in nor graduated (10.93) (10.72) (8.47) from high school Percent of adults (25+) with a 8.07 15.39 23.02 college degree (3.43) (10.36) (13.05) Percent of population (16+) in the 38.13 55.57 67.94 civilian labor force (7.25) (11.41) (6.78) Number of tracts 118 53 67 Note: All figures are means, and standard deviations are in parentheses. All differences between MTO and Section 8 destination tract characteristics are significant at least at the p 0.05 level (one-tailed t-tests). How do the destination tracts compare with the origin tracts? For the Section 8 families, the data indicate that the average tract to which families in this group moved was still quite racially segregated, with 90 percent of the population reporting black race in the 1990 census. This finding is notable given that very few respondents overall (8.8 percent) preferred to move to mostly black neighborhoods. However, the fairly large standard deviation (20.51) for this mean suggests that Section 8 families actually moved to neighborhoods with a wide variety of racial compositions. A similar story emerges with respect to the economic indicators. For example, while the average poverty rate in Section 8 destination neighborhoods remained quite high at almost 37 percent, some families moved to very low poverty neighborhoods (with poverty rates as low as 4 percent), while others moved to areas that were almost

332 Emily Rosenbaum and Laura E. Harris as poor as their origin neighborhoods (with poverty rates as high as 70 percent). Similarly, while homeownership was far more prevalent in Section 8 families destination neighborhoods than in their origin neighborhoods, the standard deviation is almost equal to the mean, suggesting that some families ended up in areas with virtually no owner-occupied homes. With respect to social context indicators, the areas to which Section 8 families moved had far more adults who had completed college and were actively engaged in the labor force, although the standard deviations again hint at the considerable variety in Section 8 families destination tracts along these dimensions. Finally, only 1.9 percent of Section 8 families (which, in fact, amounts to a single Section 8 family) relocated to a suburban location. Given the program requirement that MTO families move to low-poverty neighborhoods, it was expected that these families would have relocated to more advantageous areas than their counterparts in the comparison group. The data provide clear support for these expectations. That is, relative to Section 8 families destination tracts, those to which MTO families moved were better off economically, with far lower rates of poverty and public assistance, and higher median incomes and homeownership. These tracts also were characterized by a lower prevalence of female headship and teens who were neither enrolled in school nor in the labor force, and higher levels of adult labor force participation and college completion. In terms of racial/ethnic composition, these tracts were also the most diverse on average, with just over 57 percent of the population reporting black race in the 1990 census. Just under one-third (31.3 percent) of MTO families relocated to homes in the suburbs. Such a finding suggests the operation of unmeasured (at the baseline) preferences, such as those for proximity to friends, family, or perhaps key institutions such as the family s church or health care provider (Varady and Walker 2000). However, the fact that a larger proportion of MTO families moved to the suburbs than expressed a preference for living there on the baseline survey (approximately 20 percent overall) suggests that the additional services provided to the families in the MTO experimental group may have opened up housing opportunities that were unknown to them before joining the program or were simply deemed unavailable (and thus not expressed as a preference) (Popkin, Rosenbaum, and Meaden 1993; Varady and Walker 2000). Indirect support for this interpretation comes from questions asked of MTO families about the usefulness of the counseling program. Almost threefourths (72.7 percent) of MTO respondents reported that the information provided by the housing counselors was either somewhat or very useful in their search for a house or apartment. In addition, in excess of 80 percent of MTO respondents reported that the counseling was somewhat or very useful in helping them deal with landlords (80.6 percent) and in signing a lease (84.6 percent). Alternatively, it may be that more MTO than Section 8 families relocated to the suburbs be-

Early Impacts of the Moving to Opportunity Program 333 cause there were not many affordable units available in low-poverty neighborhoods in the city (Harris 1999). In summary, while all families had been living in economically and socially marginal communities when they joined the demonstration program, the tracts to which they moved differed dramatically both from their origin tracts and from each other, depending on program-group status. In other words, while the data suggest that the Section 8 families made significant improvements, on average, in the social and economic environments of their neighborhoods as a result of moving, the tracts to which the MTO families moved were significantly better off in terms of all the indicators of economic and social context. In addition, MTO families were more likely to have moved to suburban locations, and to more racially diverse tracts. These differences by program-group status stand out, given that respondents from the two groups were statistically identical in their initial preferences for suburban and mixed neighborhood compositions. Unfortunately, because of the program s design, it is not possible to determine whether the greater gains in neighborhood conditions achieved by MTO families were due to the provision of supplemental services or were simply inherent to the program requirement that they move to low-poverty neighborhoods. However, the variability in the Section 8 families destination tracts and the greater ability of MTO families to achieve their premove neighborhood preferences are consistent with the notion that housing counseling can help assisted households relocate to substantially improved neighborhoods (Goering, Stebbins, and Siewert 1995; Turner 1998). Changes in subjective conditions. In this section, subjective measures of neighborhood conditions are used to compare neighborhood conditions before and after the move. Insofar as the differences indicated by the objective characteristics are reflected in respondents reports and perceptions of the conditions prevailing in the origin and destination neighborhoods, it was expected that the MTO families would report far better conditions in their destination neighborhoods than the Section 8 families. To enhance the ability to detect changes in neighborhood conditions as a result of moving, the postmove survey included many of the same measures used on HUD s baseline survey. Thus, the comparisons described here focus on this subset of measures, as well as a limited number of measures used uniquely in each survey. The results of the comparisons are shown in table 3. Beginning with reported problems, respondents to both surveys were asked the extent to which five conditions relating to social and physical disorder in the area were big problems, small problems, or no problems at all. These conditions were trash or litter on the street, graffiti on the walls, people drinking in public, drug dealers/users in the area, and abandoned buildings. Items were added to the postmove survey to tap into the presence of other problems such as incivility, crime and violence,

334 Emily Rosenbaum and Laura E. Harris Table 3. Reported Neighborhood Conditions for Origin and Destination Neighborhoods, by Program-Group Assignment (Percent) Origin Neighborhoods Destination Neighborhoods Neighborhood Condition Section 8 MTO Section 8 MTO Big or small problem with Trash or litter on the streets 100.0 95.4 54.7 a 28.4*** a or sidewalks Graffiti or writing on walls 100.0 99.1 22.6 a 10.4* a People drinking in public 98.0 95.4 43.4 a 10.4**** a Drug dealers or users 100.0 98.5 54.7 a 21.2**** a Abandoned buildings 96.1 90.6 32.1 a 7.6**** a Not enough recreational 55.8 35.0** programs for youth People saying insulting things 22.6 6.2*** or bothering other people Crime and violence 54.9 26.9*** Not enough public 20.8 37.3** transportation Lots of people who can t find jobs 79.6 46.7**** Different racial or cultural groups 13.5 4.5* who cannot get along Satisfied with neighborhood 5.9 6.2 New neighborhood is better than 81.1 89.6 old neighborhood N 51 64 53 60 a Difference between baseline and postmove survey significant at p 0.001 level (one-tailed t-test). *p 0.10; **p 0.05; ***p 0.01; ****p 0.001; one-tailed t-test. and joblessness. For these comparisons, responses were recorded into dichotomies reflecting the presence/absence of problems. Nearly all respondents in both groups reported at baseline that trash and litter, graffiti, public drinking, drug users and dealers, and abandoned buildings were problems in their origin neighborhoods. Such unanimity is striking and adds an image of severe social disorder and physical deterioration to the picture of racial, social, and economic isolation derived from the objective characteristics. By contrast, far fewer respondents in each group reported that their new neighborhoods featured similar problems, a finding consistent with the improvements as measured by the objective characteristics achieved by each group. Moreover, the gains that each group achieved in terms of moving to areas with less social and physical disorder are highly significant at the p 0.001 level (one-tailed t-test). However, the comparisons of postmove conditions also indicate that the gains made by the MTO families outpaced those achieved by the Section 8 families. That is, significantly fewer MTO than Section 8 respondents reported problems in their new neighborhoods on each of the five dimensions. Similarly, significantly fewer MTO than Section 8 respondents reported that their new neighborhoods featured problems with insufficient recreational programs,

Early Impacts of the Moving to Opportunity Program 335 crime and violence, widespread idleness, and incivility. The only indicator on which proportionately more MTO than Section 8 families reported problems was inadequate public transportation. This relative disadvantage is likely related to the far greater tendency for MTO families to have relocated to the suburbs. The differences in origin and destination neighborhoods described so far are reflected in respondents overall evaluations of their neighborhoods before and after they moved. At baseline, less than 10 percent of respondents in each group reported having been satisfied with their neighborhoods. However, after the move, statistically similar percentages of respondents reported that their new neighborhood was better than their old one. This statistical similarity suggests that, from the respondents point of view, simply moving out of public housing in Chicago translated into dramatically improved neighborhood conditions (National Commission on Severely Distressed Public Housing 1992; Popkin et al. 2000). Changes in housing conditions This section focuses on comparisons of the families housing conditions before and after the move. As was the case with the subjective indicators of neighborhood conditions, the comparisons of housing conditions rely on respondents reports of problems, and a subset of these items appears on both surveys. These comparisons are presented in table 4. Before moving, approximately two-thirds of MTO and Section 8 respondents reported problems with peeling paint and broken plaster in their units, and three-fourths reported problems with inadequate space. With respect to nonfunctioning plumbing, rodent infestation, and broken/ absent locks on front doors, slightly higher proportions of Section 8 than MTO respondents reported such problems, but the differences never attained statistical significance. Statistically similar proportions also reported problems with broken windows, malfunctioning heating systems, broken appliances, and exposed wires. In short, the data suggest that the public housing units from which the families moved were in severe states of disrepair. After moving, however, respondents from both groups were far less likely to report problems with peeling paint, nonfunctioning plumbing, rodent infestation, broken locks, and inadequate space. Moreover, the gains in housing conditions achieved by both groups of respondents were significant at least at the p 0.01 level (one-tailed t-test). While these findings of improvements are similar to those found for neighborhood conditions, what differs here is that the prevalence of housing problems after the move tends to be statistically similar for the two groups. That is, it is only on two dimensions of postmove housing problems walls

336 Emily Rosenbaum and Laura E. Harris Table 4. Reported Housing Conditions before and after the Move, by Program-Group Assignment (Percent) Premove Postmove Neighborhood Condition Section 8 MTO Section 8 MTO Big or small problem with Walls with peeling paint or 64.7 69.2 37.7 a 16.4*** b broken plaster Plumbing that doesn t work 60.8 46.9 22.6 b 19.4 b Rats or mice 62.8 52.3 26.4 b 17.9 b Broken locks or no locks on the 50.1 43.1 13.5 b 0.9*** b door to unit Too little space 76.5 76.9 34.0 b 32.8 b Broken windows or windows 62.8 52.3 without screens A heating system that doesn t work 41.2 33.9 A stove or refrigerator that doesn t 19.6 23.1 work Exposed wire or electrical problem 31.4 35.4 Overall condition of housing is 30.0 33.9 73.6 b 80.6 b good/excellent New housing is better than old housing 81.1 71.6 N 51 64 53 67 a Difference between baseline and postmove survey significant at p 0.01 level (one-tailed t-test). b Difference between baseline and postmove survey significant at p 0.001 level (one-tailed t-test). *p 0.10; **p 0.05; ***p 0.01; ****p 0.001; one-tailed t-test. with peeling paint or broken plaster and broken/missing locks on front doors that MTO respondents reported significantly better conditions than Section 8 respondents. The greater degree of similarity in housing than in neighborhood conditions after the move is likely due to requirements of the broader Section 8 program for adequate housing. 13 The improvements in housing conditions achieved by families in both groups are reflected in respondents overall evaluations of their housing units. Approximately one-third rated the condition of their public housing units as good or excellent before the move. After the move, the proportion rating the condition of their new units as good/excellent rose to about three-fourths, an improvement that is highly significant at the p 0.001 level (one-tailed t-test) for both groups. Similarly, most respondents in each group reported in the postmove survey that their new unit was better than their old one. As was the case for the parallel measure of neighborhood conditions, the difference between the MTO and Section 8 groups on this indicator is not statistically significant. 13 HUD s Housing Quality Standards require that units rented with Section 8, or vouchers, meet minimum physical standards, such as plumbing that works properly, locks on doors, and the absence of lead paint hazards, thereby eliminating the most severely dilapidated housing units from the housing options available to voucher holders.

Early Impacts of the Moving to Opportunity Program 337 In summary, the data on neighborhood and housing conditions prevailing before and after the families moved as part of the Chicago MTO demonstration program suggest that it has achieved a significant degree of success in its short-term goal of improving participating families experiences on these dimensions of well-being. Moreover, for neighborhood conditions, the gains achieved by the experimental group exceed those achieved by the comparison group by a considerable margin, suggesting an added benefit to be gained by either the use of additional services like housing counseling or the program requirement of moving to lowpoverty neighborhoods. 14 As noted earlier, while it is not possible to determine which of the program features is at the root of the greater neighborhood gains achieved by MTO families, the fact that these families were better able than families in the Section 8 group to relocate to racially mixed and suburban neighborhoods given that respondents from the two groups expressed similar preferences to do so suggests some support for the benefits to be derived from the added services. Employment of respondents Does moving out of public housing improve respondents employment prospects? This is perhaps one of the most crucial policy-related questions to be addressed in the long term with the MTO demonstration program, especially in this postwelfare era. Evaluations of the Gautreaux program indicated that women who moved to the suburbs were more likely than their city-moving counterparts to be employed after more than five years (on average) in the program, even after controlling for various employment-related factors (Popkin, Rosenbaum, and Meaden 1993; Rosenbaum and Popkin 1991). As indicated earlier, because of methodological problems with the Gautreaux evaluations, it is not clear whether such findings are generalizable. If they are, then the findings from the Gautreaux program suggest that the MTO respondents may be more likely than the Section 8 respondents to be employed after moving. However, given that the families participating in the Chicago MTO program had spent far less time in their new neighborhoods than the Gautreaux families had (about one year versus more than five years), we should be cautious in our expectations for dramatic improvements in their labor force participation and employment in the short term. Pre- and postmove labor force participation rates and employment rates for respondents in each group are shown in table 5. Before the move, MTO and Section 8 respondents were equally likely (from a statistical 14 Comparing the outcomes for Section 8 families according to the poverty level of their destination tracts would have helped us differentiate between the role played by housing counseling and that played by the program requirement faced by MTO families. However, only two of the Section 8 families we surveyed moved to tracts with a poverty rate of less than 10 percent, thus preventing us from making this kind of comparison.

338 Emily Rosenbaum and Laura E. Harris Table 5. Labor Force Participation and Employment among Householders, before and after Moving, by Program-Group Assignment (Percent) Premove Postmove Economic Activity Status Section 8 MTO Section 8 MTO In labor force 43.9 49.1 64.3 a 59.3 Employed 29.3 24.5 42.9 46.3 b N 41 53 42 54 a Difference between baseline and postmove survey significant at p 0.05 level (one-tailed t-test). b Difference between baseline and postmove survey significant at p 0.01 level (one-tailed t-test). standpoint) to be in the labor force and to be employed. The same finding holds for their rates after moving; however, both groups demonstrate increased levels of activity. After moving to their new neighborhoods, Section 8 respondents were far more likely to be actively participating in the labor force (i.e., working or looking for work), while for MTO respondents, a statistically significant increase is evident only for employment per se. Some of these increases are due to the entry into the labor market of women with no previous work histories; among those who reported on the baseline survey that they had never worked for pay, 20 percent reported on the postmove survey that they were working, and 27 percent reported either working or looking for work. 15 Thus, despite the fact that the respondents had been in their new neighborhoods for only a year or so, the change in venue appears to have had a powerful effect on their ability to join the legitimate labor market. However, the absence of differences in postmove labor force participation and employment between MTO and Section 8 respondents suggests that the robust health of the economy in the late 1990s probably contributed strongly to these short-run gains in employment. Moreover, similar pre- and postmove employment rates were reported for the New York City site, even among in-place controls (Leventhal and Brooks-Gunn 2001). 16 Similar employment increases are also documented in the research on the Boston MTO program (Katz, Kling, and Leibman 2001). The increases in labor force participation and employment by MTO and Section 8 respondents are matched by decreases in the proportion of respondents who reported that they did any small jobs to earn extra money. While 19 percent of respondents overall reported doing such small jobs as baby sitting or cleaning on the baseline survey, only 6 percent reported such off-the-books activities on the postmove survey. Clearly, being actively engaged in the labor market reduces the time available 15 These figures are not specific to program-group assignment because stratifying by this variable produced extremely small cell sizes. 16 The New York sample was interviewed approximately three years after baseline and included MTO and Section 8 families that had not yet moved.