«Mixed Messages on Mixed incoMes Volume 15, Number 2 • 2013 U.S. Department of Housing and Urban Development | Office of Policy Development and ...»
having financial strain after relocation, and having no public housing friends in destination neighborhoods) show a mixed story. Specifically, the level of being disabled or having a chronic condition is nearly 0.60, but experiencing financial strain is only 0.16. Likewise, having no public housing friends is 0.35, indicating that most did have friends in public housing. Interestingly, the dichotomous variable concerning whether a resident moved to a high-receiving neighborhood is 0.08, indicating that most residents in the sample did not move to this type of area.
The average neighborhood characteristics for the sample used in our analysis, shown in exhibit 2, are consistent with previous research. Specifically, on average, these former public housing residents are moving to neighborhoods with less poverty (but not low poverty) that are racially segregated;
the average poverty level is nearly 32 percent, and the average percentage non-Hispanic African American is more than 84 percent. On average, they are also moving to neighborhoods where the percentage of homes built more than 30 years ago is high (nearly 65 percent). These neighborhoods are also characterized by a high mobility level, with an average of nearly 71 percent living in the
same place for less than 10 years. Average unemployment is about 17 percent and the average of households that are female headed is nearly 20 percent. Average vacancy is slightly more than 21 percent, and the average of households who are renters is nearly 52 percent.
Neighborhood Characteristics by Level of Receivership To put our subsequent multivariate analysis in perspective, we also provide some descriptive information at the census tract level for the entire sample (including those residents we drop from the subsequent multivariate analysis of relocation satisfaction) by levels of receivership. Exhibit 3
shows the receivership categories by census tract. Based on the distribution of former public housing households across the destination census tracts, we came up with the following categorization:
(1) nonreceiving; (2) low receiving, meaning 1 to 5 households; (3) medium receiving, meaning 6 to 12 households; and (4) high receiving, meaning more than 12 households.
Exhibits 4 and 5 show the average tract-level population and socioeconomic characteristics by level of receivership. Exhibit 4 provides this information for the city of Atlanta and includes crime information. Exhibit 5 shows the population and socioeconomic information for the suburbs (note that the crime data are not available for the suburbs). On average, all the receiving tracts differ substantially from the nonreceiving tracts. Based on 2000 census tract boundaries, of the 660 census Exhibit 3 Public Housing Relocation Receivership, by Census Tract City of Atlanta boundary Public housing Census tract boundaries Relocated resident destinations Nonreceiving Low receiving Medium receiving High receiving
tracts in the Atlanta metropolitan region, former public housing residents moved to 84, with 64 within the city limits. Thus, less than 10 percent of our sample moved outside the city limits, and those who did so typically relocated to tracts adjacent to the city boundaries with relatively similar neighborhood characteristics to the characteristics of the tracts to which those who relocated within the city moved. The average distance moved is only 3 miles. Thus, by far, most relocated residents are not far from the former public housing locations.
We begin with the average census tract characteristics by receivership for the city, shown in exhibit 4.
In terms of racial composition, low-receiving, medium-receiving, and public housing census tracts range from 71 to 75 percent African American. High-receiving tracts average 95 percent African American, and the average percentage African American across all levels of receivership is slightly more than 80. By contrast, nonreceiving tracts are 46 percent African American, and the citywide
percentage is 60. Thus, a clear pattern of racial segregation is apparent; former public housing residents are not moving to more racially integrated neighborhoods, and, in fact, the level of racial segregation is much greater in high-receiving tracts than in the public housing census tracts.
Nonreceiving neighborhoods have the lowest vacancy and medium-receiving neighborhoods have the lowest renter household percentages. Across the other receivership categories and citywide, however, little difference emerges except in the high-receiving tracts, where the proportion of rental households is about 20 percentage points more. Homeownership percentages are less in high- and low-receiving tracts and in the public housing tracts as compared with the percentages in nonreceiving tracts and citywide. Medium-receiving tracts, however, have a homeownership percentage equal to that of the city. The average across all receiving tracts is slightly less than 29 percent as compared with the citywide average of 37 percent and the nonreceiving tract average of slightly more than 41 percent.
Poverty percentages across all receiving and public housing tracts are greater than the citywide and nonreceiving tract percentages. Whereas the citywide poverty rate is 22.4 percent, and the nonreceiving tract rate about the same, the poverty rate for the low-receiving tracts is 33.5 percent, for medium-receiving tracts is 26.0 percent, and for high-receiving tracts is 29.0 percent, with an average across all receiving tracts of 29.6 percent. The poverty rate for the receiving tracts is between
7.5 and 12.0 percentage points less than for the public housing tracts. Like those of previous studies, our findings reveal that residents are moving to neighborhoods with less poverty than public housing. The widely accepted definition of low-poverty neighborhoods is 20 percent or less, however (Goetz, 2003). All levels of receivership neighborhoods in our study exceed this threshold by between 6.0 and 13.5 percent, and the citywide and nonreceiving figures also exceed this threshold.
Crime trends indicate some interesting patterns. Specifically, the nonviolent crime rate per 1,000 people is greatest (106) in the nonreceiving tracts and least in the medium-receiving ones (68).
By contrast, the citywide, low-receiving, and high-receiving rates are nearly equivalent (95 to 97), with the average across all receiving tracts at 81. The nonviolent crime rate for public housing is greater than for all the receiving and citywide tracts, at nearly 98. Similarly, the violent crime rate is least (17) in the medium-receiving neighborhoods even compared with crime rates in the citywide (21) and the nonreceiving (19) tracts. Low-receiving neighborhoods have a rate of 24 and high-receiving neighborhoods a rate of 26, with the average across all receiving tracts at 22. The rate in the public housing tracts is the greatest, at 27.
Exhibit 5 shows the suburbanwide population and socioeconomic characteristics compared with the averages for the receiving and nonreceiving census tracts. Because most residents stayed within the city limits, the numbers of households per suburban census tract varied little, with the average being three. Therefore, we simply categorize the suburban tracts as receiving or nonreceiving.
Findings indicate that, although the receiving tracts are more disadvantaged and racially segregated on average than both the nonreceiving and metropolitanwide tracts, they are less disadvantaged than public housing. In addition, residents who moved to the suburbs are living in far less disadvantaged and racially segregated tracts than those residents who moved within the city, which in large part reflects disparities that have existed between the urban and suburban regions of metropolitan areas around the country since suburbanization began in the 1950s. The fact that so few residents in our
study (10 percent) left the city proper suggests both structural barriers (such as lack of public transportation) and individual choices (reluctance to move too far away from existing social supports).
Exhibit 6 presents the first set of logistic regressions. Model 1 presents raw logits and standard errors from logistic regression of an easy relocation process. Residents of family projects are significantly more likely than residents of projects for seniors to have experienced an easy relocation process (b = 0.96). For each additional year of age, the probability of an easy relocation increases significantly, by 0.04 logits. Having a disability is significantly associated with 0.60 lower logged odds of an easy relocation process than the nondisabled. Having a chronic health condition is not associated with the ease of the relocation process, however. Each additional year living in public housing (tenure) is associated with a 0.07-logged odds reduction in experiencing an easy relocation process. Those experiencing financial strain have 1.26 lower logged odds of an easy relocation process than those with no financial strain. Having no friends in the public housing community is associated with 0.52 higher logged odds of an easy relocation process than those with friends in public housing community.
Notes: The reference categories are living in housing for seniors or people with disabilities, having a functional limitation or disability, having no chronic conditions, and having no financial strain. Raw estimates and standard errors in parentheses.
Models 2 through 5 explore whether the hard-to-house characteristics vary by type of housing project. This examination is important because the effect of relocation on residents of housing for seniors has received little examination. All noninteracted variables change little if at all, and therefore we will not reinterpret them but focus solely on the interactions. In model 2, we interact disability status with housing type. Having a disability is associated with 0.69 lower logged odds of an easy relocation process, but the odds did not differ significantly for residents of family housing versus housing for seniors.
Model 3 interacts financial strain by housing type. Experiencing financial strain is associated with
0.11 lower logged odds of an easy relocation process, but the odds did not differ significantly for residents of family housing versus housing for seniors. Model 4 interacts tenure in public housing with type of housing project. A significant difference emerges in the association of tenure in public housing with the relocation process for residents of family housing versus housing for seniors.
Each additional year of tenure in public housing reduces the logged odds of an easy relocation process for family housing residents compared with the corresponding odds for residents of housing for seniors. Model 5 interacts having no friends in public housing with type of housing project.
A significant difference emerges in the association of having no friends in public housing with the relocation process for residents of family housing versus housing for seniors. Those in family housing with no public housing friends had a 0.85-logged odds greater likelihood of an easy relocation process than residents of housing for seniors.
In sum, being older and from housing for seniors, having a disability, experiencing financial strain, and living a longer time in public housing decreased the probability of experiencing an easy relocation process. What are the consequences in terms of the new homes and neighborhoods of not experiencing an easy relocation process? Does it lead to worse home and neighborhood conditions?
Exhibit 7 presents mean differences in reported home and neighborhood conditions and objective census tract-level measures of neighborhood conditions between those who experienced an easy Exhibit 7 Predicting Home and Neighborhood Consequences of Easy Relocation Versus Not Easy Relocation Mean for Not Easy Mean for Easy F-Test Relocation Relocation Condition of postrelocation home 1.93 (0.85) 1.57 (0.77) 9.46* (1 = excellent, 2 = good, 3 = fair, 4 = poor) Satisfaction with postrelocation neighborhood 2.27 (1.28) 1.66 (1.12) 14.19** High former public housing receiving neighborhood 0.06 (0.23) 0.11 (0.31) 4.95* Neighborhood percent of homes more than 30 years old 62.76 (21.34) 68.17 (22.39) 1.22 Neighborhood percent of homes vacant 21.48 (8.68) 20.92 (9.14) 0.37 Neighborhood percent renters 52.43 (14.40) 51.08 (15.69) 0.12 Neighborhood percent living in same place less than 73.12 (14.03) 66.80 (16.44) 5.15* 10 years Neighborhood percent in poverty 32.09 (11.32) 31.16 (11.86) 0.27 Neighborhood percent female-headed households 18.63 (8.93) 20.35 (7.90) 4.06 Neighborhood percent unemployed 16.13 (6.94) 18.75 (5.50) 10.15* Neighborhood percent non-Hispanic African American 79.93 (26.29) 92.02 (11.00) 17.16*** * p ≤.05. ** p ≤.01. *** p ≤.0001.
Note: Raw estimates and standard errors in parentheses.
Cityscape 187Oakley, Ruel, and Reid
relocation and those who did not, along with the reported ANOVA test results. Resident evaluation of both home and neighborhood is significantly greater for those who experienced an easy relocation than for those who did not. Those who experienced an easy relocation also were significantly more likely to move into neighborhoods where at least 12 others from our sample moved compared with those who did not experience an easy relocation.
In terms of the census tract measures, no significant difference emerges between the neighborhoods of those who had an easy relocation and those who did not in terms of neighborhood percentage vacancy, percentage renters, and percentage older homes. Approximately 20 percent of homes were vacant in neighborhoods for each group. For each group, slightly more than 50 percent of all occupied homes were occupied by renters. Although the neighborhoods chosen by those experiencing an easy relocation had more older homes, the difference was not significant. A significant difference does emerge, however, in the level of neighborhood turnover. Those who experienced an easy relocation moved into neighborhoods with significantly less turnover (66.8 percent) than those who did not experience an easy relocation (73.0 percent).