Location Efficiency and Affordability: A National Analysis of Walkable Access and HUD-Assisted Housing

Abstract

As walkable neighborhoods are rapidly gaining popularity, these location-efficient areas are becoming less affordable to low-income tenants. We ask to what extent project- and tenant-based federal housing assistance is keeping these areas affordable and whether tradeoffs exist. Using descriptive statistical and logistic regression analysis for a data set of 3.8 million U.S. Department of Housing and Urban Development (HUD) tenants and a variety of neighborhood-level indicators, we find that HUD assistance provides tenants with differential access to walkable neighborhoods. Tenants who are senior, Asian, White, or have disabilities have higher chances of living in higher opportunity walkable areas. However, for those tenants with the greatest disadvantages (African American and Hispanic tenants), neighborhood quality remains compromised by higher poverty, segregation, and worse school quality, even in walkable neighborhoods. We identify the type of assistance (public housing, project-based rental assistance, and Housing Choice Vouchers) that is associated with compromised or higher opportunity access for these groups. This information can help prioritize assisted housing counseling, preservation, and siting to reduce existing spatial inequalities related to walkable amenity access, especially for African American and Hispanic tenants. This research also helps advance emerging research on the conceptualization and measurement of neighborhoods that integrates urban form and socioeconomic indicators.

Keywords:

• Location efficiency
• walkable access
• affordability
• HUD-assisted housing
• built urban form
• neighborhood quality

Acknowledgments

The authors would like to thank the editor and reviewers for detailed feedback, which helped improve the quality of the article. The work that provided the basis for this publication was supported by funding under an award from the U.S. Department of Housing and Urban Development. The substance and findings of the work are dedicated to the public. The authors are solely responsible for the accuracy of the statements and interpretations contained in this publication. Such interpretations do not necessarily reflect the views of the government. We also gratefully acknowledge research contributions by Dr. Sungduck Lee, Eva (Yue) Zhang, and Dr. Daniel Arribas-Bel.

Notes

1. Unfortunately, we were not able to distinguish traditional public housing from HOPE VI in this data set.

2. Several commercial vendors sell such crime data, but the estimates we examined were derived from models that interpolated data from the citywide Uniform Crime Reports to the neighborhood level, which was not reliable enough for our analysis.

3. Based on the 2003 Office of Management and Budget definition of core-based statistical areas (Office of Management & Budget, 2003).

4. The data were purchased from Geolytics since we compared the 2005–2009 poverty rates with those of 2000 to assess changes (Geolytics, Inc., 2010).

5. To differentiate urban and suburban areas, the following definitions are applied. The 2010 Census defines 1,308 principal cities of metropolitan or micropolitan statistical areas (The pcicbsa10 variable in the 2010 Census Designated Place, 2010 tiger/Line Shapefile, U.S. Department of Commerce, U.S. Census Bureau, Geography Division, Publication_Date: 2010; 2010 Census Designated Place, 2010 tiger/Line Shapefile, http://www.ofm.wa.gov/pop/geographic/tiger10/metadata/cdp10.html). These include cities, towns, villages, boroughs, and other municipalities. This analysis is based on the subset of 1,187 principal cities in metro areas that the 2010 Census identifies as cities (i.e., excluding towns or villages; The lsad10 variable in the 2010 Census Designated Place, 2010 tiger/Line Shapefile, U.S. Department of Commerce). For the purpose of this analysis, all other neighborhoods outside of these cities but within the metro area are identified as “suburban” if there are no housing units in rural parts of the neighborhood (i.e., any neighborhoods with rural housing units are excluded here; The h2 variable in the 2010 Census Designated Place, 2010 tiger/Line Shapefile, U.S. Department of Commerce).

6. The sample size (n) and pseudo-R² value are reported for each model. The popular R² measure of explained variation in Ordinary Least Squares (OLS) regressions does not exist for logistic regressions since they are estimated by maximizing the likelihood function. A pseudo-R² measure is used instead (McFadden’s pseudo-R²). It indicates how much a model is improved by including the predictors versus only the intercept. Specifically, it is computed as 1 minus this term: the estimated maximum log likelihood value of the model with predictors divided by the estimated maximum log likelihood value of the model without predictors. The logistic regression models are estimated in Stata. A larger pseudo-R² value indicates a better model fit, but small values are common. In contrast to the regular OLS R² values, these pseudo-R² values should not be compared across different data sets (in our case, between the three HUD programs) but can be compared for the same data set (within the same HUD program). The models predicting the odds of living in a segregated neighborhood with higher poverty rates and no walkable access have the highest pseudo-R² values for all three HUD programs (0.11–0.17). The pseudo-R² values especially for project-based housing in low-poverty areas are very low (0.04–0.05), as are the pseudo-R² values for the voucher models. However, all but one of the odds ratios are significant at the 0.01 level.

7. Based on a separate analysis of the HUD and 2010 Census data based on walkable access unit size, density, and urban/suburban locations. Results are available from the authors upon request.

8. The table with these results was removed based on reviewer feedback to avoid overloading the article with tables, but is available from the authors upon request.

9. Housing program membership came from the HUD data set whereas the population data were obtained from the 2010 Census.

10. To understand how this percentage is derived: Of all African American public housing residents who live in walkable areas in the Northeast, 10% live in low poverty and 90% in higher poverty areas.

11. These odds are higher than those for seniors in similar neighborhoods without walkable access. This is consistent with the fact that PBRA tenants in walkable areas are the oldest (see Table 1). However, PBRA has the smallest share of seniors, followed by public housing and HC Vouchers (Table 1).

12. For African American tenants with PBRA, these odds stay the same in areas with higher poverty, segregation, and no walkable access. In contrast, for Hispanic PBRA tenants, the odds drop to 0.8 in these neighborhoods without walkable access. In other words, African American PBRA tenants are 3 times more likely to live in segregated higher poverty neighborhoods regardless of whether they have walkable access. Hispanic PBRA tenants are only more likely to live in higher poverty segregated neighborhoods that also have walkable access. Asian tenants with PBRA are most likely to live in any neighborhood with walkable access. But the likelihood that this access is compromised by high poverty is higher than that of living in a walkable low-poverty neighborhood (2.4 vs. 1.8).

13. walkableneighborhoods.org/explore.

14. In separately published research (Talen and Koschinsky, 2014a), we find variations in these patterns between cities: For instance, Boston, Massachusetts, and Seattle, Washington, provide better neighborhood access than Chicago, Illinois, and Miami, Florida, do, whereas Atlanta, Georgia, and Phoenix, Arizona, provide the worst access. High-access neighborhoods were compromised by segregation in Atlanta, Boston, and Chicago but not in Miami, Phoenix, and Seattle. Subsidized housing in low-poverty areas generally had worse walkable access, and 61% of HUD units in these cities were burdened with both low access to services and low transit access.

15. For instance, see Bredderman (2015) on problems with the nation’s largest public housing provider, the New York City Housing Authority.

16. For an overview of patterns in each of the 359 metropolitan areas, see our website: http://walkableneighborhoods.org/explore/