Place Matters: Why the Location of Low-Income Housing Projects is Important to Neighborhood Revitalization

The next two charts show the variation in clustering (where project-based assistance represents a high share of all housing in the neighborhood). Among low-income tracts, we find, at one extreme, 800 tracts where clustering is substantial. In these tracts, HUD projects (with 441,700 units) are a dominating presence, accounting for 30 percent or more of all units. What happens to these projects, for good or ill, will have a powerful influence on the future of their neighborhoods overall.

In 1,400 tracts, project-based units represent from 10 to 30 percent of the total housing (not so dominant but still influential). HUD-assisted projects account for less than 10 percent of all units in 2,000 tracts. Most noteworthy, however, is that 5,500 tracts (well over half of all low-income tracts) have no HUD project-based units at all, meaning that HUD projects will not be a factor in neighborhood revitalization.

Among moderate- and high-income neighborhoods (those with poverty rates below 20 percent) HUD-assisted projects are less prevalent: 81 percent of the tracts do not have any project-based units and in most others (15 percent), such units represent less than 10 percent of the housing stock.

We also found that the share of HUD-assisted projects varies greatly among metropolitan areas. At the high end, these projects accounted for 4.5 percent or more of all housing in metropolitan Providence, New York, St. Louis, Seattle, and Boston. The absolute size of metropolitan New York’s project-based stock stands out-with 357,000 occupied units, New York has more than four times the total of its nearest rival (Chicago). At the low end, project-based units accounted for less than 0.8 percent of all units in metropolitan Dallas; Riverside-San Bernardino, California; Boise City, Idaho; Phoenix; and Bradenton, Florida.

This distribution of project-based assistance is poorly correlated with metropolitan poverty rates. Of the 100 metros we studied, five of the top 20 in terms of poverty ranked among the lowest 20 in terms of project-based assistance: Bakersfield, California; San Antonio; Tucson; Albuquerque; and Houston. This result is not explained by HUD allocations policies.  Differences in the degree to which developers and housing authorities sought HUD projects were most important; some cities (more often in the snowbelt) were aggressive in seeking this form of assistance, while others (often in the sunbelt) generally avoided it.  Performance also played a role; HUD reallocated funds from cities unable to complete approved projects to those with better production capacity.

Metros also vary widely in the densities of project-based units in low-income neighborhoods. Density is a useful measure because it indicates the degree to which HUD projects are likely to be important to neighborhood revitalization planning (which focuses on low-income neighborhoods). Metropolitan New York tops this list with 20 HUD project units per 100 total units in low-income tracts. Metropolitan Boston; Honolulu; Washington, DC; and Worcester, Massachusetts come next with 15 to 19 project-based units per 100. At the other end of the spectrum are five metros with two or fewer HUD project units per 100: Riverside-San Bernardino; Bakersfield, California; McAllen, Texas; Boise City; and Scranton, Pennsylvania.