Public Housing : Building segregation long term impacts

 The American public housing program, one of the largest and most controversial U.S. urban policies of the 20th century, involved the construction of approximately 1.4 million federally-funded units between the 1930s and the 1970s.

The program began during the Great Depression as a New Deal effort to alleviate housing shortages and provide economic stimulus. The Housing Act of 1937 expanded the program, explicitly defining its purpose as slum clearance and redevelopment, and the Housing Act of 1949 further expanded funding, framing public housing as a tool to remedy housing shortages, eliminate substandard housing, and provide a decent living environment for every American family.

However, the context shifted dramatically. Prominent historical accounts suggest the program became controversial because it created and entrenched racial and economic segregation, accelerating mid-century urban decline. Site selection was systematically targeted towards initially poorer, more populated neighborhoods with higher Black population shares, reflecting both slum clearance goals and highly racialized site selection politics. For instance, the Public Works Administration followed a "neighborhood composition rule" that formally segregated projects, reinforcing existing segregation or actively creating it. This political conflict, where white neighborhoods often blocked construction, led to the concentration of projects in predominantly Black areas. By the 1960s, working families moved out, income eligibility tightened, and projects increasingly concentrated the most disadvantaged households, leading to widespread disrepair and criticism of their austere design. Ultimately, the evidence suggests that despite intentions of revitalization, public housing reinforced existing patterns of economic and racial segregation and reduced long-run economic opportunity.

Site selection for American public housing projects was a non-random process characterized by targeting distressed areas and influenced heavily by racial politics.

Projects were systematically targeted towards neighborhoods that were initially poorer, more populated, and had higher shares of Black residents, reflecting both the program’s official goal of slum clearance and the racialized politics of housing policy in the mid-20th century. Specifically, neighborhoods that received public housing often had lower median incomes, lower rents, and higher unemployment rates in 1940.

Political dynamics consistently shaped these decisions, as historical case studies show white neighborhoods fiercely resisted construction, leading to the concentration of projects in predominantly Black areas. For example, a comparison of proposed-but-not-selected sites in Philadelphia found that rejected locations were initially much whiter than those that actually received projects.

Neighborhoods targeted for public housing were also more likely to be designated as "Hazardous" (redlined) by the Home Owners' Loan Corporation (HOLC) in the 1930s and were much more likely to be affected by Urban Renewal.

Crucially, the siting decisions reinforced existing residential segregation. Projects built in poorer and higher Black population share neighborhoods tended to have higher shares of Black residents themselves. Furthermore, early policy followed a “neighborhood composition rule” that formally segregated projects, entrenching existing segregation patterns. These dynamics ultimately contributed to public housing reinforcing existing patterns of economic and racial segregation and reducing long-run economic opportunity in the affected neighborhoods.

The causal effects of American public housing construction on the neighborhoods that received projects (treated neighborhoods) were estimated using a stacked matched difference-in-differences (DiD) approach. This methodology compared treated neighborhoods to matched control areas sharing similar pre-treatment characteristics within the same county.

The findings indicate that public housing construction had significant and persistent long-term effects on neighborhood trajectories:

1. Population and Demographics: Treated neighborhoods experienced large, persistent increases in total population, rising by approximately 15% in the immediate decade after construction and reaching 16–17% in subsequent decades. This growth was driven by a substantial increase in the total Black population (a 57.2% increase relative to controls). The construction led to significant increases in Black population shares, rising by approximately 5.8 percentage points in the long run (t=20), representing a 20.9% increase relative to the baseline share. This population influx was accompanied by a large average decline in the estimated private population (non-public housing residents), affecting both white and Black residents.
2. Economic Decline: Treated neighborhoods experienced substantial declines in median incomes and rents. Median income fell sharply, declining by 9.5% in the first decade and reaching 15.3% below controls by the third decade (t=20). Median rents also declined, reaching -10.1% by the third decade. Declines were also found in other measures of economic well-being, such as lower labor force participation rates and higher unemployment rates.

Overall, the evidence suggests that public housing construction accelerated demographic changes and socioeconomic decline, reinforcing existing patterns of economic and racial segregation in these neighborhoods.

The sources provide evidence consistent with neighborhood tipping dynamics occurring in response to public housing construction.

The analysis explored how the effects of public housing varied based on the initial racial composition of the treated and nearby neighborhoods. The hypothesis tested was that initial racial composition could influence the extent to which public housing construction precipitated white flight or overall racial transition.

Here are the specific findings regarding tipping dynamics:

• Tipping Range Defined: Neighborhoods were categorized based on their baseline Black population share measured 10 years before construction. The "medium" initial Black share group, considered to be in the tipping range, was defined as having a Black population share between 1% and 12%. This range corresponds to the tipping threshold estimated by Card, Mas, and J. Rothstein (2008).
• Response in Treated Neighborhoods: Treated neighborhoods (those receiving projects) that were initially almost entirely non-Black (less than 1% Black share) or were in the "tipping range" (1–12% Black share) saw sizable increases in Black population in response to public housing construction. In contrast, treated neighborhoods with high initial Black shares (12% or higher) saw no change in Black population, and sometimes even a slight increase in white population.
• Response in Nearby Neighborhoods (Spillovers): Nearby neighborhoods that were in the "tipping range" also saw sizable declines in white population. This finding suggests that public housing construction triggered racial tipping dynamics, leading to white resident outflows in surrounding areas that had an initial Black population share within this specific range.

Geographic spillovers (effects on nearby neighborhoods) resulting from the construction of American public housing projects were found to be relatively limited on average.

The methodology defined nearby tracts as those that shared a border with a treated tract, were not themselves treated, and lay within one kilometer of the nearest public housing project.

Key findings for these adjacent areas include:

1. Economic Spillovers: Nearby neighborhoods experienced modest declines in median incomes. Specifically, median income fell by 2.7% in the first decade and 4.8% in the second decade following construction. However, the sources found no effect on median rent in these spillover neighborhoods.
2. Demographic Spillovers (Average): On average, there was little evidence of significant changes in racial composition, population, or rents in nearby neighborhoods. They experienced small population declines overall, but no evidence of substantial changes in the Black or white population was found, suggesting that public housing construction did not, on average, precipitate large-scale white flight or racial transition in these surrounding areas.
3. Tipping Dynamics: Despite the limited average effect, the study found evidence consistent with neighborhood tipping dynamics. Nearby neighborhoods that were initially in the potential "tipping range" (between 1% and 12% Black population share) saw sizable declines in white population/outflows of white residents in response to public housing construction.
4. Opportunity Spillovers: While nearby tracts initially showed small adverse effects on upward mobility, these differences completely disappear when controlling for 1970 neighborhood characteristics (such as income and demographics). This suggests that the modest negative spillovers on opportunity occurred through earlier neighborhood changes, rather than through persistent or independent effects of being near the project itself.

Overall, these findings suggest that the program’s most pronounced negative effects were largely concentrated in the project neighborhoods themselves, challenging historical narratives that attributed broad urban decline to public housing.

The sources demonstrate that the long-term neighborhood effects of American public housing construction were heterogeneous and driven by specific mechanisms, particularly related to initial neighborhood racial composition, construction timing, and overlap with other urban policies.

Heterogeneity of Effects

1. Neighborhood Tipping Dynamics: The sources provide evidence consistent with neighborhood tipping dynamics.

   • Tipping Range Defined: Neighborhoods initially falling within a potential "tipping range" (defined as having a Black population share between 1% and 12%, based on prior research) exhibited different responses than others.
   • Treated Neighborhoods (Receiving Projects): Treated tracts that were almost entirely non-Black (<1% Black share) or were in the "tipping range" (1–12% Black share) experienced sizable increases in Black population. In contrast, treated neighborhoods with a high initial Black share (12% or higher) saw no change in Black population, and sometimes a slight increase in white population.
   • Nearby Neighborhoods (Spillovers): Nearby neighborhoods (adjacent to treated tracts) that were in the "tipping range" also experienced sizable declines in white population (white resident outflows), suggesting that public housing construction triggered racial tipping in these surrounding areas.

2. Timing of Construction: The neighborhood effects were found to be larger for projects built before 1960 compared to those built later.

   • The long-run Black population increase in public housing neighborhoods and the long-run white population exit in nearby neighborhoods were primarily observed for projects built in the early period (before 1960), but not for those built in the later period (1960 or later).

3. Urban Renewal Overlap: The long-run effects varied significantly depending on whether the tract was also subjected to the Urban Renewal program.

   • Treated neighborhoods that were also Urban Renewal tracts showed null effects on median income and population by race, suggesting the impact of urban renewal may have outweighed any effects of public housing.
   • In neighborhoods not targeted by urban renewal, the effects were similar to the main baseline analysis (substantial negative effects).

Mechanisms Driving Neighborhood Change

The sources outline several channels through which public housing might have affected neighborhoods:

1. Mechanical Compositional Change: Public housing mechanically altered the racial and socioeconomic composition of neighborhoods due to tenant selection policies and subsidized rents. The construction resulted in large, persistent increases in total population driven by the substantial increase in the total Black population (a 57.2% increase relative to controls).
2. Endogenous Sorting by Private Households: The arrival of public housing residents triggered endogenous sorting, prompting some existing private residents to relocate. The data show a large average decline in the estimated private population (non-public housing residents), affecting both white and Black residents, following construction. This decline in private residents is consistent with white flight in certain "tipping range" neighborhoods.
3. Social and Market Externalities (Opportunity): The construction generated negative social and market externalities that impacted long-run economic opportunity.
   • Low-income children raised in public housing tracts experienced significantly lower upward mobility and a higher incarceration rate compared to matched controls.
   • These adverse opportunity effects were geographically concentrated within the immediate project tracts, as effects found in nearby tracts disappeared after controlling for neighborhood characteristics like income and demographics in 1970. This implies that the spillovers on opportunity occurred through earlier neighborhood changes, rather than through a persistent or independent effect of proximity to the project.
4. Reinforcement of Segregation: The siting dynamics and project demographics reinforced existing residential segregation. Public housing projects largely matched the racial composition of their surrounding neighborhoods, rather than disrupting existing patterns of segregation. Projects built in poorer neighborhoods tended to be more heavily Black.

The sources indicate that the construction of American public housing projects reduced long-run economic opportunity for children from low-income families. These adverse effects were found to be geographically concentrated within the project neighborhoods themselves.

To assess long-run opportunity, the research linked the public housing data to the Opportunity Atlas. The analysis focused on low-income children (those whose parents were at the 25th percentile of the national income distribution) born between 1978 and 1983.

Effects on Public Housing Tracts

Children who grew up in the public housing neighborhoods (treated tracts) experienced significantly worse outcomes compared to matched control neighborhoods:

• Upward Mobility: Children had a 1.7 percentage point lower income rank in adulthood (2014-2015).
• Incarceration: Children had a 0.5 percentage point higher incarceration rate as of April 1, 2010.

About half of these adverse effects on income rank and incarceration persisted even after statistically controlling for 1970 neighborhood characteristics (such as Black share and median income). This suggests that public housing impacted opportunity both by altering observable neighborhood composition and through additional effects of the projects themselves.

Geographic Spillovers on Opportunity

The sources found that the adverse effects on upward mobility were geographically concentrated within immediate project tracts.

• Nearby Neighborhoods: While nearby tracts initially showed small adverse effects on upward mobility, these differences completely disappear when controlling for 1970 neighborhood characteristics (income and demographics).
• Interpretation: This result implies that nearby neighborhoods experienced worse outcomes simply because they were poorer and more Black by 1970, not because proximity to public housing directly reduced opportunity through persistent or independent effects.

The sources caution that since pre-treatment mobility outcomes are unobservable, these results should not necessarily be interpreted as purely causal estimates of public housing on upward mobility.

The construction of public housing projects had significant adverse effects on the long-run economic opportunity of low-income children who grew up in the treated neighborhoods.

The research focused on low-income children, defined as those whose parents were at the 25th percentile of the national income distribution, born between 1978 and 1983.

Specific Outcomes

1. Upward Mobility (Income Rank): Children raised in public housing tracts experienced a 1.7 percentage point lower income rank in adulthood (measured in 2014–2015) compared to children in matched control neighborhoods.
2. Incarceration Rates: These children also had a 0.5 percentage point higher incarceration rate as of 2010.

Concentration and Mechanisms

• Persistence of Effects: Approximately half of the adverse effects on income rank and incarceration persisted even after statistically controlling for 1970 neighborhood characteristics (such as Black share and median income). This suggests that public housing reduced opportunity not only by changing observable neighborhood composition but also through additional effects of the projects themselves.
• Geographic Concentration: The adverse effects on upward mobility were geographically concentrated within immediate project tracts. Any small initial adverse effects found in nearby (spillover) tracts completely disappeared once 1970 neighborhood characteristics were controlled for. This indicates that nearby neighborhoods experienced worse outcomes because they were already poorer and more Black by 1970, not because proximity to public housing directly reduced opportunity.

These findings demonstrate that public housing significantly reduced the long-run economic opportunities for low-income children residing in the project areas.

The supplementary analysis using high-resolution land value data in Chicago showed that project size was a source of substantial heterogeneity in neighborhood impacts.

Projects were categorized based on a median project size of 202 units, comparing 55 smaller projects to 18 larger developments. The research found evidence that smaller projects seemed to have more persistent positive effects on local land values. This land value analysis, performed using a stacked spatial difference-in-differences design in concentric rings around projects, contrasts the broader finding that public housing, on average, showed positive effects on local land values, though these average effects mostly did not reach statistical significance.