Panel Paper: Housing Assistance and Neighborhood Disadvantage in Families with Children

Friday, November 9, 2018
Jackson - Mezz Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Sarah Gold, Princeton University


The public housing program provide rental assistance to about two million households in the United States and impact the neighborhoods to which families have access. Compared other low-income families’ neighborhoods, families with public housing typically live in neighborhoods with higher poverty rates and minority concentration. The effects of neighborhoods can persist across a range of outcomes including cognitive skills, academic achievement, and economic mobility.

This paper uses longitudinal data to examine the quality of neighborhoods to which families receiving public housing have access and whether neighborhood quality differs within families, depending on housing assistance receipt.


This project utilizes the Panel Study of Income Dynamics (PSID), a large, nationally representative panel survey of households in the United States. In addition to the detailed longitudinal data available in the PSID core, the PSID’s restricted-use Assisted Housing Database has matched standardized addresses for each family unit through 2009 to records of assisted housing to verify housing assistance receipt. Neighborhood characteristics are drawn from the Longitudinal Tract Database.

The study sample is limited to low-income, renter families with children born between 1970 and 1992. Families are categorized as low-income if their average income from the child’s birth to age 15 is at or below 50% of the area median income. Housing assistance at each wave is coded categorically (no assistance; public housing; other assistance). Both a continuous measure of neighborhood poverty and a combined measure of high census-tract level poverty and minority concentration are used as outcome variables.

The associations between housing assistance and neighborhood disadvantage are explored, taking advantage of the panel nature of these data, and controlling for a comprehensive set of covariates at the individual-, family-, and household-levels. In all models, housing assistance is lagged by two years to address potential reverse causality. Associations of housing assistance type and neighborhood disadvantage are estimated using pooled cross-sections (n=6,348 with observations on 945 children) of the data and adjusting standard errors for clustering at the individual level. Individual fixed effects models are incorporated to address unobserved differences between those with and without housing assistance.


Results from both pooled cross-sectional and fixed effects models indicate that children living in public housing live in neighborhoods with higher poverty rates than low-income children without housing assistance (11.2 and 4.3 percentage points, respectively). Random effects models show that children living in public housing reside in neighborhoods that are 4.3 percentage points more racially and economically segregated that children not receiving housing assistance. The fixed effects models suggest that double segregation in which children living in public housing reside does not differ from the neighborhoods in which they lived without housing assistance.


These findings would suggest that neighborhood-level improvements be made around public housing.