Panel Paper: Does the Value of Housing Assistance Impact Health Outcomes?

Thursday, November 8, 2018
Harding - Mezz Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Daniel Miller1, Thomas Byrne2 and Margaret M. C. Thomas1, (1)Boston University, (2)U.S. Department of Veterans Affairs

Housing is a key social determinant of health, and the array of U.S. Department of Housing and Urban Development (HUD) low-income housing assistance programs represent a potentially powerful policy lever to promote population health and reduce health disparities. Yet, research on how federal low-income housing programs impact health remains highly limited. Much previous research has considered housing assistance in its most simplistic form: as a yes-no indicator. Because HUD-assisted households typically pay 30% of their income towards rent, the monetary value of housing assistance varies across households as a function of household income, the prevailing market rents that HUD-assisted households would otherwise have to pay, and the rents of units in which assisted household reside. Variation in the value of housing assistance is likely to translate into meaningful differences in health outcomes. Thus, this study addresses the following questions: 1) Does the value of housing assistance impact health outcomes? and 2) Does this impact vary across federal housing assistance programs?

To address these questions, we capitalize on a newly available dataset that links responses from the nationally-representative National Health Interview Survey (NHIS) for the period from 1999 to 2012 with HUD administrative records from 1999 to 2014. We use these data to identify a sample of nearly 59,000 NHIS respondents who were 1) current or 2) future recipients of HUD-assisted housing or 3) who met income eligibility criteria for HUD housing at the time of their participation in the NHIS. We create novel measures of both the “explicit” and “implicit” value of HUD housing assistance by combining information from the HUD records on tenant income, tenant and HUD rent payments and data on prevailing fair market rents. We adjust these measures both for inflation using the Consumer Price Index and to account for geographic differences in cost of living by using the Census’ Supplemental Poverty Measure.

We use two analytic approaches to assess the impact of the value of housing assistance on adult and child measures of overall health and mental health, chronic health conditions, acute health conditions, and access to health care. First, in following prior research, we compare NHIS respondents who were in HUD-assisted housing at the time of their interview with those who subsequently entered HUD-assisted housing within two years. Second, we employ instrumental variables using the local supply of HUD-assisted housing as an instrument for the value of housing assistance. In both approaches, we model the impact of the value of any type of housing assistance, and test whether there are differential effects of this value across the HUD’s Housing Choice Voucher, public housing and multi-family housing programs. Study findings stand to have high policy relevance, particularly in the context of recent proposals to increase the rent contribution of HUD-assisted housing tenants from 30% to 35% of their income. The study also has broader long-term implications for informing how the existing (limited) resources dedicated towards housing assistance can be used in the most efficient and effective manner to promote population health.