Federal Housing Assistance Programs, Food Insecurity and Health Care Hardship among Children in the United States: Differences Across Rental Market Affordability Contexts

Technical Summary: Federal Housing Assistance Programs, Food Insecurity, and Health Care Hardship among Children in the United States

Problem Statement
Food insecurity and health care hardship represent critical public health challenges for families with children in the United States, often exacerbated by a worsening rental housing affordability crisis, particularly in urban areas. While housing costs constitute the largest share of household budgets for low-income families, forcing trade-offs between rent and essential needs like food and medical care, the specific impact of federal housing assistance on child health outcomes remains underexplored. Existing literature has established links between housing assistance and adult food insecurity and unmet health needs, as well as various child health metrics (e.g., asthma, lead levels), but lacks evidence regarding child-specific food insecurity and health care hardship. Furthermore, it is unclear whether the benefits of housing assistance are uniform across communities or if they are amplified in areas with severe rental affordability constraints, where low-income renters dedicate a disproportionate share of income to housing.

Methodology
This study utilizes a linked dataset combining the nationally representative National Health Interview Survey (NHIS) from 1999–2018 with U.S. Department of Housing and Urban Development (HUD) administrative records from 1996–2019. The analysis focuses on children aged 0–17.

• Study Design: To address selection bias inherent in observational studies of housing assistance, the authors employ a "pseudo-waitlist" comparison group. Current HUD assistance recipients (those residing in HUD housing at the time of the NHIS interview) are compared against future recipients (those who entered HUD-assisted housing within two years of the interview). This approach leverages the average two-year wait time for assistance to create a comparable control group.
• Variables:
  • Outcomes: Measures include food insecurity (using USDA 10-item module), various indicators of health care hardship (delayed care, unaffordable care for child or family members, unaffordable prescriptions/mental health care), and a composite measure of unmet health needs due to cost. Overall child health is based on parent-reported status.
  • Independent Variables: The primary independent variable is current receipt of HUD assistance. The study further stratifies by program type: Housing Choice Voucher (HCV) versus a combined "project-based" category (Public Housing and Multifamily).
  • Contextual Stratification: Counties are categorized into "low," "moderate," or "severe" rent burden strata based on the share of cost-burdened renters (spending >30% of income on housing).
  • Covariates: Models adjust for child demographics (age, sex, race/ethnicity), family characteristics (type, education, employment), and economic factors (income, SNAP receipt, health insurance, Census region).
• Analysis: Binary logistic regression models are estimated for each outcome. Separate models are run for the full sample and stratified by rent burden levels and program type.

Key Results

• Overall Impact: Current HUD recipients demonstrated significantly lower odds of food insecurity (28% lower odds) compared to future recipients. They also showed significantly lower odds of various health care hardships, including a 19% reduction in the odds of having any unmet health need due to cost. No significant association was found for overall self-rated health in the aggregate model.
• Heterogeneity by Rent Burden: The protective effects of housing assistance were most pronounced in counties with "severe" rent burden. In these areas, current recipients had a 34% reduction in the odds of food insecurity and a 31% reduction in the odds of any unmet health need due to cost. In contrast, associations were generally non-significant in counties with low or moderate rent burdens.
• Program Type Differences:
  • Housing Choice Vouchers (HCV): HCV recipients showed significant reductions in food insecurity and health care hardship, particularly in severe rent burden counties. In these areas, HCV receipt was associated with an approximate 40% reduction in the odds of both food insecurity and very low food security, as well as a 30% reduction in unmet health needs due to cost.
  • Project-Based Housing: Findings were less consistent. While project-based housing was associated with reduced odds of delaying care and unaffordable medical needs in severe rent burden counties, it showed mixed results in other strata, including a counter-intuitive association with higher odds of fair/poor health in low rent burden counties.
• Statistical Significance of Interaction: While coefficients were consistently larger in magnitude for severe rent burden strata, interaction terms between rental affordability strata and HUD assistance receipt were not statistically significant in pooled models. Thus, the authors caution that while the effect appears larger in severe areas, they cannot definitively conclude the association differs significantly across strata.

Key Contributions and Significance
The paper claims to be the first study to utilize linked NHIS-HUD administrative data to examine child health care hardship and the first to investigate whether the benefits of federal housing assistance vary by local rental market affordability.

• Mechanism of Action: The findings support an income-driven mechanism. Because HUD programs cap tenant payments at 30% of income, the assistance generates a stronger "income effect" in high-cost markets where non-assisted families face steeper housing cost burdens. This freed-up income appears to translate into reduced food insecurity and improved access to health care, particularly in the most expensive rental markets.
• Policy Implications: The results suggest that housing assistance, particularly the HCV program, is a potent tool for mitigating material hardship among children. The study indicates that the benefits of such assistance may be maximized when targeted toward communities facing the most severe affordability crises.
• Novelty regarding HCV: Contrary to prior work linking HUD data to NHANES (which found only project-based housing reduced food insecurity), this study finds that HCVs are associated with a substantial reduction (approx. 40%) in food insecurity and very low food security, specifically among children in the most cost-burdened areas.

Limitations
The authors acknowledge that while the pseudo-waitlist design addresses major sources of endogeneity, selection bias may persist. Additionally, all outcome measures rely on survey self-reports, which are subject to misreporting. Although the study finds larger effect sizes in severe rent burden areas, the lack of statistically significant interaction terms means the authors remain modest in claiming that the association differs significantly across strata, rather than simply being larger in magnitude.