Does Improving Housing Quality Reduce Health Care Utilization?

Dragan (2026) examines whether improving housing quality reduces health care use and spending among low-income residents. She asks whether a large New York City housing remediation policy—the Alternative Enforcement Program—led to changes in health care utilization. Using Medicaid enrollment and claims data from 2007–2019 linked to building-level housing violation records, she applies a regression discontinuity design around the program’s eligibility cutoff. She finds no evidence of short-run reductions in emergency department visits or health care spending, despite housing violations falling by roughly 50%. Longer-run analyses show modest declines in emergency department visits of about 10–15% after three to four years, with no spending reductions.

Why This Article Was Selected for The Policy Scientist

This article addresses a policy question of broad and growing importance: whether investments in housing quality can realistically reduce health care utilization and spending. As health systems increasingly fund nonmedical interventions, credible evidence on downstream health effects is essential. Dragan has written extensively on housing, health, and social policy, and this study fits squarely within that body of work. The article is timely given Medicaid and Medicare pilots that assume near-term cost savings from housing remediation. The linked administrative housing and Medicaid claims data are unusually rich, and the New York City setting provides scale. The regression discontinuity design is a strong causal approach, strengthening confidence in the null short-run findings and their relevance beyond this jurisdiction.

Full Citation and Link to Article

Dragan, K. L. (2026). Housing quality improvement and health care utilization: A regression discontinuity study. Journal of Policy Analysis and Management, 45(1). https://doi.org/10.1002/pam.70074 

Central Research Question

The central research question of this article is whether large-scale housing quality remediation can causally reduce health care utilization and spending among low-income residents. Specifically, the study asks whether New York City’s Alternative Enforcement Program (AEP)—which mandates intensive remediation in the city’s worst-quality private apartment buildings—leads to short-run or longer-run reductions in emergency department use, housing-sensitive health visits, or Medicaid expenditures. The question is motivated by the growing use of housing interventions by health payers, particularly Medicaid and Medicare, which often justify such programs on the expectation of near-term health care cost savings. The article seeks to move beyond correlation by estimating whether improving housing conditions in extremely distressed buildings actually alters patterns of health system use.

Previous Literature

The existing literature documents strong cross-sectional associations between poor housing quality and adverse health outcomes, including asthma, injuries, mental health conditions, and lead exposure. Prior studies consistently show that low-quality housing is correlated with higher health care utilization, especially emergency department visits. However, much of this evidence relies on observational designs that cannot rule out confounding or reverse causality, such as health-based sorting into poorer housing.

Quasi-experimental studies and randomized trials examining housing improvements are comparatively rare and have produced mixed results. Some quasi-experimental studies suggest modest improvements in health or utilization following housing upgrades, particularly for children or narrowly defined interventions. Randomized controlled trials of specific remediation efforts—such as insulation, pest management, or air filtration—often find improvements in self-reported symptoms or indoor air quality but little evidence of changes in clinical outcomes or utilization. Many prior studies focus on public housing, redevelopment rather than remediation, or bundled interventions that include health education, limiting generalizability. This study contributes by evaluating a real-world, large-scale policy affecting privately owned housing and by applying a strong causal inference design.

Data

The analysis links multiple administrative datasets spanning 2007–2019. Individual-level Medicaid enrollment, claims, and encounter data for New York City residents are merged with building-level housing data from the Department of Housing Preservation and Development. Housing data include detailed records on maintenance violations, emergency repair charges, and building characteristics. Buildings are ranked annually according to a formal algorithm used to determine AEP eligibility. Medicaid enrollees are assigned to buildings based on verified residential addresses at the start of each program year.

The resulting dataset includes over 48,000 Medicaid enrollees living in buildings considered for AEP, with approximately 24,000 individuals falling within the optimal bandwidth around the treatment cutoff. Outcomes include emergency department visits, visits for housing-sensitive conditions, total health care expenditures, Medicaid enrollment continuity, and residential mobility. The administrative nature of the data minimizes recall bias and measurement error in utilization and spending.

Methods

The study employs a regression discontinuity design that exploits the sharp cutoff used by the city to select buildings for AEP. Buildings ranked just above the threshold are required to remediate serious housing violations, while those just below are not, creating quasi-random assignment near the cutoff. The identifying assumption is that buildings and tenants on either side of the threshold are otherwise comparable.

Local linear regressions with triangular kernel weights are used within an empirically selected bandwidth of 25 ranks. The primary estimand is the local average treatment effect for tenants living in buildings at the margin of selection. Models adjust for demographic characteristics, baseline health spending, building characteristics, borough, and program year to improve precision. Robust standard errors are bias-corrected and clustered at the building level.

The primary analysis focuses on outcomes in the 12 months following program selection, reflecting the time horizon emphasized by health payers. Secondary analyses examine longer-run outcomes up to four years later using both regression discontinuity extensions and an exploratory event study framework.

Findings/Size Effects

The housing remediation program substantially improved measured housing quality. Total and severe housing violations declined by roughly 50% within one year of program entry, confirming strong first-stage effects. Despite these improvements, the study finds no evidence of short-run reductions in health care utilization or spending. The probability of any emergency department visit, total emergency department visits, visits for housing-sensitive conditions, and total Medicaid expenditures all show estimated effects close to zero, with confidence intervals ruling out large reductions.

Subgroup analyses focusing on children, individuals with chronic respiratory or mental health conditions, and residents of buildings with mold or pest violations yield similarly null short-run results. Sensitivity analyses using alternative bandwidths, functional forms, and control groups do not alter the conclusions.

Longer-run analyses provide limited evidence of delayed effects. Event study estimates suggest reductions in total emergency department visits of approximately 10–15% three to four years after intervention, corresponding to about 7–10 fewer visits per 100 tenants annually. However, these reductions do not translate into detectable decreases in overall health care spending at any time horizon examined.

Conclusion

This study provides rigorous causal evidence on a central assumption underlying health sector investments in housing quality: that remediation will generate short-run health care savings. Using high-quality administrative data and a strong regression discontinuity design, the article finds that even intensive remediation in extremely distressed housing does not produce immediate reductions in health care utilization or expenditures. The findings align with prior randomized and quasi-experimental studies that show limited effects of housing improvements on clinical utilization, despite improvements in living conditions.

The results suggest that pathways from housing quality to health care use are long and complex, and that modest remediation may be insufficient to reverse cumulative exposure to disadvantage. While housing improvements may yield important welfare and quality-of-life benefits, the evidence here indicates that expectations of near-term cost savings should be treated cautiously. The study represents a substantial contribution to the housing and health literature by applying a credible causal design to a large-scale, real-world policy affecting private housing, with implications for similar programs in other jurisdictions.