Does Health Insurance Improve Access to Health Care?

Anderson et al. (2025) examine how gaining health insurance affects access to and use of medical care among uninsured near-elderly adults. They use 2008–2017 data from the American Community Survey, Behavioral Risk Factor Surveillance System, and hospital discharge records, exploiting Medicare eligibility at age 65. They find that gaining insurance reduces foregone care by 53.9 percentage points, increases having a regular medical provider by 32.8 points, and raises the likelihood of a recent checkup by 41.2 points. Insurance also produces 13.2 additional hospitalizations per 100 people, including 3.6 emergency and 7.4 elective admissions.

Why This Article Was Selected for The Policy Scientist

Health insurance remains a central policy issue because coverage can determine whether people obtain routine care, receive timely treatment, and undergo procedures that affect health and quality of life. The topic is particularly timely as policymakers continue to debate the affordability and reach of public insurance. This study makes an important contribution by isolating the effect of gaining insurance from simply changing insurance plans, extending earlier Medicare research by Dobkin and Maestas and others. Its large national survey and administrative datasets provide strong evidence across states, although generalization beyond the United States requires caution. The causal-inference strategy, combining regression discontinuity with instrumental variables, is especially strong.

Full Citation and Link to Article

Anderson, M., Dobkin, C., Maestas, N., & Rose, L. (in press). Health insurance and access to care for the near elderly. American Economic Journal: Economic Policy. American Economic Association. https://www.aeaweb.org/articles?id=10.1257/pol.20230644

Central Research Question

The central research question is whether gaining health insurance causally increases access to health care and medical utilization among uninsured Americans approaching age 65. The analysis focuses on people who become eligible for Medicare at 65, when insurance coverage changes abruptly. The central empirical difficulty is that Medicare eligibility produces several changes simultaneously: some previously uninsured people gain insurance, while many already-insured people replace private coverage with Medicare or add Medicare to another source of coverage. The study therefore seeks to isolate the effect of becoming insured from the effects of merely changing the type or generosity of existing insurance. It also examines how much gaining coverage closes preexisting gaps in medical access and utilization between insured and uninsured near-elderly adults.

Previous Literature

A substantial literature examines how health insurance affects access, utilization, health, and financial security. The RAND Health Insurance Experiment (Manning et al., 1987) and Oregon Health Insurance Experiment (Finkelstein et al., 2012) provide especially important experimental evidence. However, both included relatively few near-elderly participants, limiting their statistical power for studying this population. RAND also did not include an uninsured control group suitable for estimating the effect of moving directly from no insurance to insurance.

The study builds particularly closely on research exploiting Medicare eligibility at age 65. Card, Dobkin, and Maestas (2008) documented substantial changes in health-care utilization when people become Medicare eligible, while Card, Dobkin, and Maestas (2009) examined Medicare's effects on mortality. McWilliams et al. (2003, 2004, 2007a, 2007b) similarly examined changes in access, utilization, and health surrounding Medicare eligibility. Other research has produced less evidence of improvements in health or mortality, including Polsky et al. (2009) and Black et al. (2017).

A limitation of this literature is that turning 65 changes coverage for both uninsured and insured people. Consequently, conventional regression-discontinuity estimates around age 65 combine the effect of gaining insurance with the effects of switching insurance arrangements. This distinction became particularly important after the Affordable Care Act reduced uninsurance unevenly across states. The authors also draw methodologically on Kline and Walters (2016), who addressed the problem of separating multiple treatment channels generated by a single policy intervention.

Data

The empirical analysis combines several unusually large datasets covering 2008 through 2017. The American Community Survey (ACS) provides information on insurance coverage and allows respondents to report multiple forms of insurance. The authors use these data to estimate how insurance arrangements change precisely at age 65 within states and across time. They classify the principal transitions into three categories: uninsured to Medicare; non-Medicare insurance to Medicare; and non-Medicare insurance to Medicare plus another source of coverage.

The Behavioral Risk Factor Surveillance System (BRFSS) provides measures of access to medical care. Outcomes include whether respondents report foregoing medical care because of cost, whether they have a regular medical provider, and whether they received a checkup during the previous year. BRFSS access measures are available across all 50 states and the District of Columbia.

Administrative hospital records from the Healthcare Cost and Utilization Project (HCUP) provide inpatient utilization measures for 23 states. These records permit the researchers to distinguish among total, emergency, urgent, and elective hospital admissions. Combining survey and administrative data provides both broad measures of access and objective evidence of actual medical treatment.

Methods

The research design exploits the sharp increase in Medicare coverage when individuals turn 65. A conventional fuzzy regression-discontinuity design could compare changes in insurance with changes in medical utilization at that threshold. The complication is that approximately 7 percent of people move from being uninsured to insured at age 65, while much larger percentages of already-insured people also change insurance arrangements. Treating all utilization changes as consequences of newly obtaining insurance would therefore violate the single-channel assumption underlying a conventional instrumental-variables interpretation.

The authors develop a multi-channel extension of the two-sample instrumental-variables approach. First, regression-discontinuity models using ACS data estimate the size of each insurance transition at age 65 for individual states and time periods. The design exploits variation across states and between the pre- and post-ACA periods because the proportion of people entering Medicare from uninsurance differs substantially across these settings.

The researchers then estimate corresponding age-65 discontinuities in medical access and utilization using BRFSS and hospital data. They relate these outcome discontinuities to the proportions of people experiencing each insurance transition. This approach generates separate local average treatment effects for gaining insurance, switching from another insurance source to Medicare, and moving from another insurance source to Medicare plus secondary coverage.

The authors conduct numerous robustness and falsification tests, including alternative bandwidths, separate analyses of within-state and between-state variation, tests for nonlinear relationships, and corrections for measurement error. The identifying assumptions are stronger than those required by a standard regression-discontinuity design, but the analyses generally support the validity of the principal estimates.

Findings/Size Effects

The estimated effects of gaining insurance are large. Among previously uninsured near-elderly adults, obtaining Medicare reduces the probability of foregoing medical care because of cost by 53.9 percentage points. It increases the probability of having a regular medical provider by 32.8 percentage points and increases the probability of having received a checkup during the previous year by 41.2 percentage points.

Insurance also produces substantial increases in hospital utilization. Moving from no insurance to Medicare generates an estimated 13.2 additional hospital admissions per 100 people annually. For comparison, uninsured 64-year-olds experience approximately 4.9 hospitalizations per 100 people, meaning the estimated increase exceeds twice the pre-Medicare hospitalization rate.

Much of the increase involves elective treatment. Gaining insurance produces approximately 7.4 additional elective admissions per 100 people, compared with a baseline elective hospitalization rate of only about 0.48 per 100 among uninsured 64-year-olds. The analysis also estimates approximately 3.6 additional emergency hospitalizations per 100 people. Emergency admissions include treatment for serious conditions such as sepsis, cardiac arrhythmias, and renal failure. Elective admissions include hip and knee replacements and cardiac procedures, treatments that can substantially affect quality of life and, in some cases, longevity.

These findings indicate that lack of insurance is associated not simply with postponement of discretionary services but with substantial reductions in treatment for consequential medical conditions. By contrast, the estimated effects of changing insurance arrangements among people who were already insured are generally much smaller. Thus, much of the increase in utilization observed at age 65 appears attributable specifically to previously uninsured people gaining coverage rather than to insured individuals switching to Medicare.

The results are generally robust across alternative specifications, although the multi-channel estimates are less precise than estimates from a simpler single-channel instrumental-variables model. The estimate for receiving a recent checkup is also less robust across alternative sources of identifying variation. Nevertheless, both approaches produce broadly similar evidence that gaining insurance substantially increases access and utilization.

Conclusion

The analysis concludes that health insurance has large causal effects on medical access and utilization among uninsured near-elderly Americans. Medicare eligibility at age 65 substantially reduces cost-related barriers to care, increases connections to regular medical providers, and produces large increases in both elective and emergency hospital treatment.

The study's principal contribution is its separation of the consequences of gaining insurance from the effects of changing insurance arrangements among people who already have coverage. This distinction resolves an important limitation of earlier studies using the age-65 Medicare threshold. Because far more people change insurance type at age 65 than move from uninsurance to insurance, failing to distinguish these channels could obscure the actual effect of gaining coverage.

The large samples spanning multiple states and years also provide sufficient statistical power to identify specific types of hospital treatment that uninsured adults appear to forgo. The findings therefore extend the Medicare and health-insurance literature beyond aggregate utilization measures. For the near-elderly population studied, the evidence indicates that insurance coverage closes a substantial portion of the access and utilization gap between insured and uninsured adults and changes the receipt of medical services with potentially important consequences for health and quality of life.