Causal Impact of Medicaid on Poverty: The Importance of Poverty Measures

What is the impact of Medicaid on poverty? Prior estimates have different shortcomings. Sommers and Oellerich (2013) estimated a causal effect of Medicaid on poverty, as measured by the supplemental poverty measure (SPM). However, the SPM captures poverty reduction due to reduced out-of-pocket spending but not due to meeting health care needs.

Korenman and Remler (2016) developed a health inclusive poverty measure, which includes both an explicit need for health insurance and health insurance benefits as resources to meet that need. Remler, Korenman and Hyson (2017) estimated the impact of Medicaid and other health insurance benefits on health inclusive poverty. However, their estimates were non-causal “accounting” estimates, assuming that in the absence of Medicaid, households would have no other health insurance benefits to meet coverage needs. To the extent that low-income individuals can gain access alternative insurance benefits, these non-causal estimates likely overstate the impact of Medicaid on poverty.

This paper combines the strengths of these prior two approaches to estimate the causal impact of Medicaid on a measure of poverty that incorporates health insurance needs and benefits. We follow Sommers and Oellerich and use propensity score matching to simulate counterfactual health insurance coverage and out-of-pocket spending on medical care and insurance from the distribution of otherwise similar individuals without Medicaid coverage, using the Current Population Survey. We then measure poverty in the counterfactual no-Medicaid state using the Health Inclusive Poverty Measure (HIPM).

We find Medicaid reduced HIPM poverty by 2.5 percentage points among the under-age-65 U.S. population in 2015, more than double the estimated 1.0 percentage point reduction in poverty using the SPM and lower than the 3.9 point non-causal accounting estimated impact using the HIPM.

Among those who, absent Medicaid, would become uninsured, Medicaid reduced HIPM and SPM poverty by 6.4 and 1.2 percentage points, respectively. Among those who would transition to private non-group insurance, Medicaid reduced HIPM and SPM poverty by 4.1 and 1.8 percentage points, respectively. In contrast, among those transitioning to employer-sponsored insurance, Medicaid reduced HIPM and SPM poverty by 1.4 and 0. 8 points.

We find that the difference between Medicaid’s impact among African Americans and Hispanics and its impact among Non-Hispanic Whites is much larger when measured by the HIPM than the SPM. Specifically, we estimate that Medicaid reduces SPM poverty among Blacks, Hispanics and Whites by 1.7, 1.7 and 0.6 percentage points, respectively, but reduces HIPM poverty by 4.1, 4.8 and 1.4 points. The measure of poverty also affects how Medicaid’s impact varies by family structure. In particular, the HIPM shows much greater Medicaid impact on single parents, relative to other families, than the SPM.

We also investigate the realism of the counterfactual distribution of insurance status and out-of-pocket medical expenditures, and explore sensitivity to alternative counterfactuals. These results suggest major impacts of Medicaid on poverty, especially for persons likely to lack alternative health insurance benefits.