Former RSF visiting scholar Helen Levy (University of Michigan) and RSF grantee Thomas Buchmueller (University of Michigan) have recently published an NBER working paper, "Medicaid Expansion and the Unemployed," based on a 2018 grant from RSF and the Robert Wood Johnson Foundation. Levy, an expert on public health insurance and health care reform, answers some questions for us on findings and implications from their recent report.
How would you describe your empirical approach to understanding the connection between unemployment and Medicaid expansion after the passage of the Affordable Care Act?
We compared what was happening in states that did and didn’t expand Medicaid under the ACA. The first step was to establish that before the ACA, trends in health insurance, health care access, and unemployment exits were similar across the two groups of states. This doesn’t mean they were identical – they weren’t; for example, non-expansions states had consistently higher rates of uninsurance than expansion states – but they moved together over time in the same way.
Having established that, we then rely on the post-ACA trends in the non-expansion states to tell us what would have happened in the expansion states after the ACA if they hadn’t expanded Medicaid. The comparison between this "counterfactual" outcome and the outcome that we actually observed in the expansion states is what gave us an estimate of the effect of Medicaid on different outcomes.
It’s particularly important in this case to have a counterfactual because there were other changes happening at the same time as Medicaid expansion that affected both expansion and non-expansion states. The implementation of other ACA reforms that made private insurance more accessible and the ongoing recovery from the recession are the two most important of these. Because of these, you can’t just point at the decline over time in uninsurance among the unemployed in expansion states and say "Look what a difference Medicaid expansion made!" But analyzing how the difference between expansion and non-expansion states changed over time does allow us to make that statement.
This approach is called "difference-in-differences" analysis.
Did the expansion of Medicaid eligibility improve access to health care for the unemployed?
It did. Medicaid expansion increased the fraction of the unemployed who had insurance coverage by eight percentage points more than what would have happened without Medicaid expansion. About 40 percent of unemployed individuals in expansion states were uninsured before the ACA, so Medicaid expansion reduced uninsurance among the unemployed by about one-fifth. Medicaid also reduced the fraction of the unemployed who reported that they had needed to see a doctor at some point in the last year but had not done so because of the cost.
How does the availability of government-subsidized health insurance impact the duration of unemployment spells and the likelihood of people detaching from the labor market altogether?
It didn’t! This is surprising because many people have argued that providing free health insurance must reduce the incentive to work. But we find no evidence that Medicaid expansion meant that unemployed people were less likely to find jobs. And for some subgroups of the unemployed – specifically, those with children and those who had been unemployed less than a year – Medicaid expansion actually made them less likely to exit the labor force (that is, to give up on finding a job). In other words, the increases in coverage and access to care that resulted from Medicaid expansion did nothing to discourage unemployed workers from working.
Can you expand on your research findings that the expansion of Medicaid did not prompt moral hazards for the unemployed – i.e., the availability of Medicaid did not make it more likely that people would be less likely to take job offers or keep on searching? How might your findings inform efforts by some states to tie Medicaid eligibility to workforce requirements?
One reason states are considering such requirements is out of concern that Medicaid reduces the incentive for beneficiaries to work. This study shows that for one important subgroup – the unemployed – there is no evidence to support the idea that Medicaid discourages work. So our finding undercuts that rationale for imposing work requirements.
What other current research projects and/or projects do you have in the pipeline?
I have a couple of papers on the ACA that should be coming out fairly soon. One, coauthored with Nicholas Bagley and Andrew Ying, is called "What’s Left of the ACA?"; we offer a progress report on the ACA after ten years, discussing what has been rolled out as intended, what hasn’t, and why – including but definitely not limited to partisan attacks on the law’s high-profile provisions. That is slated for an upcoming issue of RSF: The Russell Sage Foundation Journal of the Social Sciences. Tom Buchmueller, who was one of my collaborators on the unemployed project, and I have a paper in the pipeline at Health Affairs looking at what has happened to racial and ethnic disparities in coverage and access since the ACA. (Spoiler alert: they’ve declined.) At a less-polished stage are a couple of projects analyzing how social insurance programs (Medicare, Social Security) have affected food insecurity among older Americans.