A recent Buckeye Institute report claiming that Medicaid work requirements would significantly increase beneficiaries’ work hours and lifetime earnings is based on flawed assumptions, misrepresents existing research, and ignores actual experience regarding Medicaid and work from states like Arkansas, Michigan, and Ohio. [1] Thus, its estimates are invalid.
Using national survey data, the authors identify two groups of enrollees who received Medicaid through the Affordable Care Act’s (ACA) Medicaid expansion: those working at least 20 hours per week and those working less than 20 hours per week. The authors’ conclusions are entirely driven by two assumptions. First, they “assume that all enrollees would satisfy the work requirement.” In other words, no one would lose Medicaid due to the requirement, nor would anyone be exempt; instead, the average work hours for those now working under 20 hours per week (12 hours for women, 13 hours for men) would increase to match the average work hours for those now working over 20 hours per week (34 hours for women, 38 hours for men). Second, they assume that enrollees would remain on Medicaid, working the increased number of hours, until they turned 65.
Not only do the authors offer no empirical evidence to justify these assumptions, but a range of relevant data and studies refute them.
- The great majority of adult Medicaid beneficiaries who aren’t working face significant barriers to work.[2] Yet the Buckeye Institute study explicitly assumes that every single beneficiary not currently working due to a disability, caregiving responsibilities, or other barriers would start working more than 20 hours per week due to work requirements.
- In Arkansas, the only state to implement Medicaid work requirements to date, nearly 17,000 low-income adults — 22 percent of those subject to the work requirement — have lost Medicaid coverage due to non-compliance.[3] There has been no discernible increase in labor force participation in the state since the work requirements took effect in June 2018,[4] and only 0.5 percent of those subject to the work requirements have newly reported work hours, many of whom would likely have found work without the requirement.
- The Temporary Assistance for Needy Families (TANF) cash assistance program, which requires most adult beneficiaries to participate in work activities as a condition of eligibility, is often held up as a model for work requirements in other programs. However, even the best-performing city programs, where beneficiaries had access to federally financed work training and supports (conditions not typically present in the Medicaid work requirement proposals to date), produced no evidence that work requirements have a lasting positive impact on work.[5]
- Most adults enrolled in Medicaid spend less than two years on the program,[6] calling into question the Buckeye Institute study assumption that expansion adults will remain enrolled for their entire working lives.
- Health coverage supports work, so taking it away may actually impede work for some beneficiaries. For example, survey results show that among non-working adults gaining coverage through the ACA’s Medicaid expansion in Ohio and Michigan, majorities said having health care made it easier to look for work; among working adults, majorities said coverage made it easier to work or made them better at their jobs.[7]
There are other concerns with the Buckeye Institute study as well. First, the survey data reflect a period before any state had a Medicaid work requirement, so there is no reason to attribute the difference in work hours between the two groups of beneficiaries even in part to a work requirement. Second, the authors base their theory on the unsupported premise that beneficiaries knowingly limit their work to remain income-eligible for Medicaid, but they then assume, inconsistently, that beneficiaries would work well over the required 20 hours per week under a work requirement. And in fact, working at minimum wage at the elevated hours the study assumes under a work requirement would make low-income adults ineligible for Medicaid expansion.
As backing for their approach, the authors cite studies purporting to show that Medicaid depresses work participation. But this assertion is also inconsistent with the best available evidence.
A landmark study of Oregon’s expansion of Medicaid to low-income adults through a lottery finds no evidence that Medicaid enrollment significantly affects labor force participation.[8] Likewise, another evaluation finds no evidence that employment rates rose among adults after they lost Medicaid due to Tennessee’s termination of its Medicaid expansion program in the 1990s.[9] (This evaluation is more rigorous and more recent than a similar evaluation cited by the Buckeye Institute.) Finally, a Kaiser Family Foundation meta-analysis finds no credible evidence that Medicaid expansion reduced employment.[10]
The Buckeye Institute report gives disproportionate attention to one study showing that participation in Wisconsin’s BadgerCare program reduced labor force participation.[11] But that study reflected some unique factors, which may explain why it’s inconsistent with the preponderance of the evidence. Its results are taken from the Great Recession, when barriers to employment were significantly higher. Also, unlike in current Medicaid expansion programs, those disenrolled from BadgerCare due to increased earnings could not re-enroll later if their earnings fell.
The Buckeye Institute report does include the following caveat:
The lifetime profile of earnings by individuals on Medicaid assumes Medicaid enrollment for the entire life of the individual and that, in the absence of work requirements, individuals who work less than 20 hours per week will continue to do so for the remainder of their lives. In reality, people improve their earnings enough to leave Medicaid, increase their hours worked on their own, or just enroll in Medicaid while between jobs only to dis-enroll when they find work.
This characterization of the critical role that Medicaid plays for working adults in the absence of work requirements is well founded. But the report’s assumptions are inaccurate, its methods flawed, and thus its analytical results unusable.
