In June 2026, the Centers for Medicare & Medicaid Services (CMS) issued an interim final rule (IFR) for implementation of the Medicaid work requirement (also called the community engagement requirement) established by last year’s One Big Beautiful Bill Act.
As part of the rulemaking process, federal agencies must prepare a regulatory impact analysis that assesses the anticipated costs and benefits of economically significant regulatory actions. The regulatory impact analysis for the new rule attempts to quantify the benefits and costs of the additional hours Medicaid program participants are expected to work as a result of the policy. It also estimates the effect of the work requirement policy on Medicaid enrollment and expenditures.
In this article, we summarize findings from our recent comment letter and explain how the impact analysis is inconsistent with evidence from prior state-level implementation of Medicaid work requirements and contains apparent errors that lead to a vast overestimation of potential employment effects. Consequently, the rule overstates the provision’s economic benefits, even as it ignores the costs associated with the stated losses of coverage.
How the interim final rule understates potential losses in Medicaid coverage
Beginning in January 2027, Medicaid coverage through the Affordable Care Act expansion will only be available to adults who can demonstrate that they are engaged in work, school, community service, or a work program for at least 80 hours in specified months or that they qualify for an exemption from the work requirement.
In the rule, CMS projects that approximately 3.1 million people will lose Medicaid coverage in the first full year of implementation of the work requirement and that the annual enrollment reduction will increase to 3.3 million by 2033 and then stabilize at this level.
This estimate is substantially lower than estimates from other sources. For instance, the Congressional Budget Office and RAND project that the work requirement will reduce Medicaid enrollment by approximately 5.7 million and 5.3 million, respectively, in 2034. Similarly, an Urban Institute analysis projected that work requirements could reduce Medicaid expansion enrollment by 5 million or more in 2028 if states do not adopt certain policies and procedures to mitigate coverage losses among people in compliance with or exempt from the provision. However, the rule limits states’ ability to minimize coverage losses. For instance, restrictions and documentation requirements placed on medical frailty exemptions could lead to substantial loss of coverage for people with significant health needs.
In estimating coverage losses, the rule assumes at least 26 percent of the 20.4 million expansion enrollees would qualify for exemptions based on medical frailty, caregiving status, and other criteria. The remaining 74 percent, about 15.1 million enrollees, would have to demonstrate compliance.
Among those 15.1 million expansion enrollees, the rule estimates compliance rates based on a KFF analysis of employment in 2023 among Medicaid enrollees ages 19 to 64. The rule’s assumptions imply that approximately 79 percent of these enrollees would be deemed compliant based on employment, 8 percent would be deemed compliant based on school attendance, and between 12 and 13 percent would be deemed noncompliant.
Based on these assumptions, the rule estimates that 9 percent of all Medicaid expansion enrollees, or about 1.8 million beneficiaries, would lose coverage because of noncompliance. The rule further implies that 1.3 million people who should be deemed compliant or exempt will be procedurally disenrolled because they have difficulty with the verification and reporting process. Thus, the total projected coverage loss is approximately 3.1 million, or about 15 percent of enrollees, over 4 in 10 of whom would lose Medicaid coverage despite being in compliance with or exempt from the new work requirements.
Notes: Applicable individuals include Medicaid expansion enrollees who are not excluded from the work requirement. Estimates are applied to the rule’s projected number of enrollees in fiscal year 2027. Population estimates are rounded to the nearest hundred thousand. Percentages are rounded to the nearest whole number. The overall share not meeting the work requirement is adjusted from 9.2 percent to 9 percent to match estimates in the rule, and the share of applicable individuals not meeting the work requirement is adjusted from 12.7 percent to 12 percent to match estimates in the rule.
The rule’s assumptions likely underestimate Medicaid enrollment declines. For instance, the rule does not account for employment instability among working enrollees, how difficult it may be for states to verify employment and income for self-employed or gig workers, or challenges verifying compliance for students given the current lack of integration between Medicaid eligibility systems and school enrollment data.
In addition, the rule assumes most people who are not working because of medical frailty will be excluded from the work requirements. But the rule’s medical frailty exemption criteria, which are more stringent than the criteria laid out in the statute, will make it difficult to identify people with health conditions that impair their ability to meet the work requirement. If consistent with prior experiences with Medicaid work requirements in states such as Arkansas, Georgia, Michigan, and New Hampshire, the expected procedural disenrollment rate for people with health-related work limitations will be much higher than assumed in the rule.
Why the estimated employment effects are inconsistent with prior evidence
The rule’s estimated costs and benefits assume that approximately 5 million Medicaid participants will increase their work effort or transition to work or other community engagement activities because of the policy. This assumption is inconsistent with the rule’s own analysis of coverage impacts and is not grounded in evidence.
First, more than 90 percent of Medicaid enrollees are working, attending school, or have characteristics that should qualify them for an exemption. As the rule itself shows, this implies that the number of people who could increase their work effort because of the policy is limited to 1.8 million or fewer individuals. Analyses of other data sources have similarly found that less than 10 percent of expansion enrollees are in a position to transition to employment because of the work requirement.
Second, multiple independent, rigorous evaluations of the impact of work requirements in Arkansas’ Medicaid expansion program found the policy had no effect on employment but did lead to a significant increase in the number of uninsured adults in the target population. The rule does not reference available evidence showing that conditioning Medicaid eligibility on participation in work or other activities has not led to meaningful changes in employment and has caused rapid losses in coverage.
Third, the 5-million-person projected increase in employment and other community engagement is far larger than the current pace of overall job growth in the economy.
Finally, the rule does not acknowledge that the One Big Beautiful Bill Act includes no new funding for job training, education, job search assistance, or other supports to enhance job skills or address employment barriers facing the Medicaid population. The interim rule also does not address structural issues in the low-wage labor market that can cause difficulty finding and maintaining stable employment.
How the rule appears to overestimate employment impacts
The rule’s analysis of employment impacts also has several apparent flaws. As noted above, the rule implies that no more than 1.8 million enrollees would be deemed noncompliant based on current patterns of employment or school enrollment. Because other enrollees are excluded or already compliant, these 1.8 million enrollees represent the maximum number of people who theoretically could transition to employment or increase their work hours or other community engagement because of the policy. Other, more detailed survey data suggest the number of expansion enrollees who are not working because of lack of interest in a job is even lower—no more than a few hundred thousand. Thus, the estimate that 5 million people would be moved to work or other community engagement because of the policy, which is provided in IFR tables 42 and 43, is implausibly high.
The source of the discrepancy appears to stem from the rule’s reliance on a separate estimate from a 2025 White House Council of Economic Advisers (CEA) report purporting to show that, according to the rule, “at least 50 percent of able-bodied adults enrolled in Medicaid did not work.” The rule appears to take the difference between the 50 percent assumed not to be working according to its interpretation of the CEA report and the 75 to 81 percent of applicable individuals assumed to be meeting work requirements through employment under the four scenarios in table 39 of the IFR to calculate that 25 to 31 percent of people in each scenario would be “moved to work” by the policy.
But the estimates that are derived in the rule do not represent the share of “able-bodied” enrollees who would be moved to work; rather they represent applicable individuals who are already working enough to be compliant with the work requirement according to the rule’s own analysis. These percentages, and a similar calculation for the share “moved to other community engagement,” are apparently applied to the 15 million individuals who must demonstrate compliance with the work requirement to determine that about 5 million would increase work effort or transition to work or other community engagement, which in turn informs the rule’s cost–benefit analysis. Therefore, the cost–benefit analysis included in the rule appears to be built on an error in estimating how many individuals could be moved to employment or other community engagement by the policy. When the apparent error is corrected, the estimated share of expansion enrollees who are not working but may be able to work would decline substantially.
Because of these issues, the analysis of employment impacts appears to assume a much larger pool of individuals who could move from nonwork to employment because of the policy compared with its own coverage impact analysis. To estimate coverage impacts, the rule seems to assume most people who can work are already working based on the KFF report, and therefore coverage losses will be low. But for the purpose of estimating employment impacts, it seems to assume only half of “able-bodied” adults are working based on the CEA report (which, as we show in our full analysis, is incorrectly measured in the CEA report and misinterpreted in the rule).
Given the evidence from prior literature on Medicaid work requirements and the small size of the group that could be incentivized to work, employment gains resulting from implementing the work requirement are likely to be negligible. Therefore, the interim rule has significantly overstated the policy’s potential benefits.
More broadly, the rule does not measure any of the costs associated with coverage loss in the form of worse health outcomes, higher mortality, lower rates of productivity and employment, higher uncompensated care costs, and administrative burdens on providers who assist patients with documentation or who face greater reimbursement challenges as their patients experience coverage instability.
Read the full comment letter here.
This analysis was funded by the Robert Wood Johnson Foundation. We are grateful to them and to all our funders, who make it possible for Urban to advance its mission. The views expressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. Funders do not determine research findings or the insights and recommendations of Urban experts. Further information on the Urban Institute’s funding principles is available at urban.org/fundingprinciples.
