Over 3 in 4 adults potentially subject to Medicaid work requirements may already have a route to exclusion or compliance Authors: Joshua Varcie, Shamik Bhat, Teddy Knox, Heather Howard Published: 2026 ABSTRACT AbstractIntroductionStarting in 2027, 18.5 million Medicaid enrollees will likely be subject to new work requirements that could lead to significant coverage losses.MethodsUsing American Community Survey data, we estimate the share of the population of interest that will have a route to exclusion from or compliance with the requirements. We also explore how rates of exclusion from or compliance with work requirements vary by state and demographic group.ResultsWe find that 46% of those in our proxy population will meet the criteria for at least one exclusion, 58% will meet the criteria for compliance, and 78% will have at least one route to exclusion or compliance. The “exclusion or compliance” share varies from 70% in New York to 94% in North Dakota. It also varies by demographic group, as women and younger adults are more likely than their counterparts to have a route to exclusion or compliance.ConclusionThe large “exclusion or compliance” share implies that an important driver of coverage loss will be states’ capacity to verify exclusions or compliance, as <1 in 4 enrollees are estimated to not have a route to exclusion or compliance.Keywords: state Medicaid programs, work requirements, eligibility verifications, administrative burden, American Community Survey Introduction H.R. 1 introduces the first federal work requirement for certain Medicaid beneficiaries. Beginning January 1, 2027, enrollees in the new adult group created by the Affordable Care Act (ACA) must demonstrate compliance with work requirements or qualify for an exclusion to maintain coverage. The impact will be substantial: the Congressional Budget Office projects these requirements will cause nearly 5 million of the 18.5 million subject to the requirements to lose coverage.1,2 Impacts will vary considerably across states, depending on the composition of each state's Medicaid population and on policy and administrative choices made during implementation. The stakes for coverage, health outcomes, and healthcare systems, however, are high in every state. State Medicaid agencies face tight deadlines, high system costs of verification, and staff capacity constraints as they implement work requirements.3 Furthermore, the Interim Final Rule (IFR) issued on June 1, 2026 introduces new requirements that will make it more difficult for states to implement than anticipated.4,5 To support state implementation efforts and allow for comparisons across states, this article estimates the share of the population potentially subject to work requirements that will be compliant with or excluded from the requirements, providing results for 40 Medicaid expansion states, D.C., and Wisconsin. (We provide estimates for Wisconsin because we expect the law to apply to its partial expansion population). It also produces estimates of the share potentially qualifying for each route to exclusion or compliance and analyzes this share by demographic group. We estimate that 78% of Medicaid enrollees potentially subject to work requirements may already have a route to exclusion or compliance. Based on the CBO's estimates, this would equate to roughly 14 million Americans for whom state governments may need to verify exclusion or compliance. This undertaking will involve designing and operating new systems for automatic verification and—where automated methods fail—conducting manual case management and targeted outreach to vulnerable or hard-to-reach subgroups. Our results provide crucial baseline estimates upon which states can develop a strategy to minimize coverage losses for eligible enrollees. Background on H.R. 1 work requirements H.R. 1 introduces community engagement (“work”) requirements on enrollees who receive coverage through the new adult group created by the ACA. For the first time in the program's history, Medicaid enrollment for this group will be conditional on an enrollee's ability to demonstrate community engagement through work, school, or other activities. According to the IFR, the requirement's purpose is to incentivize long-term employment, increase program integrity and reduce fraud, and align Medicaid requirements with other social services (eg, SNAP).4 Work requirements apply to a subset of those enrolled in Medicaid, known as the “new adult group”: individuals ages 19-64 who qualify for Medicaid by having income below 138% of the Federal Poverty Level (FPL). This group does not include individuals eligible through other mandatory groups, such as parents with very low incomes (varying by state), pregnant women, and individuals with disabilities. It also excludes residents of non-expansion states and anyone enrolled in Medicare Part A or B. To keep Medicaid coverage, enrollees for whom work requirements are applicable (“applicable enrollees”) must be determined compliant with or excluded from work requirements. States must offer several mandatory exclusions (see Data and Methods); there are also optional short-term hardship exclusions that states can choose to offer. (We focus on the mandatory exclusions in this article but present an estimate for the high county unemployment exclusion in the Appendix). States must first assess whether applicable enrollees qualify for an exclusion before assessing compliance. Those not determined excluded must comply with work requirements by working 80 hours per month, earning at least $580 per month, being enrolled in an educational institution at least half-time, volunteering at least 80 hours per month, or participating in certain other activities. Enrollees can also comply if they participate in a combination of these activities for at least 80 hours per month.4 To verify enrollees’ compliance or exclusion, states must use existing data sources or solicit documentation from enrollees when data is unavailable. A key concern is that excluded or compliant enrollees will be erroneously determined non-compliant because of burdensome, complex, or inefficient administrative processes implemented by state Medicaid agencies.6 Automatic verification may be costly for states because of insufficient IT systems and limited capacity to implement policy changes.7,8 Manual verification—which requires enrollees to submit documentation—will create administrative burdens for residents, some of whom will not take the needed actions for verification due to information barriers or logistical challenges.9,10 Each of these scenarios would lead to incorrect determinations resulting in coverage loss. States must develop effective implementation strategies to target eligible enrollees at risk of losing coverage due to administrative hurdles. Understanding how many Medicaid enrollees are likely to be compliant with or excluded from the requirements, and how those patterns disaggregate among different subgroups of interest, can help states prioritize investments in policy implementation efforts and create data-driven strategies for targeted outreach. Data and methods Data Our analysis uses 2024 American Community Survey (IPUMS) data to estimate the share of enrollees compliant with or excluded from work requirements.11 The ACS provides a large number of detailed, individual-level records, allowing for precise, statewide, population-weighted estimates. We supplement our analysis with other data sources (see Appendix). This article used AI for coding, graphics, and citation assistance. All output was reviewed by the authors for correctness. Defining the population of interest We focus our analysis on the population for whom work requirements are applicable. The denominator for our estimates is the “population of interest,” which includes any non-institutionalized adult age 19-64 who lives in a health insurance unit (HIU) with income below 138% of the FPL and self-reports enrollment in Medicaid. We remove individuals who self-report SSI income, are enrolled in Medicare, or are parents eligible for the mandatory parent and caretaker group. We also remove individuals who live in institutional group quarters, since these individuals are likely to be in jails and prisons or qualify for Medicaid through the disability group. We use a HIU-level variable from SHADAC to calculate income relative to the FPL.12 This variable is used to include individuals with income below 138% of the FPL and to exclude individuals with income below the state threshold for the mandatory parent and caretaker group from the population of interest.13 Our population of interest likely includes some enrollees for whom work requirements are inapplicable, making our definition overly broad and not a direct reflection of the statute. For example, some people enrolled in Medicaid through the disability group will be in our population of interest. Note also that we do not adjust our population of interest to account for the Medicaid undercount or citizenship-related eligibility requirements. Determining exclusions and compliance in survey data We use the ACS to calculate whether an individual would likely qualify for each route to exclusion or compliance. These measures are imperfect proxies; we discuss how our proxies differ from the statute and corresponding limitations throughout this section. In the Online Appendix, we offer multiple estimates where feasible to reflect the uncertainty of our primary estimates. Throughout this section, “self-report” means that an individual responded affirmatively to the corresponding question when answering the ACS. Exclusions H.R. 1 creates several groups of specified excluded individuals. Of these, we are not able to provide estimates for 4 categories: former foster care youth under age 26, participants in a drug or alcohol rehabilitation or treatment program, individuals complying with TANF work requirements, and inmates of public institutions. The remaining categories are constructed using ACS data unless otherwise noted. American Indians and Alaska natives We consider individuals to meet this exclusion if they self-report American Indian or Alaska Native descent. We expect this definition to closely align with the statute. Veterans with a total disability rating We consider individuals to meet this exclusion if they self-report a disability rating greater than or equal to 70%. Because the statute requires a disability rating of 100%, we expect our measure to overestimate this exclusion. We cannot provide a more precise estimate due to data limitations. Pregnant and postpartum women We consider individuals to meet this exclusion if they self-report having given birth to a child in the last 12 months. This measure may not include those who are currently pregnant, resulting in an underestimate. But many pregnant and postpartum women qualify for Medicaid through the pregnancy group; for these women, work requirements are inapplicable and our measure will thus overestimate the exclusion (see Appendix). Individuals enrolled in SNAP or TANF and not excluded from those programs’ work requirements To estimate the population enrolled in SNAP and not excluded from SNAP work requirements, we include all individuals reporting household SNAP enrollment in the past 12 months. Then, we remove individuals eligible for an exclusion from SNAP work requirements. This includes individuals who usually work at least 30 hours per week (or have wages greater than the federal minimum wage multiplied by 30), are enrolled in an educational institution, are pregnant, are an American Indian/Alaska Native, have a disability affecting ability to work, or are a primary caretaker of a child under 14.14 These exclusions are generally constructed in the same way as the Medicaid exclusions (see Appendix). We further subset the population to individuals working at least 20 hours per week (or earning wages at least equal to the federal minimum wage multiplied by 20 hours). As with Medicaid work requirements, we cannot determine whether an individual meets all the SNAP work requirement exclusions due to data limitations, causing us to include some enrollees who are excluded from SNAP work requirements in the estimate. At the same time, restricting the data to those who usually work 20 hours per week or more may inadvertently exclude some enrollees eligible for the exclusion. We cannot estimate individuals who would meet this exclusion through TANF enrollment due to data limitations, though we expect this population to be small. Parents of a child 13 or under We consider individuals to meet this exclusion if their youngest child is 13 or younger. Note that work requirements will not apply to many parents who qualify under the mandatory parent and caretaker group. Since we exclude this group from the population of interest, we expect our estimate of this exclusion to be a good approximation of the statute, although it may not precisely match the statute's definition of parent. Individuals meeting a medical frailty condition (proxy) Individuals can qualify for a medical frailty exclusion if they (1) are blind or disabled; (2) have a substance use disorder; (3) have a disabling mental disorder; (4) have a physical, intellectual, or developmental disability that significantly impairs their ability to perform one or more activities of daily living (ADL); or (5) have a serious or complex medical condition. The IFR further requires the individual's condition to significantly impair their capacity for community engagement. Since the ACS does not ask respondents about mental or physical health conditions, we rely on the survey's disability definition to create a proxy measure for this exclusion. First, we define as disabled all individuals who self-report severe limitations to vision, hearing, participation in basic physical activities (eg, walking or climbing stairs), satisfying personal needs (eg, bathing or dressing), or doing errands alone. Prior literature has found ACS disability measures are unbiased and effective in identifying disability burden.15,16 However, disability and medical frailty are not identical, and some people who are disabled will not be considered medically frail. Still, we believe there is sufficient overlap between medical frailty and the ACS disability definition to consider it a reasonable proxy. Because the IFR requires medical frailty to impair a person's ability to work, we perform an imputation using CPS data to proxy whether an individual's disability is severe enough to limit their ability to work by using a CPS question asking if a respondent's disability limits or prevents their ability to work.17 We calculate that 56% of those with a self-reported disability (as defined in the ACS) are limited in their ability to work because of their disability. We then perform an imputation to assign 56% of those reporting a disability in the ACS to have a disability that significantly impacts their ability to work (see Appendix). We expect some people in our medical frailty proxy to be enrolled through the Medicaid disability group, rather than the new adult group. However, a recent review suggests that 70% of disabled individuals may receive Medicaid outside the disability group.18 Potential caretakers of a disabled individual We consider individuals to meet this exclusion if they are in a household where at least one member besides themselves self-reports a condition that imposes difficulties with ADLs. Despite ambiguity around the IFR's caretaker definition, disability is more inclusive for the caregiving exclusion than for medical frailty. Our definition likely overestimates the population eligible for the caretaker exclusion because someone living with an individual with a mild disability may not “provide care within a broad range of assistance”.4 But it may also underestimate the eligible population because it does not include caretakers not residing with the individual they take care of. Compliance H.R. 1 creates 4 routes to community engagement compliance. Of these, we cannot provide estimates for 2 routes: community service and job training participation for at least 80 hours per month. We also cannot estimate the population compliant through a combination of routes. The 2 primary compliance routes, income or hours worked and educational programs are constructed as follows: Working at least 80 hours per month or earning the equivalent of 80 monthly hours of work at the federal minimum wage We consider an individual compliant through this route if they self-report income >$580 per month or at least 80 hours of work per month. We estimate income by taking total (earned and unearned) income and deducting Supplemental Security Income and welfare income to approximate MAGI income, as defined in the IFR. For the work hours route, we multiply usual hours worked per week by the usual number of weeks worked per year. Since the statute requires monthly income verification, but we use an annual measure, our method may introduce inconsistencies. We consider these routes together because they both capture compliance with the requirements through work. Enrollment in an educational program We consider an individual who reports at least half-time enrollment in an educational program compliant with work requirements through this route. We estimate this population by including individuals who self-report enrollment in school. This measure is likely an overestimate, as it includes individuals enrolled less than half-time; we expect the population enrolled less than half-time—and therefore incorrectly included in our estimates—to be small. Outcomes of interest Our primary outcomes are: the share of the population of interest that observably (1) meets the criteria for an exclusion, (2) has a route to compliance, or (3) meets the criteria for at least one route to exclusion or compliance. Importantly, these outcomes reflect meeting the criteria for exclusion or compliance and not being excluded or compliant; people who meet the criteria may not have their eligibility recognized until the state successfully verifies their route to exclusion or compliance. When defining the “any exclusion” variable, we rely on our main estimates, not the broad or narrow estimates. We explore how using our broad and narrow estimates changes the results in our sensitivity analysis and in the appendix. We focus on mandatory exclusions in the article but provide estimates for the optional county unemployment exclusion in the appendix. Limitations Our findings are estimates based on survey data and proxies for statutory eligibility, rather than measurements of true population values and direct observation of status. Each exclusion and compliance estimate relies on proxies from survey data; the directions and approximate sizes of the biases introduced are discussed above. Furthermore, certain exclusions rely on a small sample size. Additionally, self-reported survey data may produce systematic misreporting, especially due to underreporting of Medicaid enrollment.19 Finally, our population of interest may be influenced by the ongoing unwinding of the continuous eligibility provisions passed by the Families First Coronavirus Response Act (FFCRA). Broad comparisons with related studies produce estimates in a similar range, although methodological differences limit direct comparison (see Appendix). Ultimately, despite the uncertainty, we believe state governments facing tight implementation timelines and capacity constraints will find our estimates useful. Results Over 3 in 4 enrollees potentially subject to work requirements may have a route to exclusion or compliance Our estimates suggest 78% of the population of interest has at least one route to compliance or exclusion from work requirements (Figure 1). Specifically, 46% have at least one route to exclusion and 58% have a route to compliance. There is significant overlap between the excluded and compliant groups; in mutually-exclusive groupings, 20% have a route to exclusion (but not compliance), 32% have a route to compliance (but not exclusion), and 26% qualify for both exclusion and compliance. The most prevalent exclusions include being a potential caretaker of a disabled individual (16%), being a parent or guardian (16%), and having a disability that could qualify an individual for the medical frailty exclusion (13%). Compliance via income or hours worked (47%) is much more common than through education (17%). Income compliance through the monthly income threshold (45%) is more common than through meeting the hours requirement (30%). Share of the population of interest compliant with or excluded from medicaid work requirements, 2024. Note: The figure shows the share of the population of interest who qualifies for at least one exclusion (first row), meets at least one route to compliance (second row), and qualifies for at least one route to exclusion or compliance (third row). Each dot represents a state. The far left (far right) dot represents the state with the lowest (highest) share of the population qualifying for an exclusion or compliance. The vertical bar represents the U.S. average. The categories are not mutually exclusive; we estimate that 26% of the population of interest will have a route to both exclusion and compliance. Percentages are ACS population-weighted estimates. The impact of Medicaid work requirements will be felt differently across states The share with at least one route to exclusion or compliance varies by state, from 70% in New York to 94% in North Dakota (Figure 1). Exclusion rates range from 29% in Connecticut to 72% in North Dakota, and compliance rates range from 45% in Alaska to 67% in Vermont (Figure 2). State variation in routes to exclusion or compliance with work requirements. Note: The figure shows state-by-state variation in the share of the population of interest who qualifies for at least one route to exclusion or compliance (a), qualifies for at least one route to compliance (b), and meets at least one exclusion (c). Darker colors represent states with a higher share meeting the outcome shown in each panel. Non-expansion states are shaded in gray. Percentages are ACS population-weighted estimates. States also vary widely in the prevalence of specific routes to exclusion or compliance (Figure 3). In some states, large American Indian populations (46% in Alaska and 29% in South Dakota) drive high overall exclusion rates. In other states, many people may be excluded under the medical frailty exclusion, such as in North Dakota and New Hampshire (27% each, according to our proxy). We expect compliance through income or hours worked to vary from 33% in D.C. to 60% in Nebraska. We expect education compliance to vary from 6% in Montana to 22% in California. Share of the population of interest qualifying for each route to exclusion or compliance. Note: The figure shows the share of the population of interest who qualifies for each route to exclusion or compliance. Each gray dot represents a state. The far left (far right) dot represents the state with the lowest (highest) share of the population qualifying for an exclusion or compliance. The vertical bar represents the U.S. average. Routes to exclusion or compliance are not mutually exclusive. Percentages are ACS population-weighted estimates. For a full description of our definitions of each route to exclusion or compliance, see the Online Appendix. Routes to exclusion and compliance differ across demographic subgroups Figure 4 shows subgroup analyses by gender, age, and race/ethnicity. Women are more likely to meet a route to exclusion or compliance than men (79% vs 77%), mostly due to 2 exclusions: postpartum (4% vs 0%) and having a child 13 or younger (20% vs 11%). Women are also more likely to be compliant, with a higher share satisfying the income or hours worked requirement (49% vs 46%) and the education requirement (18% vs 16%). Variation in observed routes to exclusion or compliance by demographic group, 2024. Note: The figure shows the share of the population of interest who qualifies for each route to exclusion or compliance by demographic group. Within each group, the top row represents the share of the population qualifying for at least one exclusion, the middle row represents the share of the population meeting at least one route to compliance, and the bottom row represents the share of the population with at least one route to exclusion or compliance. Race and Hispanic ethnicity are measured separately; respondents identified as Hispanic may be included in any racial category. Percentages are ACS population-weighted estimates. Adults over age 45 are less likely to have at least one route to exclusion or compliance. 82% of adults ages 19-25 are likely to have a route to exclusion or compliance compared with 75% of adults ages 45%-54% and 72% of adults ages 55-64. Educational enrollment drives younger adults’ relatively high compliance rates: 45% of adults ages 19-25 are expected to comply with work requirements through education compared with <3% of adults over age 45. The parenthood exclusion peaks at ages 26-44 (over 25% qualify) and is much lower (2%) in the 55-64 age range. We find small differences in overall exclusion and compliance rates across racial groups, although disparities by race do appear for certain routes to exclusion or compliance. Sensitivity analyses To check the robustness of our findings, we provide sensitivity analyses here and in the appendix. First we explore whether the “unwinding” of the continuous eligibility provisions, still underway in 2024, confounds our findings using 2024 data. To do this, we apply our same methods to the 2019 ACS (before unwinding began) and find minimal changes. Each of our main outcomes of interest remains virtually unchanged (Appendix Figure S1). Second, we re-estimate our outcomes of interest using alternative approaches that re-construct certain routes to compliance and exclusion or remove routes that have a high degree of uncertainty. These sensitivity analyses produce estimates of the share of the population of interest with at least one route to exclusion or compliance that range from 63% to 88%, respectively (Appendix Figure S2). Discussion Our estimates suggest over 3 in 4 Medicaid enrollees potentially subject to work requirements already have at least one route to exclusion or compliance; however, that does not mean enrollees are guaranteed or even likely to keep their coverage. Work requirements impose new burdens on enrollees (to present information proving their excluded or compliant status) and state governments (to verify eligibility). States must decide how to allocate scarce resources in a short time frame to ensure eligible enrollees maintain their coverage; the relatively large size of the group we estimate to be excluded or compliant makes clear the stakes of those efforts. The extent of coverage loss from work requirements will depend heavily on how effectively states can conduct verification. The Urban Institute's model of coverage loss simulates a “high mitigation scenario” where state governments successfully deploy “extensive automatic data-matching” to verify eligibility and a “low mitigation” scenario where they use less, and finds that in the latter scenario work requirements cause 7.0 million disenrollments, more than twice the 3.0 million disenrollments in the former scenario.20 The RAND Corporation more conservatively estimates that 1.8 million enrollees (10%) who are excluded or compliant according to statute will lose coverage due to “administrative and informational barriers.”21 RAND's estimate is based on historical rates of “procedural or administrative” disenrollments; thus it likely underestimates future procedural disenrollments, as new eligibility rules will present new challenges to state capacity. Our subgroup-level findings carry important implications for state policy implementers. The significant variation in routes to exclusion or compliance across states suggests that states will face different tradeoffs when setting up systems to verify eligibility. States can use these results to find which routes to exclusion or compliance affect the largest segments of their Medicaid populations and invest in verification methods accordingly (Appendix Table S1). States concerned with helping specific demographic groups maintain coverage can also consult the disaggregated data to understand which groups will be more affected by which exclusions and compliance routes. Conclusion Our estimates find that 78% of our population of interest will have at least one route to exclusion or compliance. These routes will vary by state, and men and older enrollees will be less likely to have a route to exclusion or compliance. Ultimately, our results imply that states must adopt targeted, state-specific implementation strategies to effectively mitigate coverage losses for eligible enrollees. While our estimates are uncertain, they provide a useful starting point for state policymakers to begin crafting these strategies. Supplementary Material Contributor Information Joshua Varcie, Princeton School of Public and International Affairs, 20 Prospect Ave., Princeton, NJ 08544, United States. Shamik Bhat, Princeton School of Public and International Affairs & Yale School of Medicine, 333 Cedar St., New Haven, CT 06510, United States. Teddy Knox, Princeton School of Public and International Affairs, 20 Prospect Ave., Princeton, NJ 08544, United States. Heather Howard, Princeton School of Public and International Affairs, 20 Prospect Ave., Princeton, NJ 08544, United States. Supplementary material Supplementary material is available at Health Affairs Scholar online. Funding The authors do not report any funding. Conflicts of interest Please see ICMJE form(s) for author conflicts of interest. These have been provided as supplementary materials. The authors do not report any conflict of interest disclosures. Disclosure: The views in this article are solely the authors and do not necessarily reflect the views of Princeton University or any other organization or employer with which the authors are affiliated. Data availability All data used in this article is publicly available. Additional calculations and summaries of the data (including state-level estimates) are available from the authors upon request. Notes Associated Data Supplementary Materials Data Availability Statement All data used in this article is publicly available. Additional calculations and summaries of the data (including state-level estimates) are available from the authors upon request.