Medicaid Work Requirements Will Take Away Coverage From Millions: State and Congressional District Estimates

Harsh Medicaid work requirements in the Republican reconciliation legislation will put 9.9 million to 14.9 million people at risk of losing Medicaid coverage in 2034. (See Table 1 for state estimates and link in footnote below for congressional district estimates.^[1]) The final law applied Medicaid work requirements to parents of children over age 13,^[2] more expansive than the Senate Finance Committee version that included parents of children over age 14, and significantly more expansive than the original House version that did not include any parents. Roughly 200,000 to 500,000 more people will be at risk under the legislation than under the original House bill.^[3]

If states experience coverage losses at rates observed when Arkansas implemented work requirements in 2018-2019, we estimate that 7.1 million people among those at risk will lose coverage. But coverage losses could be higher because of the work requirements provision’s draconian nature. If a state implements the policy poorly or chooses to implement it stringently, nearly all of those at risk could lose coverage.

The new law will take Medicaid coverage away from people, largely those enrolled in the Medicaid expansion group, who can’t document that they meet rigid, red-tape-laden work requirements.^[4] The legislation will also prevent people subject to the requirements from enrolling in Medicaid unless they are already employed, which is more extreme than recent bills such as the Limit, Save, Grow Act of 2023, and experiments that states such as Arkansas tried during the first Trump Administration.^[5]

The widening of the requirements to include some expansion group parents, relative to the original House proposal, will have highly differential impacts across states. In particular, there are larger impacts for states with lower income thresholds for non-expansion parents, because in these states more parents are in the expansion group.^[6]

We define “at risk” as including enrollees who will be subject to work requirements and who will likely not be automatically exempted by states from the reporting requirements. Our lower estimate of 9.9 million people at risk of losing coverage is based on a scenario in which states more effectively use data matching to reduce the number of people who need to take action (e.g., by submitting proof of their work hours) to comply. Our higher estimate of 14.9 million is based on a scenario in which states are less effective with data matching. (States with older eligibility systems, or those with less integration across health and human services programs, may be less effective at data matching.) Under both scenarios, we assume that states will automatically exempt those parents who are not subject to the requirements.

Effective January 1, 2027, the new law requires states to deny coverage to people applying for Medicaid if they are not already working (or participating in another qualifying activity) at least 80 hours per month, as well as terminate Medicaid for people already enrolled if they cannot document that they are meeting work requirements.^[7] The provision applies to Medicaid expansion group enrollees aged 19 to 64 — the entire age range of the expansion group. The provision says it exempts caregivers of dependent children aged 13 and below, people with disabilities, people who are pregnant, and several other groups.

However, evidence shows that much of the coverage loss due to work requirements will occur among people who work or should qualify for an exemption but nevertheless will lose coverage due to red tape (states should be able to exempt most people with children aged 13 and below automatically, but many others who should be exempt, such as some people with disabilities, will not be automatically exempted).^[8]

If coverage loss is on par with Arkansas’ work requirement experience in 2018-2019, we estimate that 39 percent of expansion enrollees will lose coverage, even though only 13 percent of expansion enrollees either did not work in the past year or did not potentially qualify for an exemption. (See Table 1 for state-level estimates.) In other words, at least 2 in 3 enrollees losing coverage will either be workers or will likely qualify for an exemption based on having a disability, going to school, or other factors.

Meanwhile, research shows — and the CBO previously concluded — that work requirements do not increase employment.^[9] Instead, they lead enrollees who lose coverage to take on more medical debt, delay getting needed medical care, and delay taking medications.^[10]

The law is more stringent than Arkansas’ failed work requirements policy during the first Trump Administration. It applies to expansion adults aged 19-64 while Arkansas’ policy applied to adults aged 19-55. The law also requires that people demonstrate compliance with the requirement for at least one month, and up to three months (at the discretion of the state), prior to enrollment.^[11]

On the other hand, while Arkansas’ policy required monthly reporting, the reconciliation legislation lets states choose whether to require reporting every month or a longer period, up to every six months. Unlike Arkansas, the law allows states to request waivers to exempt people in counties with very high unemployment rates and in disaster-declared areas. (The law also includes a provision that will make it possible for some seasonal workers to comply with the requirement even if they do not have sufficient hours on a monthly basis.^[12]) In practice, only a very small fraction of counties have unemployment rates that consistently meet the criteria, so states would have to elect to apply for a waiver, and the waiver could need to be approved on a month-to-month basis.^[13]

CBO estimated that the work requirements provision in the House-passed version of the bill would have led to 5.2 million people losing Medicaid coverage, with 4.8 million people becoming uninsured.^[14] CBO’s estimates of coverage loss under Medicaid work requirement proposals have been lower than estimates from CBPP, the Urban Institute, and the Brookings Institution.^[15] For example, the Urban Institute estimated that a policy similar to the Limit, Save, Grow Act of 2023 would lead to coverage loss of between 5.5 million and 6.3 million people when applied to the full, aged 19-64 expansion group, and the authors noted that the House reconciliation bill’s proposal would lead to even higher coverage losses.^[16]

Our coverage loss estimates have been more consistent with the Brookings and Urban findings than with CBO’s estimates. Using American Community Survey and Medicaid administrative data, we follow a similar approach to the Urban Institute analysis. We then allocate our state-level estimates to congressional districts for the 119^th Congress using Census custom tabulations of Medicaid enrollment by age and parental status from the 2023 American Community Survey. Our estimates do not include interactions with other provisions in the reconciliation legislation. A detailed description of the methodology follows the table below.

Methodology

We estimate expansion enrollment in fiscal year 2034 using MBES data collected by the Centers for Medicare & Medicaid Services and CBO’s June 2024 Medicaid baseline.^[17] For states that adopted Medicaid expansion before 2019, enrollment is projected from fiscal year 2019 to account for differences in pandemic-era enrollment trends; otherwise, we project enrollment from fiscal year 2023. For Wisconsin, which covers adults without children who have incomes up to 100 percent of poverty through their 1115 waiver, we project adult enrollment through non-disability pathways from fiscal year 2019 using Medicaid and CHIP Payment and Access Commission analysis of T-MSIS data.^[18] Estimates for North Carolina and South Dakota, which expanded Medicaid in 2023, were calculated using state administrative enrollment data.^[19]

To estimate characteristics of expansion enrollees, we identify adults potentially enrolling through the expansion pathway in the Census Bureau’s 2023 American Community Survey (ACS).^[20] Following our coverage gap methodology,^[21] we estimate eligibility based on Medicaid rules for defining income, income eligibility thresholds,^[22] and lawful immigration status. We exclude adults who report receipt of Supplemental Security Income or Medicare, as they are likely enrolled through the disability pathway and not the Medicaid expansion.

Explanation of Exemptions and Coverage Loss Assumptions

To estimate the proportion of enrollees who could potentially be automatically exempted from work requirements, we calculate the proportion of expansion adults who are parents with children age 13 and under, have wages consistent with working 80 hours per month under the federal minimum wage, and are potentially complying with Supplemental Nutrition Assistance Program (SNAP) work requirements (i.e., people who are subject to the federal time limit on SNAP benefits unless they meet a work requirement). Following the methodology from the Urban Institute, we determine potential SNAP work requirement compliance based on whether an individual works an average of 80 hours per month during the year, is between ages 18 and 54, lives in a household receiving SNAP, and does not have children.^[23]

We present estimates of the population at risk of losing coverage from two scenarios based on the degree to which states can successfully match data to automatically exempt enrollees who are subject to requirements from reporting. For both the limited and greater data matching scenarios, since most parents have their children included in their Medicaid household, we assume states can identify parents with a child aged 13 or below based on initial Medicaid applications and automatically exempt enrollees based on parental status.

Evidence suggests that states’ capacity to match their Medicaid databases with their wage databases and SNAP databases vary widely. For the greater data matching scenario, which aligns with the Urban Institute’s approach, we assume states automatically exempt enrollees based on wage data and receipt of SNAP benefits. We assume under this scenario that states can obtain earnings information for non-self-employed workers from state and commercial wage databases, and that states can identify individuals complying with SNAP work requirements in SNAP databases. Under the limited data matching scenario, we assume states are not able to use wage and SNAP data to automatically exempt enrollees.

We then apply state-specific estimates of these proportions to estimated enrollment in fiscal year 2034 to estimate the population who could be automatically exempted from a work requirement.^[24] We assume coverage loss to be 72 percent among the population who would be subject to and not automatically exempted from work requirements under the greater data matching scenario, based on Arkansas’ experience with work requirements.^[25]

To estimate the proportion of expansion enrollees who worked in the past year or might qualify for exemptions, we calculate the proportion of expansion adults who are parents of a child aged 13 or below, worked at least one month in the past year, are part-time or full-time students, are looking for work, have a disability, or live with a household member with a disability. These characteristics are similar, but not the same, as those estimated by the Urban Institute,^[26] and they do not capture all the exemptions and exceptions permitted by the legislation (e.g., medical frailty, living in a county with a high unemployment rate, experiencing short-term hardship).

Congressional District Allocation

We allocate state-level estimates to congressional districts for the 119^th Congress using custom tabulations acquired from the Census Bureau of adults ages 19 to 64 without children who report enrolling in Medicaid and do not report enrolling in Medicare or receiving Supplemental Security Income.^[27] The Census custom tabulations are based on combined 2019-2023 ACS data to increase sample sizes, so we scale total adult Medicaid enrollment in the custom tabulations up to our estimates of adult Medicaid enrollment by 119^th congressional district, which are based on the 2023 ACS only.^[28] This allocation assumes that, in each state, the estimated populations subject to and losing coverage due to work requirements are geographically distributed the same way as the aforementioned ACS congressional district estimates.

We also note whether a congressional district overlaps at least one county with an average unemployment rate greater than or equal to 1.5 times the national unemployment rate for 12 consecutive months from May 2024 to April 2025, the most recent period of data available from the Bureau of Labor Statistics.^[29] People living in these counties could be exempted from the work requirement if their state decides to seek an exemption.^[30]