Medicaid Expansion Under the ACA: A State-by-State Guide

Last updated 1 week ago. Our resources are updated regularly but please keep in mind that links, programs, policies, and contact information do change.

A woman in Texas had to borrow $200 from her father to see a doctor in 2012. She went in for a prescription for antidepressants after a divorce, and came out having been told she had signs of hypertension and prediabetes, problems she knew she could not afford to have.

Her five children were covered by Medicaid. She was not: her part-time job paid too much to qualify for Medicaid, but too little to qualify for the tax credits that discount a private plan.

That combination is the thing most people find hardest to believe about American health coverage, and it is not a glitch. Yes, there is a version of Medicaid built for a working-age adult with no children and no disability. It is called the expansion group. Whether a person can get it depends almost entirely on a choice their state made and can still unmake.

The eligibility line is written as a percentage of the federal poverty guideline rather than as a dollar figure. That is why the cutoff moves every year, and why it is often quoted two different ways.

One subclause, one group of people

The category lives in a single subclause of the Medicaid statute, at 42 U.S.C. 1396a(a)(10)(A)(i)(VIII). It covers people under 65 years of age, not pregnant, not on Medicare, not already eligible under an earlier category, and with income at or below 133 percent of the poverty line for their family size, and it took effect on January 1, 2014.

HealthCare.gov, the federal marketplace site, gives the Medicaid line in an expansion state as income below 138 percent of the federal poverty level, and both numbers are right. A companion clause tells the state to knock the dollar value of 5 percentage points of the poverty line off a person’s counted income before comparing it to the ceiling, so a 133 percent limit behaves in practice like a 138 percent one.

In dollars, the Department of Health and Human Services set the 2026 poverty guideline at $15,960 for a household of one and $33,000 for a household of four in the 48 contiguous states and the District of Columbia. Because that guideline is reissued for each year, the dollar cutoff in paycheck terms is not a fixed number.

Notice what the clause does not ask about. The expansion group is defined by age and income, not by whether a person has children, a diagnosis, or a job.

Which states took it, and what a no looks like

The federal Medicaid agency publishes a table of income eligibility standards with a column headed Expansion to Adults. In a state that adopted, the entry reads 133 percent. In Alabama, it reads No.

The same row shows what is left over in a state that declined. Alabama covers a parent or caretaker only up to 13 percent of the poverty line, which for a working parent is barely a category at all.

KFF, the health policy research group, keeps the running tally, with an implementation date for each adopter: January 1, 2014, for Arizona and Arkansas, September 1, 2015, for Alaska, and the label Not Adopted for the rest. Ten states have never adopted it.

Wisconsin is the odd one out. The Congressional Research Service describes it as the one non-expansion state that provides Medicaid coverage of adults without dependent children, up to 100 percent of the poverty line. So the answer to did my state refuse is not always the same as the answer to am I covered.

North Carolina shows how late a state can arrive. On December 1, 2023, Governor Roy Cooper and the state health secretary, Kody H. Kinsley, launched expansion for more than 600,000 newly eligible North Carolinians, and 300,000 people already getting limited family planning benefits were switched to full coverage automatically that day.

Getting there took ten years and a change of position by the man who had blocked it. Republicans in the state legislature had opposed expansion until the year before, and Senate Leader Phil Berger carried the 2023 deal and said his chamber would move with all deliberate speed. Berger said the urgency of approving the hospital funding program was a major factor in bringing health care industry groups along, and North Carolina expected $8 billion annually plus a $1.8 billion federal signing bonus.

Too poor for the discount

The woman’s problem was not only that Texas is strict. It was that the marketplace subsidy has a floor as well as a ceiling.

The premium tax credit statute defines who may claim it as a taxpayer whose household income for the year equals or exceeds 100 percent but does not exceed 400 percent of the poverty line for a family of that size. Below 100 percent of poverty, there is no credit at all. Read that twice, because it runs backwards from every intuition about how help is supposed to work: earning too little is what disqualifies a person.

Congress wrote that floor on the assumption that Medicaid would catch everyone underneath it. When the expansion became optional, the floor stayed exactly where it was, and the space that opened between the two is what people mean by the coverage gap.

The Congressional Research Service reports the Kaiser Family Foundation estimate that 2.2 million uninsured individuals fell into the coverage gap in 2019, with people in Texas and Florida together accounting for more than half of the national total.

One hinge matters for anyone whose income is unpredictable. A Treasury regulation treats an enrollee as eligible for the credit even when their income turns out below the poverty line, provided the Exchange estimates at the time of enrollment that it will land between 100 and 400 percent, the person enrolls in a qualified health plan, and advance credit payments are authorized and paid.

In plain terms, the estimate a person gives at sign-up can decide the year, and a good-faith estimate that turns out too high is protected rather than punished.

The 2012 ruling that turned an order into an offer

None of this would be optional but for a lawsuit filed the day the Affordable Care Act was signed. One of the individual plaintiffs owned and operated an auto repair shop in Florida. By the time the case reached the Supreme Court she had filed for bankruptcy, her business killed by the recession and the Gulf oil spill, which raised an awkward question about whether she still had standing to sue.

The Court did not strike the expansion down. Chief Justice Roberts held that the problem was the enforcement lever: the statute let the Secretary cut off all of a state’s existing Medicaid money if the state declined the new group, and that, he wrote, was not relatively mild encouragement but a gun to the head. The syllabus calls the threatened loss of over 10 percent of a state’s overall budget economic dragooning.

The remedy was narrow, and it is the reason this article exists. The Court fixed the violation by precluding the Secretary from applying §1396c to withdraw existing Medicaid funds for failure to comply with the expansion, and left the other provisions of the Act untouched. What had been a condition on every state’s Medicaid money became an offer a state could take or leave.

Four justices thought that fix was itself the error, and their objection is a serious one. The joint dissent of Scalia, Kennedy, Thomas and Alito agreed the expansion as enacted was unconstitutional and argued that the natural remedy was to invalidate it rather than convert it into something Congress had not written, on the premise that Congress legislated believing no state could realistically refuse. Roberts answered them in the opinion, saying this was not to say, as the joint dissent suggests, that we are rewriting the Medicaid Expansion.

Whichever side had the better of that exchange, its practical result has governed ever since. Congress passed a national eligibility category, and the Court handed each state a veto over it.

When the voters said yes and the state did not move

Maine voters approved a citizen initiative entitled An Act to Enhance Access to Affordable Health Care on November 7, 2017, over Governor LePage’s objection. He did not implement it. Maine’s Business and Consumer Docket entered a partial judgment in June 2018 mandating implementation of the state plan amendment provision, and the Law Court dismissed the Department’s appeal as interlocutory, meaning it had come too early to be heard.

In November 2018 Justice Michaela Murphy ordered the department to begin expanding.

Maine was not alone in going around its statehouse. Six of the seven Medicaid expansion measures that reached a statewide ballot before November 2022 were approved, in Maine, Utah, Nebraska, Idaho, Oklahoma and Missouri, while Montana’s, which paired expansion with a tobacco tax, was defeated. Oklahoma’s was the narrowest, passing 50.49 percent to 49.51 percent.

Oklahoma’s voters also wrote it into the state constitution. Article XXV-A requires the state to provide medical assistance to Low Income Adults, and forbids imposing any greater or additional burdens or restrictions on eligibility or enrollment for that group than on any other Medicaid population.

That second clause is a lock, and a narrow one. A condition the state adds on its own, aimed only at expansion adults, runs into the Oklahoma constitution; a condition Congress imposes on every state is a different matter.

Why a state would turn down the money

Governor Tate Reeves of Mississippi put the political objection in a sentence when the state House advanced an expansion bill in 2024. The bill, he wrote, was straight Obamacare Medicaid Expansion, applying to as many as 300,000 able-bodied adults who could work but may choose not to, with no real work requirement.

The Foundation for Government Accountability supplies the number behind that view, estimating that 6.8 million ObamaCare expansion enrollees are not working at all, and describing the expansion as a welfare trap for more than 12 million able-bodied adults. The research is by Victoria Eardley, an FGA research fellow, and Nic Horton, its research director.

The fiscal case runs deeper than a slogan, and it is about the match rate rather than about anyone’s work ethic. Its sharpest version comes from Brian Blase, the president of Paragon Health Institute, a former special assistant to the president for economic policy at the White House, and Drew Gonshorowski, a senior research fellow there.

They argue that federal spending per expansion enrollee in 2018 came in 56.0 percent above the projections made by the program’s own actuaries, while spending per aged enrollee ran 25.4 percent below projection and per disabled enrollee 11.3 percent below.

From that, Paragon draws the argument that lands hardest: paying a higher federal share for a working-age adult than for a child or a person with a disability means a diversion of resources away from traditional Medicaid enrollees, and its proposed fix is to equalize the federal reimbursement percentage across every eligibility group.

The force in that holds regardless of anyone’s view of the Affordable Care Act. A state gets more federal help covering a healthy 40-year-old in the expansion group than covering a child with a disability in an older category, and nobody designed that on the merits of who needs care most.

What it costs the hospital in town

Fort Scott, Kansas lost Mercy Hospital, and the substitute turned out to be thin. On a cold February evening Robert Findley fell and hit his head on a patch of ice, his wife Linda called 911, and the delays that came next exposed the frayed patchwork that sometimes stands in for rural health care.

A free-standing emergency room and a new community health center took the hospital’s place, and Linda Findley did not think they were enough. “I mean, my gosh, you need to feel like you’re safe and could be taken care of where you’re at,” she said.

In Greenwood, Mississippi, the hospital is still open and has been cutting itself back to stay that way. Greenwood Leflore Hospital, founded in 1906, has cut labor and delivery, the intensive care unit, the wellness center and its after-hours clinic. The closest hospital is UMMC Grenada, more than 30 miles away. Neither Medicaid nor Medicare fully reimburses a hospital for the care it gives, and the difference is known as the Medicaid and Medicare shortfall.

The American Hospital Association reports that higher rates of Medicaid coverage for adults have been associated with improved hospital financial performance and lower likelihood of closure, and that 74 percent of rural hospital closures happened in states where expansion was not in place or had been in place for less than a year.

That is an association rather than a controlled experiment, and the hospital association is an interested party. What it points at is the same thing Fort Scott and Greenwood show from the ground: the cost of treating people without insurance does not disappear when a state declines the coverage, it moves onto whoever is left treating them.

Ninety cents on the dollar, and where the other ten come from

For the expansion group, and only for the expansion group, Washington pays a share no other Medicaid population gets. The statute set it at 100 percent for calendar quarters in 2014, 2015, and 2016, then stepped it down to 95 percent in 2017, 94 percent in 2018, 93 percent in 2019, and 90 percent from 2020 onward.

Everyone else on Medicaid is matched by a formula tied to state per capita income, with a hard band around it: the federal medical assistance percentage shall in no case be less than 50 per centum or more than 83 per centum, with the District of Columbia fixed at 70 percent and the territories at 55.

The Centers for Medicare & Medicaid Services (CMS) wrote to state health officials on August 30, 2021, setting out temporary increases to the federal medical assistance percentage available under the American Rescue Plan Act of 2021. The letter explains the late-adopter incentive: a state that newly adopts the entire adult group under its state plan or a demonstration waiver gets an 8-quarter, 5 percentage point increase to its ordinary match, calculated on its regular population rather than on the 90 percent expansion rate.

That bump is the signing bonus Berger was counting on. It is designed to make the first two years of a late adoption pay for themselves.

The remaining ten percent is the state’s problem, and Medicaid is already the largest single thing in most state budgets. The 2024 State Expenditure Report put Medicaid at 29.8 percent of total state expenditures and 18.7 percent of general fund expenditures in fiscal 2024.

States raise much of their own share by taxing the health care providers Medicaid then pays, and federal law polices that arrangement closely. A provider tax fails the test if there is a hold harmless provision, which includes the case where the Medicaid payment to the taxpayer varies based only upon the amount of the total tax paid, or where the state hands money back in proportion to what was collected.

The 2025 reconciliation law tightened the screw. Section 71115 of Public Law 119-21 replaces the flat 6 percent safe harbor threshold, for fiscal years beginning on or after October 1, 2026, with an applicable percent determined under subparagraph (D), stepping the allowance down for expansion states while freezing a non-expansion state at whatever percentage it already imposes, or at 0 percent if it imposes none.

Read from a state treasurer’s chair, that is a squeeze aimed specifically at the states that expanded. Whether it reaches residents’ taxes depends on what each legislature does when the provider tax stops covering the gap, and that is a decision still to come.

Work requirements are federal law now, not a waiver

Before 2025 a state could only try this by asking the federal government for a waiver, and the asking was where it usually came apart. Section 71119 of the same law removed the question. Beginning no later than the first day of the first quarter that begins after December 31, 2026, or earlier if a state chooses under its state plan or a demonstration, every state must make community engagement a condition of eligibility for the expansion group.

Compliance means eighty hours in a month, and the statute lists the routes. A person qualifies if they work not less than 80 hours, complete 80 hours of community service, take part in a work program for 80 hours, enroll in an educational program at least half-time, or combine those activities to reach 80 hours.

There is also an earnings route: monthly income at or above the federal minimum wage multiplied by 80 hours, or for a seasonal worker an average at that level over the preceding six months.

So a student carrying a half-time course load is inside the rule as written, and so is someone whose pay clears the threshold in fewer hours than eighty.

The requirement reaches applicable individuals only. Outside it entirely are specified excluded individuals, whom the state must identify in accordance with standards specified by the Secretary, a list beginning with former foster care youth and with an Indian or an Urban Indian and California Indians as defined in the Indian Health Care Improvement Act.

Caregiving and serious illness live in the further exceptions rather than in the eighty-hour test. Either way the practical rule is the same, and it is the one that catches people: an exemption a person does not report is an exemption they do not get.

CMS settled the paperwork question in an interim final rule titled Medicaid Program; Community Engagement Requirement for Certain Individuals, published at 91 FR 33348. Under the interim final rule, states must first attempt to verify community engagement on an ex parte basis, including hours worked, using reliable information the state already has, and only then request more from the applicant or enrollee. Verification happens at application and at renewal, and more often if a state chooses.

Ex parte means the state checks its own records instead of mailing the enrollee a form. Whether that holds up is the whole question, and one state has already run the experiment without those records.

Georgia is already running it

Georgia Pathways covers adults 19 to 64 with household income up to 95 percent of the federal poverty level with a 5 percent income disregard who are not otherwise eligible for Medicaid, and who work or engage in employment-related activities for at least 80 hours per month. Employment, education, job skills training and community service count, and the applicant has to document and report those hours both to get in and to stay in.

That federal approval is temporary. In September 2025, CMS extended the demonstration only through the end of 2026, so Georgia needs a new federal approval to keep Pathways running after that.

Getting in has been the hard part. A convenience store clerk who worked irregular hours called the Pathways customer service hotline for help and was sent through a phone tree that ended in a voicemail asking callers to leave a message. You’d go from one robot voice to another, she told ProPublica. She said no one called back, and she gave up after nearly a month of trying.

According to ProPublica, Georgia had paid Deloitte Consulting more than $50 million for software that froze and wiped applicants’ personal information, forcing them to start over. The state had told the federal government it hoped to enroll 25,000 of the 246,000 Georgians eligible in the program’s first year.

A Government Accountability Office analysis found the program had spent $54.2 million on administrative costs since 2021, against $26.1 million spent on health care, with nearly 90 percent of the administrative money coming from the federal budget. That is roughly two dollars of overhead for every dollar of care.

Georgia’s conditions did survive a court fight, and the ruling is the strongest thing supporters of the approach have. CMS had approved the eligibility conditions, then reversed itself and rescinded the work and premium requirements.

Judge Lisa Godbey Wood, ruling in August 2022, held the reversal arbitrary and capricious, finding that the agency committed at least six critical errors: first, that CMS failed to consider or weigh the likely possibility that rescinding would mean less Medicaid coverage in Georgia, and second, that it measured Pathways against a baseline of full Medicaid expansion rather than taking the demonstration on its own terms.

The court was reviewing how an agency changed its mind, not whether work requirements are wise. But the first error it named is a real argument: in a state that will not do full expansion, a conditioned program may be the only coverage on the table.

Arkansas ran this in 2018

One man held down a full-time job at a chicken processing plant and dutifully reported his hours to Medicaid officers when the Arkansas rule went live in June 2018. About three months later a pharmacist told him the state had shut off his Medicaid and he would have to pay to refill his prescription for COPD, a serious lung condition, on his own.

It was something like $800 out of pocket for his breathing treatments, said Trevor Hawkins, an attorney at Legal Aid of Arkansas who appealed to the state and got the coverage reinstated. The man lost the chicken plant job, could not hold later part-time work as his health worsened, and eventually went onto disability.

He had complied. The system simply did not know it, and the coverage went off anyway.

He was not a rounding error either. State reports put the total at 18,164 people who lost Arkansas Works coverage in 2018 due to work rule noncompliance, the last 1,232 of them effective January 1, 2019.

The published evaluation, led by the Harvard T.H. Chan School of Public Health, found the policy failed to achieve its stated goal of boosting employment and instead harmed health care coverage and access. Adults ages 30 to 49 had to work 20 hours a week, take part in community engagement activities, or qualify for an exemption. A federal judge paused the program in April 2019.

That judge was James Boasberg, and his reasoning was about process rather than policy. He held that the Secretary had failed to consider adequately the impact of the Arkansas amendments on Medicaid coverage, because he neither offered his own estimates of coverage loss nor grappled with comments in the record projecting that a substantial number of Arkansans would be disenrolled.

That door is now shut from the other side. Congress has done by statute what the Secretary could not sustain by waiver, and the argument that persuaded Boasberg, that the agency never weighed the coverage loss, does not run against an Act of Congress the way it runs against an agency approval.

Supporters read the same Arkansas year in reverse. The Foundation for Government Accountability writes that when Arkansas implemented work requirements for a small number of able-bodied adults in 2018, more than 9,000 went to work, and more than 14,000 left, and it points to Tennessee, where it says removing 170,000 able-bodied adults from Medicaid in 2005 raised labor force participation, employment and private coverage.

Both accounts can be built from the same spreadsheet, and the disagreement is not really about the count. It is about whether a person who leaves the rolls has been moved into a job or has failed a form.

The National Health Law Program put that objection to CMS directly in comments on the interim final rule filed in August 2026, arguing that Medicaid work requirements do not increase employment and instead result in substantial coverage loss which in turn leads to health and financial harm, and urging the agency to write a final rule that minimizes coverage loss.

Coverage written to switch itself off

The last thing to know is that a state that took expansion can lose it without anyone casting a fresh vote. KFF counts twelve states with trigger laws in place that would automatically end expansion or require changes if the federal match rate were to drop.

North Carolina’s runs about a sentence. If the federal medical assistance percentage for that category of individuals falls below ninety percent (90 percent), coverage for that category shall be discontinued as expeditiously as possible, and the Department must notify the legislature’s Medicaid oversight committee, the state budget office and the fiscal research staff, post a notice with a proposed effective date, and send CMS the documents needed to shut it down.

Nobody debates. The condition trips and the machinery runs.

The picture is not uniformly grim, though. Not all trigger laws would immediately end the expansion, and states are debating them in both directions, some working to remove them and others to put them in place.

Montana went the other way and removed its end date. Its expansion carried a 2025 sunset date, so the program would lapse unless the legislature acted, and in 2025 Rep. Ed Buttrey, a Republican from Great Falls, sponsored House Bill 245 to lift that sunset. The bill drew hours of testimony from program supporters and minimal appearances from opponents.

Governor Greg Gianforte signed it without a press release. Montana’s 10 percent share had come to slightly less than $100 million in 2024, against rolls that had swelled past 125,000 during the public health emergency.

How to check a state and apply

Two pages settle it for any given state. Medicaid.gov’s State Overviews page has a Select state dropdown that opens each state’s Medicaid and CHIP profile, and the related resources there link to the Medicaid, CHIP, and Basic Health Program Eligibility Levels table.

Then apply, even if you have been told you are under the line. HealthCare.gov says that in a state that has not expanded Medicaid, a person who is not eligible may have fewer options for coverage and, depending on income, might not qualify for savings on a Marketplace plan.

Apply regardless, for two reasons that hold everywhere. A household can qualify on other grounds, through children, pregnancy or a disability. And the Marketplace application does the referring: if it looks like anyone in a household qualifies, healthcare.gov sends that information to the state agency, which then contacts the applicant about enrollment.

The category itself is federal and it is not going anywhere: one subclause, one income test, written as a share of a poverty guideline that gets reissued every year. The variable is the state, and that is the one part of this anyone can look up in about two minutes.