Asking Able-Bodied Adults to Work for Medicaid Is Not Cruelty, It Is Dignity

There is a sentence that has been shouted at cameras more often than any other in the Medicaid debate, and I would like to read it back slowly: "Asking people to work is cruel." Say it out loud in an empty room. Notice how strange it sounds. We ask nurses to work. We ask teachers, welders, truck drivers and the person who makes your coffee at 5:30 in the morning. Somewhere along the way, a requirement shared by nearly every adult American became a form of cruelty only when attached to one particular benefit.

I think that claim collapses on inspection, and I will show you why. But I am going to do it the honest way, which means starting with the strongest thing the critics say, because they have one strong thing, and it is not what the cameras are shouting.

The critics' best case

Here it is, stated as its defenders would state it.

Work requirements, the critics say, do not mainly screen out people who refuse to work. They screen out people who fail to complete a reporting process. A person who works 80 hours a month still has to prove it, on time, in a format the state demands, often through an online portal. A person who works irregular hours, who has no reliable internet, or who simply does not open the right letter, loses coverage even though he met the requirement. When Arkansas introduced a requirement in 2018, researchers writing in the New England Journal of Medicine in 2019 reported that coverage dropped by many thousands of people and that the study found no significant gain in employment. If the people who lose coverage are mostly people who were already working or who were exempt, the critics conclude, then the policy fails by its own logic: it burns eligible people and produces no new work.

That is a real argument and I do not intend to caricature it. A policy should be judged by what it does, not what it intends. If the only effect of a requirement were to take insurance away from working people because of a missed form, I would oppose it myself.

So the question is not whether the paperwork critique has force. It does. The question is whether it is an argument against the principle or against a particular design. I say it is the second, and here is the rest of the answer in numbered form.

One: the principle is not new, and it worked in 1996

Start with the history, because it is the best evidence we have. On August 22, 1996, President Bill Clinton signed the Personal Responsibility and Work Opportunity Reconciliation Act, which replaced the old Aid to Families with Dependent Children program with a block grant called Temporary Assistance for Needy Families. The new law set work participation requirements for states, a five-year federal lifetime limit on most cash assistance, and a basic idea: aid is a bridge, not a destination. The bill passed with large bipartisan majorities in both chambers. Its critics predicted catastrophe. Some of the voices who opposed it said children would be thrown into destitution.

What followed was a large drop in welfare caseloads and, through the late 1990s, a rise in employment among single mothers. Honesty requires the caveat: the late 1990s were also a period of strong economic growth, and Congress had expanded the Earned Income Tax Credit earlier in the decade, so credit has to be shared among the work requirements, the tax credit and the economy. I do not pretend the 1996 law did it all alone. But the central prediction of the critics, that requiring work would produce widespread destitution, did not come to pass in the way they said it would. The law's authors expected that making aid conditional on work would move people toward work, and for many it did.

The 1996 law mattered for Medicaid in a second way. It formally separated Medicaid from cash welfare, so that a family leaving welfare for work would not automatically lose health coverage. Medicaid for low-income families continued under a provision of the Social Security Act, section 1931. That detail matters because it shows that the architects of welfare reform did not regard work and health coverage as enemies. They designed the transition so that work would not cost a family its health care. The principle, work as the expectation and support as the bridge, has a track record.

Two: what the Arkansas case and the court actually decided

Now to Arkansas, since the critics' strongest evidence comes from there.

In 2018 the Centers for Medicare & Medicaid Services approved Arkansas's waiver, under section 1115 of the Social Security Act, to require certain adults in the Medicaid expansion group to work or engage in qualifying activity for 80 hours a month. Arkansas started with adults aged 30 through 49. Section 1115 lets the Secretary of Health and Human Services approve experimental or demonstration projects that are likely to assist in promoting the objectives of Medicaid.

In Gresham v. Azar, the D.C. Circuit ruled in 2020 against the approval. The court affirmed a district court decision that had set aside the Arkansas approval; a companion Kentucky case was dismissed as moot after that state's new governor ended its project. Its reasoning is important, and it is not what the headline says. The court did not hold that a work requirement is unconstitutional, or that Congress cannot require work for Medicaid. It held that the Secretary's approval was arbitrary and capricious under the Administrative Procedure Act because the Secretary had not adequately considered whether the project would cause people to lose coverage, and the court read the central objective of Medicaid as providing health coverage. That is a ruling about how an agency reasons under a statute, not about whether the idea is legitimate.

Think about what that means. The court's reasoning was that the Secretary could not approve this under a general waiver provision without grappling with the coverage effects. I draw a separation of powers lesson of the kind conservatives tend to draw about administrative agencies: if something this significant is going to happen, Congress, not a secretary reading a waiver clause, should say so. The Supreme Court agreed to hear the case in December 2020, but the Arkansas project had lapsed and in 2022 the Court vacated the lower-court judgments and dismissed the case as moot, so no ruling on the merits came from the Court.

The remedy, then, is to ask Congress to write the requirement into the statute. And Congress has moved in that direction: the budget reconciliation law of 2025 included Medicaid community engagement requirements of at least 80 hours a month for certain expansion adults aged 19 through 64, beginning around the end of 2026 unless a state wins a delay. The details and dates are the kind of thing that is changing, so check the text on Congress.gov rather than trusting a blog post, including this one.

Three: the paperwork problem has a design answer

This is the part the cameras skip. If the central complaint is that eligible people fall through the cracks, the solution is to fix the cracks, not to abandon the principle.

How? Think of what a competent state already knows. States hold wage records from the unemployment insurance system. They see enrollment in SNAP and cash assistance. They hold data on school enrollment and on people with disabilities. An intelligent system does not ask a working person to prove what the state's own wage data already shows. It checks the data first, a process often called ex parte verification, and only asks the individual to document the activity when the data does not answer.

Good design has several features:

• Verify from government data first. If the state's records show earnings that meet the hours, the person is done.
• Exemptions written plainly. Those who are medically frail or disabled, caregivers of young children or of an incapacitated family member, full-time students, people in treatment for substance use disorder, and pregnant women are exempt. Programs that have been built on this model all list such groups. I would write them into any version I supported.
• Qualifying activities, not just jobs. Volunteering, job training, education, and job search count. The point is engagement, and the requirement should be wide enough to fit real life.
• A cure period. A person who misses a report gets notice and time to fix it before coverage ends.

That is what Georgia's Pathways program, which started in July 2023, tried to do. Georgia did not expand Medicaid fully. It created a program for adults with incomes up to the federal poverty level who engage in 80 hours a month of work, education, job training or community service. Critics noted, fairly, that enrollment was slow and that the reporting burden was a factor. Supporters responded that the program is new and still working out the verification process. I think both are right, and the useful lesson is the one I have been pressing: the principle can survive a bad form, but the form is the thing to fix.

Four: the moral argument is not an afterthought

Now the part the cameras do not know how to handle. Why do I call work a dignity?

Because of what work does for a person. A job is not only a paycheck. It is a schedule, a set of colleagues, a skill that compounds, a reference, and the plain experience of being needed. A person who works for 80 hours a month, which is about twenty hours a week, is building something even if the wage is low: a record, a habit, a ladder. The research on long gaps in employment is consistent in one way: the longer a person is out of work, the harder it becomes to get back in. A policy that treats a healthy 35-year-old as someone who cannot be asked to do anything is a policy that treats him as a permanent dependent. That is not compassion. That is a verdict on his abilities, and it is one I would not want applied to me.

The Judeo-Christian tradition that I and many of my readers hold treats work as part of what it means to be human, not as a punishment. Genesis puts Adam in the garden "to dress it and to keep it" before the fall, in the language of the King James Version, which is to say that work comes first and toil is what the fall added. I would not use Scripture as a club, and the same tradition commands generosity to the poor, the widow and the stranger. The verse people quote at this debate, from Paul's second letter to the Thessalonians, is about those who will not work, which is different from those who cannot. A fair work requirement draws exactly that line. It supports the person who cannot and asks something of the person who can. That distinction is not cruelty. It is the oldest moral distinction in the book.

There is a Founding-era point here too. The states are free to try different approaches, and one of the glories of the federal structure is that they can. Justice Louis Brandeis, dissenting in New State Ice Co. v. Liebmann (1932), described the states as laboratories where a single courageous state might try a social experiment without risk to the rest of the country. Welfare reform in the 1990s began in state experiments, and Congress then wrote the winners into law. Medicaid work requirements are a case where state experiments should be run, measured and reported, not thrown out because the word "work" offends someone.

Five: answering the other objections

A few more claims deserve a quick reply.

"Most Medicaid enrollees already work." Many do, and that is a reason the requirement should be easy to satisfy and easy to prove, not a reason for it not to exist. If most already comply, a well-designed rule would affect few people other than those who need a nudge. The people who would be affected by a well-built rule are the ones who are not working and are not exempt.

"Work requirements do not increase employment." The evidence on this is mixed, and I will not claim a result I cannot support. The Arkansas data did not show a clear employment gain over the short period studied. But a policy can be justified by reciprocity and by dignity, not only by a measured jump in employment. We do not require evidence of a jump in attendance before we have a rule that children go to school.

"People have unpredictable hours." Then the requirement should be monthly rather than weekly, count volunteer and training hours, and allow a good-cause exception for illness or a family emergency. That is a design instruction, not an argument for no requirement.

"The money saved is small." Perhaps. I am not selling this as a budget gimmick. If the goal were purely to save money, there are bigger levers, and I discuss one of them in The Medicaid Match That Pays States More for Adults Than for Disabled Kids. The argument here is about what a program says to the person who uses it.

A word about the coverage

One last jab, because I cannot help myself. When a policy is described as "cruel," that word does an enormous amount of work. Notice how often the story begins with the one hard case, the person who lost coverage because of a form, and ends there, without asking whether anyone gained a job, a skill or a better schedule. I do not suggest that the hard case is made up. I suggest that a story that counts only one kind of outcome is not a study; it is an advertisement. The fix for the missed form is a better form. The fix for the missing work is a requirement. We can have both.

For related arguments about what the incentives of welfare programs do to families, see the marriage penalty buried in welfare programs, with the math shown. For the moral argument about work itself, Work as a Calling, What Scripture and the Reformers Said About an Ordinary Job covers the tradition, and Booker T. Washington Built Tuskegee on the Dignity of Work and Ownership covers one of its great American voices.

The case in brief

1. The critics' best point, that paperwork drops eligible people, is real, and it argues for good design: verify from state wage data first, write the exemptions plainly, count training and volunteering, and give a cure period.
2. The 1996 welfare law, which linked aid to work and kept families' Medicaid intact through section 1931, did not produce the destitution its critics predicted, though a strong economy and the Earned Income Tax Credit shared the credit.
3. Gresham v. Azar (2020) struck down the Arkansas approval because the Secretary failed to weigh coverage effects under the Administrative Procedure Act, not because Congress cannot require work, which is a reason for Congress to write the rule itself.
4. Work builds a record, a habit and a ladder, and the moral line between those who cannot work and those who will not is the oldest in the tradition. A requirement with proper exemptions respects it.

What you can do: read the text of your state's Medicaid work or community engagement rules on your state Medicaid agency's website, and if you see a verification step that asks a worker to prove what the state's own wage records already show, send that example to your state legislator.