The Attrition Policy

Work Rules, Lost Children, and a Coverage Cliff Nobody Timed

Somewhere this month, in a state that has not yet made the news, a mother will miss a reporting deadline. She works. She has always worked. The notice arrived in an envelope she mistook for an advertisement, or in an online portal she could not log in to, or in a language she does not read most easily. The window to confirm her hours closes. Nothing happens that day. Nothing happens for weeks. Then her child develops an ear infection, and the pharmacy declines the prescription, and she learns that her family lost its Medicaid coverage sometime in the interval between the missed form and the fever.

She is no one in particular, and she is everyone the rule will catch. No one decided to take that child’s coverage. That is the point. The system did not target her. It simply waited for her to fail a clerical task, and when she did, it processed the result.

This is the part of the reckoning that never reaches a courtroom. I wrote in the first installment that the funding side of the wound had a court docket, that the cuts to rural hospitals and clinics would be fought in litigation and appropriations fights anyone could follow. This installment is about the part that produces neither a filing nor a headline. It is about coverage that simply lapses. And the central claim is uncomfortable but documented: the new Medicaid work requirements do not remove people who are ineligible — they remove people who are eligible but cannot survive the paperwork. The first casualties are already visible, before the requirements have even started. The clearest casualties are children, a cohort that is shrinking even as we withdraw its support.

We already ran this experiment once. Arkansas imposed Medicaid work requirements in 2018. Over four months, roughly 18,000 adults lost coverage. The overwhelming majority subject to the rule, on the order of ninety-five percent, were already working or should have qualified for an exemption. Employment did not rise. Uninsurance did. A federal court halted the program in 20191,2. The experiment failed on its own stated terms. It did not move people into jobs. It moved people out of coverage.

Here is the detail that should stop the reader cold. The federal rule now implementing nationwide work requirements recites Arkansas history in its own text. The precedent predicting the policy’s failure is documented in the document that imposes it. That is not an accusation. It is a sequence, and the sequence is the story.

The Design Is the Point

On June 1, 2026, its statutory deadline, the Centers for Medicare and Medicaid Services published the Interim Final Rule that turns the work-requirement provision of H.R. 1 into operating instructions3,4. The rule takes effect July 31, 2026. States must begin outreach by August 31. Full implementation is required across all forty-one expansion states and the District of Columbia by January 1, 2027.

The mechanism is straightforward to describe and punishing to comply with. An enrollee in the adult expansion group, ages nineteen to sixty-four, must demonstrate eighty hours a month of work, community service, a qualifying work program, half-time education, or roughly $580 in monthly earnings. Compliance is verified at application and again at renewal3.

The rule does not stop at the statute. It exceeds it. Congress wrote an exemption for the medically frail. The rule sets a verification standard stricter than the law requires, obligating the frail to prove every six months that their condition prevents them from performing eighty hours of qualifying activity. That standard reportedly surprised the states now scrambling to build the systems to administer it5. The filter is tighter than the statute it implements. That gap is a fact, not a charge.

Now the numbers, drawn from the agency’s own projections rather than its critics’. CMS estimates that roughly 2.3 million people will be disenrolled in fiscal year 2027 alone, a combined loss of about 15% of adult-group enrollment3. The Congressional Budget Office separately estimates that the work-requirement section alone will raise the number of uninsured Americans by 5.3 million in 20346. Set against those figures is a finding the agency does not dispute: roughly nine in ten expansion enrollees subject to the requirement already work or qualify for an exemption7. The rule is not built to find the small fraction who could work and do not. It is built to process the large majority who already comply but will struggle to prove it.

For a long time, that last claim rested on inference. It no longer does. The Commonwealth Fund finds that roughly seventy percent of Medicaid disenrollments since the pandemic-era unwinding were procedural, meaning paperwork failures rather than actual ineligibility, and that one in ten enrollees lose and regain coverage within twelve months. Moving expansion adults from annual to six-month redeterminations, the same analysis warns, will only increase that churn. Most striking of all, the rule’s own analysis “expects a significant share of disenrolled members to be exempt individuals who simply fail to document compliance”8. The design defeats the eligible by its own admission.

Follow the money to its actual source. The savings this policy claims are not savings from reduced fraud. They are savings from reduced enrollment among eligible people. The dollars are recovered by shrinking the rolls, and the rolls shrink fastest where the paperwork is hardest. This is not a work policy. It is an attrition policy, and its most efficient targets are the people least able to file.

The Cliff Nobody Timed

The harm is arriving ahead of schedule. That was the spine of the first installment, and it holds here. The deepest cuts in H.R. 1 have not yet taken effect. The work requirements have not started. And yet the damage to children is already measurable.

As of April 2026, two million fewer children were enrolled in Medicaid and CHIP than in January 2025, a decline that, in the words of the analysts who tracked it, “landed before the H.R. 1 cuts have largely even kicked in”9.

Honesty requires a clear statement here, because the argument is stronger for it. That two-million figure is not yet caused by work requirements. The drivers so far are redetermination churn, the chilling effect on mixed-status families, and a decline in subsidized marketplace coverage. The work requirements are the accelerant still to come. The cliff is visible because the ground has already started to give way beneath it. A Congressional Budget Office projection reported by the same analysts has the number of children on Medicaid falling by three million between 2026 and 20369.

The clock is no longer hypothetical. Nebraska became the first state to begin Medicaid work requirements on May 1, 2026. Other states are moving early. Arkansas, the state whose 2018 experiment failed, runs a soft launch on July 1, 2026, checking compliance but not yet disenrolling anyone until January 202710. The state that proved the policy does not work will run it again first.

Look at where the children are disappearing, and the partisan reading does not hold. The ten states with the largest absolute losses in child enrollment, which together account for sixty-seven percent of the national decline, span both parties without distinction: California, Texas, Indiana, Florida, Georgia, Illinois, Ohio, Arizona, Virginia, and Kentucky. Indiana leads in percentage terms, down 19.8 percent and more than 174,000 children, after layering its own administrative requirements on top of the federal ones. California leads in absolute terms, more than 382,000 children gone11. Red states and blue states, governed differently and voting differently, are producing the same result. That is the clearest evidence available that this is a problem of design, not of party. The administrative architecture does not care who built it.

Walk it down to a single state, and the abstraction acquires a face. In Colorado, the uninsured rate among low-income children doubled in a single year, to roughly ten percent. More than 616,000 Colorado children rely on Medicaid or the state children’s plan, and more than 350,000 on SNAP12. A doubling is not a statistical wobble. It is one in ten poor children in a single state losing the thing that pays for the doctor, in twelve months, before the policy that will accelerate it has even begun.

The strongest objection deserves a fair hearing, so here it is. The Paragon Health Institute argues that child Medicaid and CHIP enrollment “remains above pre-pandemic levels,” and, on the raw number, that is true13. But the comparison hides the mechanism. A return toward pre-pandemic enrollment is not neutral when it is engineered through paperwork attrition among children who still qualify, and when the trend line is pointed down ahead of the very rule designed to bend it further. The question is not whether the number is higher than it was in 2019. It is who is being removed, why, and in which direction the line is now moving. The answer to all three is the same, and it is not reassuring.

The Welcome Mat in Reverse

Children are not subject to work requirements. No one proposes that a six-year-old document eighty hours of labor. And yet children are among the policy’s most reliable casualties, because coverage in a household is connected. When a parent falls off the rolls for failing to report, the children frequently follow, even though the children’s own eligibility never changed.

The estimate is precise enough to quote. For every hundred expansion adults who lose Medicaid, roughly five children lose it too14. Applied to the work-requirement population, that mechanism translates to between 480,000 children, counting only the children of expansion adults, and 914,000 children, counting all non-disabled adults affected. The children lose coverage not because anyone decided they should, but because the adult who files the paperwork could not.

Return to the institutional side of this same failure. In the first installment, Bluestem Health, a community health center in Nebraska, projected losing roughly $600,000 a year as patients lost coverage, most of them not because they became ineligible but because they could not navigate the renewal15. Set the two facts side by side. Bluestem is the institutional version of the bleed. The two million children are the household version. Same paperwork failure, viewed from two vantage points. No one argues that a child should be working, which is precisely why children expose the design so cleanly. Strip away the rhetoric about personal responsibility, and what remains is a clerical filter that catches the people it was never honestly meant to catch.

The Same Hand on a Second Program

It would be a mistake to read the Medicaid changes in isolation, because they are not arriving in a vacuum. SNAP, Medicaid, and CHIP are not three separate cuts. They are one design: expanded work requirements, narrowed exemptions, and a cost shift onto the states, applied across three programs that land on the same households at the same time.

The SNAP changes follow the Medicaid template almost exactly. The work-requirement age ceiling rises from fifty-four to sixty-four. For the first time, the requirement applies to parents whose youngest child is between fourteen and seventeen. Long-standing exemptions for veterans, the homeless, and former foster youth are struck16,17. The scale is the largest in the program’s history: a $187 billion cut and a projected average monthly drop of 2.4 million in participation18.

This, too, is already happening. More than 3.5 million people, roughly eight to nine percent of all recipients, have lost SNAP since July 2025, leaving about 38 to 39 million still enrolled19,20. And the pediatric share of that loss is now counted, not estimated. The Center on Budget and Policy Priorities finds that at least 700,000 children are among those already cut from food assistance21. CBO estimates that the changes will also reduce child nutrition program subsidies for roughly 96,000 children18.

Here, the mechanisms compound. The family that misses the Medicaid reporting window is, very often, the same family that hits the new SNAP rules. Lose the coverage, then lose the food benefit, and the child absorbs both. The less a parent can navigate the paperwork, the less the child eats and the less the child sees a doctor. One paperwork failure, two programs, one child. The convergence is not a rhetorical flourish. It is now visible in the same 700,000 children counted on the food side and the two million counted on the coverage side, often the very same households.

A Generation We Are Choosing Not to Count

Step back far enough and a harder question comes into view, one that turns this from a humanitarian concern into a strategic one. We are withdrawing health and nutrition support from a generation of children that is, in raw numbers, shrinking.

The United States’ total fertility rate fell to 1.62 in 2025, well below the replacement level of roughly 2.1, and the Congressional Budget Office projects it to remain flat through 205522,23. Births continue to fall. This is not a temporary dip. It is the demographic baseline the country will live with for a generation or more.

A nation does not under-protect a shrinking cohort by accident of arithmetic. These children are the human-capital base of every future the country claims to want — the workforce, the tax base, the enlisted ranks, the caregivers for an aging population. To contract the safety net beneath them at the precise moment they are becoming scarcer is not only a moral failure. It is a strategic self-injury, the same confusion of short-term savings with lasting strength that the first installment on the funding side is named after.

I will not resolve the larger question here because it has not yet earned its answer. But it is worth naming plainly. We are tightening the net beneath a generation we are already failing to replace, and a choice like that does not stay a line item in the budget. It becomes a verdict, and the verdict comes due whether or not we ever read it back to ourselves. The series will return to that question when the cost is no longer a projection. For now, it is enough to name it and let it stand.

The Wreckage and Who Buys It

There is a financial story downstream of all this, and it is not incidental. It is the destination toward which the coverage cliff flows.

Start with the balance sheet, because the household failure is also an institutional one. When eligible patients lose coverage due to paperwork, each lost patient can turn a reimbursable visit into a write-off. Rural hospitals feel that loss as delayed reimbursement, uncompensated care, and mounting pressure on services that were already fragile8. Some 41.2 percent of rural hospitals are already operating in the red, and 417 facilities are considered vulnerable to closure. The coverage cliff and the closure cliff are the same cliff, viewed from different floors of the same building.

The people who deliver the babies have said so. The major obstetric and gynecologic societies, led by the American College of Obstetricians and Gynecologists, issued a joint statement opposing the Interim Final Rule24. When the clinicians who staff the labor-and-delivery units are on record against the policy, that is not a political objection. It is a clinical one.

Now, the part that follows from financial logic rather than malice. Roughly 756 rural hospitals are at risk of closure, with more than 40% at immediate risk, and reserves that cover at most two to three years of losses. H.R. 1 is projected to cut rural Medicaid spending by about $137 billion over ten years25. A distressed asset is an acquisition target, and the private-equity playbook for distressed hospitals is well documented. Acquire the hospital. Sell its real estate through a sale-leaseback to generate immediate cash. Expand on leveraged debt. Extract returns while the operator runs the business and pays rent on buildings it once owned. When reimbursement tightens and labor costs rise, there is no cushion left because the cushion was sold for cash years earlier.

We have already watched this run to its conclusion. Steward Health Care, private-equity-acquired in 2010, sold its real estate to a hospital landlord, saw roughly $800 million extracted by its sponsors, and filed for Chapter 11 in 2024 with $9.2 billion in liabilities. Thirty-one hospitals across eight states were caught in the failure. Five ultimately closed, displacing about 5,000 workers, and the communities they served got nothing25. Steward was not rural, but it is the model, the same financial engineering aimed at the same kind of dependent community.

The timing favors the buyers. PitchBook’s first-quarter 2026 healthcare-services report finds deal volume down but dry powder and exit pressure building, with the payer subsector tied to Medicaid and Medicare exposure the hardest hit as reimbursement risk moves to the center of every diligence file26. Distress plus capital waiting to deploy is the classic acquisition setup, and the policy is manufacturing the distress.

The same pressure is reshaping the practices where patients actually receive care. Just 42.2 percent of U.S. physicians worked in private practice in 2024, down from 60.1 percent in 2012, the lowest share the American Medical Association has ever recorded, an eighteen-point drop in a dozen years. Only 35.4 percent held any ownership stake, down from 53.2 percent27,28. The share of physicians in private-equity-owned practices rose from 4.5 percent in 2022 to 6.5 percent in 2024. And the reason physicians cited most often for selling was inadequate payment rates, named by 70.8 percent of them. That is the precise mechanism this series indicts: withdraw the revenue, and the practice that serves the community is handed to private capital that did not build it.

Follow the full sequence, and the design becomes legible. The law withdraws the revenue. The asset fails. The failed asset is bought cheaply. And federal rural-rescue dollars then flow toward the acquirer that bought it at the bottom. No villain is required to explain any single step. The sequence explains itself.

Do not leave that sequence at altitude, because its endpoint is a person. When a rural hospital closes or consolidates, the attached primary care, obstetric, and emergency capacity goes with it, and the surrounding population loses proximate access to care. The realized chain is already documented, not projected. Since 2020, more than 116 rural labor-and-delivery units have closed. Only 41 percent of rural hospitals still provide maternity services. In roughly a third of U.S. counties, there is not a single obstetric provider or birthing facility25. This is the bridge back to the children. A shrinking generation is being born into counties with no place to be born. The institutional failure becomes a parent driving an hour further to deliver a child, in a country that has decided it can spare neither the hospital nor the coverage.

What We Will Not Be Able to Say

Return to the mother and the missed form, because the whole policy is contained in that small, undramatic event. She did nothing wrong that the system was honestly trying to catch. She worked, and she lost her coverage, and her child lost it with her, and the savings the policy will claim are in some measure the value of her family’s care, recovered by her inability to file.

Across three programs, the architecture is identical. Expand the work requirement. Narrow the exemptions. Shift the cost. Then let attrition do the rest, quietly, without anyone having to cast a vote on any single family. SNAP, Medicaid, and CHIP are not three policies. They are one method, applied three times, to the same households, in the same months.

What makes this installment different from the last is that nothing here is a forecast. The two million children are already gone. The 700,000 are already hungry. The hospitals are already in the red, and the buyers are already circling. The work requirements that will deepen all of it have not even started. When they do, and the numbers turn from projection into casualty, we will not be able to say we did not know. The number of children was the warning, and it was delivered before the policy began. We were told, in advance, exactly what this would cost and exactly who would pay it. The only open question is what we do with a warning we cannot claim we never received.

Endnotes

1. Sommers, B. D., et al. (2019). Coverage losses, substantial confusion in Arkansas following implementation of Medicaid work requirements. Harvard T.H. Chan School of Public Health / New England Journal of Medicine. https://hsph.harvard.edu/news/coverage-losses-substantial-confusion-in-arkansas-following-implementation-of-medicaid-work-requirements/

2. Urban Institute. (2025). New evidence confirms Arkansas Medicaid work requirement did not boost employment. https://www.urban.org/urban-wire/new-evidence-confirms-arkansas-medicaid-work-requirement-did-not-boost-employment

3. JD Supra. (2026). CMS issues interim final rule imposing Medicaid work requirements. https://www.jdsupra.com/legalnews/cms-issues-interim-final-rule-imposing-4510270/

4. National Law Review. (2026). CMS issues interim final rule: Medicaid work requirements. https://natlawreview.com/article/cms-issues-interim-final-rule-medicaid-work-requirements

5. Politico Pulse. (2026, June 8). States scramble on Medicaid work rule. Politico. https://www.politico.com/newsletters/politico-pulse/2026/06/08/states-medicaid-work-rule-scramble-00953054

6. Congressional Budget Office. (2025). Supplemental cost estimate: Medicaid provisions of P.L. 119–21 (H.R. 1). https://www.cbo.gov/system/files/2025-10/PL-119-21-Medicaid%20_0.pdf

7. Georgetown University Center for Children and Families. (2025, May 19). Medicaid work requirements could threaten parents’ and children’s coverage and well-being. https://ccf.georgetown.edu/2025/05/19/medicaid-work-requirements-could-threaten-parents-and-childrens-coverage-and-well-being/

8. Commonwealth Fund analysis, reported via AmeriTrust. (2026). New Medicaid rules threaten to deepen the rural healthcare crisis. PR Newswire. https://www.prnewswire.com/news-releases/new-medicaid-rules-threaten-to-deepen-the-rural-healthcare-crisis-302800067.html

9. Georgetown University Center for Children and Families. (2026, May 28). Two million fewer children are enrolled in Medicaid since the start of 2025. https://ccf.georgetown.edu/2026/05/28/two-million-fewer-children-are-enrolled-in-medicaid-since-trump-took-office/

10. KFF. (2026). An early look at policy decisions as states get ready to implement work requirements. https://www.kff.org/medicaid/an-early-look-at-policy-decisions-as-states-get-ready-to-implement-work-requirements/

11. Georgetown University Center for Children and Families. (2026, June 10). Which states are seeing the largest declines in child Medicaid and CHIP enrollment. https://ccf.georgetown.edu/2026/06/10/which-states-are-seeing-the-largest-declines-in-child-medicaid-chip-enrollment/

12. Denver7 / Colorado Children’s Campaign. (2026). Colorado children losing health insurance as new report reveals widening gaps in care (Kids Count 2026). https://www.denver7.com/news/local-news/colorado-children-losing-health-insurance-as-new-report-reveals-widening-gaps-in-care

13. Paragon Health Institute. (2026). Child Medicaid and CHIP enrollment remains above pre-pandemic levels in 2026. https://paragoninstitute.org/paragon-pic/child-medicaid-and-chip-enrollment-remains-above-pre-pandemic-levels-in-2026/

14. Center for American Progress / Manatt Health. (2025). The collateral damage of Medicaid work requirements. https://www.americanprogress.org/article/the-collateral-damage-of-medicaid-work-requirements/

15. Georgetown University Center for Children and Families. (2026, May 1). Rural hospitals and communities feeling impact of H.R. 1 Medicaid cuts as rural health fund falls short. https://ccf.georgetown.edu/2026/05/01/rural-hospitals-and-communities-feeling-impact-of-h-r-1-medicaid-cuts-rural-health-fund-falls-short/

16. Penn Leonard Davis Institute of Health Economics. (2026). Even when SNAP payments resume, more food cuts will affect millions of Americans. https://ldi.upenn.edu/our-work/research-updates/even-when-snap-payments-resume-more-food-cuts-will-affect-millions-of-americans/

17. National Association of Counties. (2025). H.R. 1 and the Supplemental Nutrition Assistance Program (SNAP): What counties should know. https://www.naco.org/resource/hr-1-and-supplemental-nutrition-assistance-program-snap-what-counties-should-know

18. Congressional Budget Office. (2025). Estimated budgetary effects of SNAP provisions (Pub. №61367). https://www.cbo.gov/system/files/2025-08/61367-SNAP.pdf

19. Ms. Magazine. (2026, June 9). SNAP cuts under the One Big Beautiful Bill Act. https://msmagazine.com/2026/06/09/snap-trump-one-big-beautiful-bill-act-food-stamps-hunger-congress-republicans/

20. CNBC. (2026, May 30). SNAP food stamps and the Big Beautiful Bill. https://www.cnbc.com/2026/05/30/snap-food-stamps-big-beautiful-bill.html

21. Common Dreams, reporting Center on Budget and Policy Priorities estimates. (2026). Supplemental Nutrition Assistance Program coverage. https://www.commondreams.org/tag/supplemental-nutrition-assistanc

22. Congressional Budget Office. (2026). The demographic outlook: 2026 to 2056 (Pub. №61164). https://www.cbo.gov/publication/61164

23. NPR. (2026, April 9). U.S. birthrate, babies, and immigration. https://www.npr.org/2026/04/09/nx-s1-5779627/birthrate-united-states-babies-immigration

24. American College of Obstetricians and Gynecologists. (2026, June). Ob-gyn societies issue joint statement opposing Medicaid interim final rule. https://www.acog.org/news/news-releases/2026/06/ob-gyn-societies-issue-joint-statement-opposing-medicaid-interim-final-rule

25. Gavin/Solmonese. (2026). Rural hospital bankruptcy: distress, consolidation, and private capital. https://gavinsolmonese.com/insights/rural-hospital-bankruptcy/

26. PitchBook. (2026). Q1 2026 healthcare services report. https://pitchbook.com/news/reports/q1-2026-healthcare-services-report

27. American Medical Association. (2025). Smaller share of doctors in private practice than ever (Physician Practice Benchmark Survey). https://www.ama-assn.org/practice-management/private-practices/smaller-share-doctors-private-practice-ever

28. American Medical Association. (2025). Policy Research Perspectives: Physician practice arrangements, 2024. https://www.ama-assn.org/system/files/2024-prp-pp-characteristics.pdf