Medicaid Work Requirements 2027: Why CMS-2454-IFC Forces a Managed Care Capitation Reset

CMS published CMS-2454-IFC in June 2026 (Federal Register Document No. 2026-11094), implementing the community engagement requirement mandated by Section 44141 of the One Big Beautiful Bill Act, with a national compliance deadline of January 1, 2027.

Managed care rates for periods beginning July 2026 through June 2027 are being built now against 2024 and 2025 claims data that contains none of it. The people who exit first are not a random cross-section of the enrolled population.

Key Takeaways

• The rule imposes an 80-hour monthly community engagement standard on non-pregnant adults aged 19 to 64 in the ACA expansion group, satisfiable by documenting monthly earned income of at least $580.

• CMS projects enrollment losses of 2.3 million in FY2027, rising above 3 million later; CBO put the steady-state figure at 5.2 million by 2034 with 4.8 million becoming uninsured.

• Arkansas's 2019 pilot disenrolled roughly 18,000 people in seven months, about 1 in 4 of those subject, though more than 95% already met the requirement or should have been exempt.

• Section 41106 of OBBBA blocks premium tax credits for work-requirement noncompliers regardless of income, so the healthy exits do not reappear in the individual market.

• MACPAC measured a mean 7.4-month lag between a program change's effective date and the certification update reflecting it.

What the Rule Requires and Whom It Exempts

The exemption list is the actuarially interesting half, because it defines by exclusion the population MCOs keep.

CMS-2454-IFC applies an 80-hour monthly community engagement standard to non-pregnant adults aged 19 to 64 enrolled through the ACA expansion group or certain Section 1115 demonstrations. Qualifying activities include employment, a qualifying work program or job skills training, half-time postsecondary or vocational enrollment, and volunteer service. Documented monthly earned income of at least $580, which is 80 hours at the $7.25 federal minimum wage, also satisfies it.

Exempt are the medically frail, pregnant and postpartum individuals through 12 months post-delivery, parents and caretakers of children under 14, current and former foster youth through age 26, individuals within three months of release from incarceration, American Indians and Alaska Natives, veterans with a total disability rating, individuals in substance use disorder treatment, and those whose TANF participation already satisfies the requirement. Inability to find work in a high-unemployment labor market is not an exemption.

Forty-three states plus the District of Columbia have expanded Medicaid and are subject to the rule; the seven non-expansion states are not. Implementation generally begins January 1, 2027, with extensions available through December 31, 2028 for states that document an operational justification CMS approves.

The Residual Reprices Itself Without Anyone Getting Sicker

Under 42 CFR 438.4 capitation rates rest on base period experience trended forward for the expected covered population. This rule changes the population, not the trend.

Compliance correlates with health. Adults who can work 80 hours a month are on average younger, healthier, and cheaper than the full expansion population, and every exemption category names a higher-acuity group. So the members who leave are disproportionately work-capable and moderate-acuity, and the members who stay are disproportionately the ones the exemptions protect.

Arkansas is the closest evidence. Its 2019 requirement disenrolled roughly 18,000 people in seven months, about 1 in 4 of those subject, even though more than 95% already met the requirement or should have been exempt. A Commonwealth Fund analysis found a 4.4 percentage point rise in uninsurance among 30-to-49-year-olds below 300% of FPL, with no measurable increase in employment.

That points to two disenrollment layers with different pricing consequences. The procedural layer, people who miss notices, lack documentation capacity, or fall through incomplete electronic verification, is mixed in health status. KFF found approximately 69% of 2023-2024 unwinding disenrollments were procedural, which applied to CMS's 2.3 million FY2027 estimate implies roughly 1.6 million procedural losses in year one. The compliance layer that follows is the systematic one, concentrated in working-age moderate acuity.

The mechanical result is risk score migration. CDPS+Rx scores, which most states use to acuity-adjust capitation, are calibrated prospectively on diagnostic history. As lower-acuity members exit, the residual's average score rises with no change in any individual's health, and a certification that adjusts enrollment volume without adjusting the assumed average score prices the wrong book.

Volume and cost do not scale together here. An MCO managing 200,000 expansion members and retaining 180,000 does not hold a proportionally smaller version of the same book. Stop-loss attachment points calibrated on the pre-rule mix are breached more often, and specialty drug and gene therapy exposure concentrates in exactly the medically frail and disabled categories the exemptions retain. Where specialty pharmacy is carved out of capitation, that shift lowers per-member cost inside the capitation base while raising carve-out liability, which can read as improved MCO performance if the two are not tracked against composition together.

The Valve That Would Normally Relieve This Is Closed

In an ordinary adverse selection scenario, some of the departing healthy population reappears elsewhere and the shift is partly absorbed. Not here.

Section 41106 of the One Big Beautiful Bill Act bars premium tax credits for individuals disenrolled from Medicaid for failing the community engagement requirement, regardless of income. A 35-year-old earning $28,000 who could not document hours cannot buy a subsidized marketplace plan. He can buy an unsubsidized one, which CMS enrollment data suggest fewer than 3% of people at that income do. That is the mechanism behind CBO's finding that 4.8 million of 5.2 million losers become uninsured rather than moving to other coverage, at $344 billion of federal Medicaid savings over ten years.

For managed care pricing this removes the offset. The healthy exits do not stabilize an adjacent pool; they land in uncompensated care at hospitals and safety-net providers, while the MCO keeps the higher-acuity residual on rates built for a population that still included them. It also means ACA rate filings should not expect meaningful inflow from this group, which removes an anticipated morbidity pressure and the partial redistribution at the same time.

The correction lag makes it durable. MACPAC measured a mean 7.4-month delay between a program change's effective date and the certification update reflecting it, which for a January 2027 change puts corrected rates around August 2027. MCOs whose contracts allow renegotiation only at annual renewal have no interim mechanism and carry a full rating period at pre-rule rates.

And there is no methodology to point at. CMS issued the 2025-2026 Managed Care Rate Development Guide in August 2025 and the 2026-2027 guide in February 2026, neither written against an effective rule and neither addressing work requirement composition. States certifying January 2027 rates are building the adjustment for the central problem in their certifications without federal guidance on how, alongside a parallel six-month redetermination cycle competing for the same eligibility systems.