Section 44108 of the One Big Beautiful Bill Act requires every state to redetermine Medicaid eligibility for expansion adults every six months rather than annually, effective December 31, 2026, alongside 80-hour monthly work requirements for most of that population. The Congressional Budget Office puts the Medicaid and CHIP provisions at 7.5 million additional uninsured by 2034. The rate-setting problem is not the level. It is that the denominator now moves twice a year.

Key Takeaways

  • 69% of unwinding-era disenrollments were procedural, meaning coverage was lost to paperwork rather than to a finding of ineligibility, across more than 25 million disenrollments through September 2024.
  • 2.0 to 3.1 million expansion adults lose coverage from six-month redetermination alone by 2028 in Urban Institute modelling, rising to 4.9 to 10.1 million once work requirements are included.
  • 97% of Arkansas enrollees who lost coverage under its 2018-2019 work requirement were either compliant or exempt, and lost it on reporting failure.
  • $200 million in FY 2026 is the federal implementation allocation, which HMA's analysis says "comes nowhere close" to what states must spend.

What the Provision Changes

The six-month cycle applies to adults aged 19 through 64 who qualify through the ACA expansion pathway, a group that peaked near 21.2 million during continuous enrollment and stood at roughly 18.7 million in early 2026. Traditional Medicaid eligibility groups keep the annual cycle.

Exemptions from the work requirement are narrower than many state proposals contemplated: parents or caretakers of children under 14, pregnant and postpartum individuals through 12 months postpartum, current and former foster youth to age 26, people released from incarceration within three months, American Indians and Alaska Natives, veterans with total disability ratings, the medically frail, TANF work-requirement completers, certain SNAP households, and people in substance use treatment.

The verification design is the part that determines the numbers. States must use electronic data sources to check compliance "to the greatest extent possible" before asking the enrollee for documentation, and where matching fails the burden moves to the enrollee. That makes the disenrollment rate a function of each state's data infrastructure across 40 expansion states rather than a function of eligibility.

Metric Unwinding (12-Month) Projected (6-Month)
Disenrollment rate per cycle 31% 25-35% (est.)
Procedural share of disenrollments 69% 72-78% (est.)
Redetermination cycles per year 1 2
Annual exposure to procedural loss 1x 2x (compounding)
Expansion enrollment reduction (Urban Inst.) Baseline 2.0-3.1M by 2028

The Filter Removes the Cheap Members First

The 2023-2024 unwinding is the closest empirical analogue, and its headline is the split rather than the total. KFF tracked more than 25 million disenrollments through September 2024, 31% of completed renewals, of which 69% were procedural. State variation was extreme: 93% procedural in Nevada and New Mexico against 22% in Maine, which had paused procedural disenrollments in August 2023.

Arkansas ran the work requirement version. In the seven months before a federal court blocked it, 18,000 enrollees lost coverage, roughly 25% of those subject to the requirement, and researchers found 97% of them were compliant or exempt. The losses concentrated among people facing housing instability, limited internet access, or cognitive and behavioral health conditions not formally classified as medically frail.

A paperwork filter does not select at random. Younger adults with fewer chronic conditions and lower utilisation are the ones who let a renewal notice go unanswered; enrollees with ongoing prescriptions and standing provider relationships complete the form. Each cycle therefore removes a disproportionately healthy slice and concentrates a sicker residual, the same morbidity transfer that shows up on the marketplace side of the move.

That lands directly in the capitation rate. If a state loses 10% of enrollment to a cycle and the departing members were 15% less costly than those retained, the surviving population's PMPM rises by roughly 1.5 to 1.7% from mix alone, before any medical trend. Under ASOP No. 49 and 42 CFR 438.4 the certifying actuary has to separate that mix shift from genuine trend, and the shift now occurs twice per rating year.

The base period cannot supply the answer. Rates developed in January 2027 off 2024-2025 encounter data extrapolate from a population that lived under annual redetermination to one that will pass through two cycles inside the prospective year. Decomposing mix from trend needs cohort-level data at the granularity of who left and when, which many state Medicaid programmes do not currently produce.

Re-entry compounds it within the period. Procedurally disenrolled members return with deferred care, medication restarts and catch-up utilisation, and under a semi-annual cycle those spikes arrive twice and can overlap where states stagger cohorts on a rolling basis.

The Plan Bearing the Risk Cannot See the Process

The Act prohibits managed care organisations from making work requirement compliance determinations. States hold sole authority, and they are not required to share compliance status with MCOs prospectively.

So the entity carrying the financial consequence of the enrollment change has no visibility into what drives it. An MCO knows which members are using services, which correlates loosely with engagement and not at all with whether someone logged 80 hours or filed for an exemption. Disenrollment can only be modelled as a population-level distribution rather than a member-level prediction, which is a coarser instrument with wider intervals around it.

The contrast with Medicare Advantage is exact and unflattering. CMS gives MA plans beneficiary-level risk scores and eligibility data, so MA actuaries build member-level projections. Medicaid MCO actuaries price a population whose composition is set by an administrative process they can neither observe nor influence.

What fills that gap is state administrative capacity, which is not an actuarial variable. States with ex parte renewal, integrated eligibility platforms and wage and tax data matching verify without enrollee action; Oregon and North Carolina held disenrollment well below the national average through the unwinding. States running paper renewals and manual verification produced the high procedural rates, and they will again.

The $200 million federal allocation for FY 2026 is what states get to close that difference while doubling redetermination frequency, standing up work requirement verification, processing exemptions, and handling appeals from people wrongly disenrolled. A state that underfunds the build processes renewals more slowly and makes more errors, and every one of those errors reaches the MCO as an unplanned member loss with the same selection signature as a genuine one. The rate certification has to carry a risk whose largest single driver is a state budget line.

Further Reading on actuary.info