Judge Richard Stearns of the federal district court in Massachusetts declined on July 30 to pause the Medicaid community engagement rule, rejecting a preliminary injunction sought by 25 states and the District of Columbia, and the rule took legal effect the next morning (Healthcare Dive, July 2026).

CMS projects the requirement will remove 2.3 million people from Medicaid in fiscal 2027. A measurable share of them will land, sicker than average, on self-insured employer plans whose 2027 assumptions were set before any of this existed.

Key Takeaways

  • 2.3 million fewer Medicaid enrollees in fiscal 2027 on CMS's own projection, rising above 3 million in later years, with a share of them arriving on self-insured employer plans mid-year.
  • 80 hours per month of qualifying activity for non-pregnant adults aged 19 through 64 in the expansion group, with every expansion state and the District of Columbia compliant no later than January 1, 2027.
  • $182 to $485 of added annual cost per person across 179 million people with employment-based coverage is the uncompensated-care pass-through estimate. The width of that range is the actuarial problem.
  • 5.3 million more uninsured by 2034 on CBO's arithmetic, the largest single Medicaid item in the law, against a $325 billion reduction in federal Medicaid spending over 2025 through 2034.
  • 9.9% medical trend for 2027, a 15-year high, is the baseline this lands on top of rather than a number already containing it. Segal's respondents answered before the rule survived its first court test.

What Became Enforceable on July 31

The interim final rule, CMS-2454-IFC, implements the community engagement requirement Congress wrote into last summer's budget reconciliation law. Non-pregnant adults aged 19 through 64 in the Medicaid expansion group must document 80 hours per month of qualifying activity, employment, education, work programs or community service, or meet equivalent income thresholds, as a condition of eligibility.

The rule exempts pregnant and postpartum women, medically frail individuals and caregivers, and lets states request a temporary good faith effort exemption when their verification systems are not operationally ready (Foley Hoag, June 2026). Every state that expanded Medicaid, plus the District of Columbia, has to comply no later than January 1, 2027.

The timeline was compressed. CMS issued the rule June 1, published it in the Federal Register June 3, and set both the effective date and the close of the comment period on the same day, July 31 (CMS fact sheet, June 2026). The agency's own projection is 2.3 million fewer enrollees in fiscal 2027, rising above 3 million later.

The Congressional Budget Office runs the longer arithmetic higher: 5.3 million more uninsured by 2034 from the work requirement alone, carrying a $325 billion reduction in federal Medicaid spending over 2025 through 2034 (CBO, via Georgetown CCF, August 2025). Judge Stearns was not persuaded the states faced irreparable harm before a merits decision, so those numbers no longer turn on litigation for the 2027 plan year.

New Lives, Deferred Care, and the Completion Factors

The first channel runs through the plan's own enrollment file. Most adults subject to the requirement already work; the projected coverage losses come substantially from documentation and reporting failures rather than refusal to work. A disenrolled worker whose employer offers coverage has a qualifying life event and a strong incentive to take the employer plan.

That migration reprices frequency and severity together, on a cohort whose experience does not yet exist in any credibility database the plan's actuary can query. Medicaid-eligible populations carry higher chronic disease and behavioral health burdens than the average enrolled employee, and coverage that churned leaves those diagnoses unmanaged.

The reserving effect is more specific than a general loading. Populations entering coverage after a gap front-load utilization: deferred imaging, delayed procedures, prescriptions restarted, specialist backlogs cleared in the first months. Early development periods of the 2027 incurral year will run steeper than the completion factors fitted to 2024 through 2026 experience, so an IBNR estimate applying historical completion patterns to the new mix understates the liability exactly as the enrollment file says exposure is growing.

That understatement surfaces late. The first valuation after the cohort arrives still leans on thin, immature paid data, so the miss shows up as adverse development in the second and third valuation cycles, after the reserve has been booked and reported. It is the receiving side of the same displaced morbidity actuary.info traced through the managed care capitation reset and the ACA morbidity load now appearing in 2027 rate filings.

Nor does the cohort arrive as a January 1 block. States will process disenrollments through existing redetermination cycles, each one opening a 60-day special enrollment window, and many affected workers then face a plan waiting period. Exposure accretes month by month from February onward. Specific and aggregate stop-loss attachments quoted now off 2026 census files inherit that shape: adding even three to five percent more lives mid-year, concentrated above the base rate, erodes the aggregate corridor from both directions at once.

Which Channel the Cost Arrives Through Is Not Yet Knowable

Not every disenrolled worker lands on an employer plan. CBO's 5.3 million figure is, by definition, people who end up with nothing, and their care becomes uncompensated. Hospitals recover it the way they always have, through the commercial rates they negotiate with payers that can pay. The AFL-CIO put that pass-through at $182 to $485 of added annual cost per person across the 179 million people with employment-based coverage (AFL-CIO, 2025).

The width of that range is the constraint. Nobody pricing a 2027 self-insured renewal can currently distinguish a 40-basis-point trend add from one three times that size, because the split between the two channels depends on employer offer rates, state implementation behavior and take-up decisions that have not happened yet.

The channels also run on different clocks. Channel one hits claims as soon as the new lives enroll and start clearing deferred care, inside the 2027 plan year. Channel two moves at the speed of hospital contracting: uncompensated care absorbed in 2027 is recovered through rate negotiations on one-to-three-year cycles, so it lands mostly in 2028 and 2029 renewals. The 9.9% trend Segal published for 2027 contains neither.

State choices then spread the exposure unevenly. How aggressively a state verifies, how it operationalizes the medically frail exemption, whether it seeks the good faith effort exemption, and how work-rule checks interact with its existing redetermination cadence all decide whether disenrollment arrives as a January cliff or rolling attrition. A sponsor with employees in a dozen expansion states holds a dozen exposure profiles under one set of plan assumptions, and a single national enrollment-growth assumption is the wrong shape for that even if its level turns out right.

The sponsor's own contribution design makes the final split. Low-wage workers weigh the employee premium contribution against going uninsured, and that contribution has historically suppressed take-up even where coverage is offered. Rich employer contributions pull more of the displaced population into channel one and buy the claims exposure directly; an employee share that prices those workers out pushes them into channel two, where the cost returns through hospital rates without ever appearing on the enrollment file.

Further Reading

Sources