Marketplace enrollment fell roughly 20% as enhanced ACA subsidies lapsed — and the people leaving are the healthier ones

Summary

Marketplace sign-ups for 2026 fell to 23.1 million, down about 1 million (4.9%) from 24.3 million in 2025 — the first year-over-year decline since 2020. But sign-ups overstate actual coverage: roughly 14% of January 2026 sign-ups had not paid their first premium, and KFF/Wakely estimate that average effectuated (paid) enrollment for 2026 will land around 17.5 million, down roughly 21% (a drop of about 4.8 million people) from 22.3 million in 2025, with plausible outcomes ranging from a 17% to 26% decline; the Urban Institute separately projects 4.8 million people becoming uninsured because of the lapse. On the health-mix question: every independent source that has looked at this — KFF's enrollee survey, Wakely's HHS-HCC risk-score analysis of 80% of the individual market, insurers' own state rate filings, and Peterson-KFF's tracking of 2027 rate filings — agrees the people leaving are healthier on average than those who stay, which is pushing the remaining risk pool toward higher morbidity and is already showing up as an explicit line item in 2027 premium filings.

How much has enrollment shifted

The 2021 American Rescue Plan enhanced premium tax credits (extended through 2025 by the Inflation Reduction Act) expired at the end of 2025 with no further extension from Congress, despite a government shutdown fight over the issue and a narrow House vote (17 Republicans joined Democrats) that failed in the Senate. This was the first year since 2020 that enrollees lacked access to the enhanced subsidies, and it produced the sharpest single-year enrollment drop in the ACA Marketplaces' history.

• Plan selections (sign-ups): CMS's final snapshot put 2026 sign-ups at 23.1 million vs. 24.3 million in 2025 (a 4.9% decline); an earlier CMS snapshot in mid-January (before open enrollment fully closed) already showed about 22.8 million signed up, down 1.5 million from the same point a year earlier.
• Effectuated (paid) enrollment is the more meaningful figure because it reflects who actually keeps coverage. Wakely's analysis of proprietary payment data covering ~80% of the individual market (75+ issuers, 30+ states) found only about 86% of January 2026 enrollees had paid their first premium (state range: 82nd to 94th percentile), and net attrition will likely continue through the year via non-renewal, grace-period lapses, and disenrollments that predate January entirely (and so never show up in sign-up counts at all). Wakely's central estimate is a 17%-26% decline in average 2026 individual-market enrollment vs. 2025; applying that range to the ACA Marketplace specifically implies effectuated enrollment of roughly 16.5-18.5 million, centered near 17.5 million — a drop of 3.8 to 5.8 million people from 2025's 22.3 million. This is close to CBO's own February 2026 projection of ~16.9 million average 2026 Marketplace enrollment (a ~25% contraction), and to the Urban Institute's separate estimate that 4.8 million people will become uninsured in 2026 because of the lapse (with another roughly 2.5 million dropping ACA coverage but finding other insurance, e.g. employer or Medicaid).
• Who left, by income: the decline is heavily concentrated just above the "subsidy cliff" (400% of the federal poverty line, where premium tax credit eligibility ends entirely under un-enhanced rules). Consumers between 400-500% FPL were only 3% of 2025 enrollment but accounted for 27% of the sign-up decline (a 44% drop in that group alone, over 321,000 people); everyone above 400% FPL was just 7% of 2025 enrollment but 48% of the total decline. Lower-income enrollees (under 150% FPL, the largest single income group) still accounted for 37% of the decline in absolute terms, even though their exit rate was much lower (about 4%), simply because that group is so large.
• Who left, by age: adults 18-34 accounted for 46% of the total decline in sign-ups (down 542,000, or 8%, from 6.7 million to 6.2 million) — more than any other age group.
• Plan-tier shifts among those who stayed: silver-plan share fell from 57% to 43% of sign-ups (a record low, and the first time under half chose silver), while bronze rose from 30% to 40% and gold from 13% to 17%, as enrollees who kept coverage bought down to cheaper, higher-deductible plans. The average Marketplace deductible rose 37% ($1,027) to a record $3,786.
• Context: total ACA Marketplace enrollment had more than doubled since 2020 (from roughly 11-12 million to 24.3 million in 2025) once the enhanced subsidies took effect, with 88% of that growth (11.4 of 12.9 million new enrollees) concentrated in states that voted for Trump in 2024 — enrollment there grew 157% on average versus 36% in states that voted for Harris. The 2026 pullback is unwinding a large share of that growth in the same states that gained the most from it.

Is the group leaving healthier or sicker than the group staying

Healthier. This is the consistent finding across four independent kinds of evidence:

• Insurer risk-score data (Wakely Consulting Group, April 2026): using data from 75+ issuers across 30+ states (~80% of the ACA-compliant individual market), Wakely compared normalized HHS-HCC risk scores — the government's own actuarial measure of how sick a population is — for enrollees who paid their January 2026 premium ("Stayers, paid") against enrollees who had not yet paid ("Stayers, unpaid," i.e., on track to lapse). The paid (staying) group's aggregate risk score was 10.2% higher than the unpaid (leaving) group's (a 6.2% gap on a state-by-state simple-average basis) — direct, individual-level evidence that people exiting the market are healthier than people who keep paying. Wakely's separate market-wide model, which also has to account for new/switching enrollees ("Joiners") whose risk scores aren't yet observable, estimates that this selection effect will raise overall market morbidity by 2.5% to 6.0% in 2026 relative to 2025 (their headline range, folding in additional model uncertainty, is 2.9% to 6.5%). The model's own working assumption is that new entrants are 5%-20% sicker than the leaver population, built explicitly on the premise that people willing to pay much higher premiums to newly join or stay are sicker than those forgoing coverage.
• Age-based selection (KFF, CMS OEP data): the group with the largest sign-up decline — young adults 18-34 — is also the demographic insurers' own rate filings describe as "typically healthier." KFF notes this pattern is exactly what insurers projected in their 2026 rate filings.
• Insurer rate filings, stated directly: Multiple insurers said this outright in their state regulatory filings. Maine Community Health Options wrote that "healthier individual consumers are more likely to lapse coverage." UnitedHealthcare's Maryland filing stated that "healthier members are expected to leave at a disproportionately higher rate than those with significant healthcare needs, increasing market morbidity in 2026." Antidote Health, filing in Texas for 2027, built in an explicit "morbidity adjustment of 6.0%" to account for the same shift.
• Measured effect on pricing, confirmed in two consecutive years of rate filings (Peterson-KFF Health System Tracker, analyzing 77 insurers in 16 states plus DC): this adverse-selection dynamic is estimated to have added about 4 percentage points to 2026 premium increases, and insurers are layering on roughly another 4 percentage points for 2027 (median proposed 2027 increase: 14%, after a finalized 20% increase for 2026) specifically because the pool that remained after the 2026 exodus is sicker than the 2025 pool was.

One caveat from Wakely's own methodology: they cannot fully separate people who switched insurers ("Joiners," who remain in the market) from people who dropped ACA coverage entirely, among their "Leaver" cohort, so their market-wide morbidity-shift range carries real model uncertainty. But the direct Stayer paid-vs-unpaid risk-score comparison sidesteps that ambiguity and is the cleanest available evidence: within the same issuer and the same year, the people not paying (and therefore losing coverage) carry meaningfully lower risk scores than those who pay to keep it — and every insurer statement, rate filing, and independent analyst reviewed here points the same direction.

Data and methodology notes

The AskAmerica/askengine connector's own tables (census.acs_health_insurance, census.sahie_insurance, and the health schema) do not carry ACA Marketplace enrollment, plan-selection, or risk-adjustment data at all — a confirmed catalog gap (search_catalog returned no match on "marketplace" or "exchange" against any table or column in the corpus). All figures in this report are therefore drawn directly from primary and near-primary sources fetched and read in full: CMS's own Open Enrollment Period reports and public-use files (via KFF's and CNBC's direct citations of the CMS releases), KFF's published analysis of those CMS files plus its own survey of 1,117 2025 enrollees (fielded Feb 12-Mar 2, 2026), Wakely Consulting Group's April 2026 report built on the Wakely National Risk Adjustment Reporting (WNRAR) project's issuer-submitted risk-score and payment data, and Peterson-KFF Health System Tracker's analysis of 2027 insurer rate filings. Effectuated 2026 enrollment is still an estimate as of this writing — CMS's own Effectuated Enrollment: Early Snapshot is expected around July 2026 and the Full Year report (net of grace-period retroactive terminations) not until summer 2027 — so the 17.5 million figure and the morbidity-shift estimates above are the best available projections, not yet a finalized government tally.