Opinion: Washington is purging Obamacare rolls just as premiums hit a breaking point

Four days before the federal government opened its annual shopping window for Affordable Care Act coverage, the Trump administration announced that it had quietly cut off subsidies for another 189,000 people. The Centers for Medicare & Medicaid Services said the enrollees had either died or were already covered by Medicare, Medicaid or the Children's Health Insurance Program, and that ending their marketplace subsidies would save taxpayers roughly $1.5 billion over the coming year. Health and Human Services Secretary Robert F. Kennedy Jr. called it a matter of basic stewardship: "Taxpayers should not foot the bill for duplicate coverage or subsidies for people who have died."

Stated that way, the action is hard to object to. Nobody should be collecting a federal insurance subsidy from beyond the grave, and nobody needs two taxpayer-funded health plans at once. But the October 14 announcement was not an isolated housekeeping exercise. It was the third major purge of the ACA marketplace rolls in just over a year, arriving at the worst possible moment for the people caught up in it: the same two weeks in which premiums are spiking harder than at any point in the program's history and enrollees are expected to make fast, high-stakes decisions about their coverage. The administration's fraud-fighting instincts may be sound. Its execution, so far, has not earned the benefit of the doubt it is asking for.

A pattern, not an incident

Start with the scale. In 2025, the administration restored twice-yearly data matching against Medicaid rolls and ended subsidies for about 551,000 enrollees it found were also enrolled in Medicaid. In September 2026, Vice President JD Vance, flanked by CMS Administrator Mehmet Oz, announced that a White House task force would cancel roughly 315,000 enrollments covering 760,000 people over suspected fraud, with another 419,000 enrollments flagged for further review and a six-month freeze on new insurance brokers and agents signing people up. That action alone, officials said, would save $2.2 billion. Then came the October 14 round of 189,000 people, broken down by CMS as roughly 120,000 with duplicate Medicaid coverage, 58,000 with duplicate Medicare coverage and 11,000 who had died. Oz said the combined total savings across these rounds has reached nearly $6 billion.

Each announcement has followed the same script: a large round number, a dollar figure in the billions, and a reassurance that no legitimately eligible person is losing anything. Kennedy, to his credit, has not pretended the process is flawless. At the October press conference he acknowledged that "some cases begin with an error, others reveal deliberate abuse." That is an honest admission, but it is also the crux of the problem. The administration has not published an error rate, an appeals outcome tally or an independent audit of its data-matching methodology. It has asked the public to accept, on its word alone, that nearly two million enrollment actions over fourteen months have mostly hit their intended targets.

The premium shock behind the purge

The timing compounds the credibility gap. Enhanced ACA premium tax credits, first adopted during the pandemic, expired at the end of 2025 after Congress failed to renew them. The consequences have been severe. According to a national survey of marketplace enrollees conducted by the health research organization KFF, 9 percent of 2025 enrollees had already dropped marketplace coverage entirely by the time the enhanced credits lapsed, and four in five of those who stayed said their costs were higher, with half describing the increase as "a lot higher." More than half said they had already cut back, or expected to cut back, on food and other basic household spending to keep paying premiums.

Now a second wave is approaching. Insurers' own rate filings, reviewed by KFF across 276 companies in every state and the District of Columbia, show a median proposed premium increase of 15 percent for 2027, the second straight year of double-digit requests after a finalized median increase of 20 percent for 2026. Insurers cite rising medical costs generally, but they also point specifically to a sicker, smaller risk pool left behind after the subsidy cliff pushed healthier people out of the market. Open enrollment for 2027 coverage begins November 1, which means the people receiving CMS's termination notices this month are being asked to absorb a subsidy cancellation and shop for pricier coverage within the same few weeks.

• Enhanced premium tax credits expired December 31, 2025, after Congress did not renew them.
• Insurers have filed a median 15 percent premium increase for 2027, following a finalized 20 percent increase for 2026.
• CMS says it ended subsidies for 189,000 people on October 14 and for 760,000 people in September, on top of 551,000 in 2025.

The fraud concern is not invented

None of this means the administration is wrong that fraud exists. A Government Accountability Office report released in December 2025 described a covert test in which investigators built 20 fictitious identities and applied for marketplace coverage. The federal exchange approved subsidized plans for all four applicants submitted for 2024 and kept 18 of 20 enrolled with subsidies through most of 2025, generating thousands of dollars a month in advance premium tax credits for people who did not exist. GAO also found it could not confirm that the IRS had reconciled more than $21 billion in advance credits paid out in tax year 2023. CMS itself estimates it paid nearly $124 billion in subsidies to about 19.5 million enrollees in 2024. Against numbers that size, a verification system that lets fictitious applicants sail through is a genuine vulnerability, and a final rule CMS issued in May tightening income verification, reinstating pre-enrollment checks for special enrollment periods and restricting broker marketing practices is a defensible regulatory response, whatever one thinks of the ACA politically.

The question is not whether CMS should be checking for duplicate coverage and dead enrollees. Federal regulations have long required exactly that kind of periodic data matching. The question is whether a process run largely through press conferences, with headline numbers but no public accounting of false positives, is the right vehicle for cutting off nearly two million people's subsidies in little more than a year.

Who actually answers the knock

CMS says enrollees flagged for duplicate coverage or death get 30 days to dispute the finding before their subsidy, and in many cases their plan, is canceled. That sounds procedurally fair until one considers who typically misses a 30-day government notice: people who have moved, people whose mail goes unopened because they are sick, elderly widows and widowers settling a spouse's estate, and lower-income households juggling multiple address changes. Patient advocates at the National Health Law Program have documented for years that government data-matching systems for public coverage are error-prone and that most resulting terminations turn out to be procedural rather than genuine findings of ineligibility, with large shares of people who lose coverage this way later shown to have been eligible all along, often after they become uninsured in the interim.

CMS Administrator Oz has insisted that nobody affected is actually losing coverage, since by definition anyone flagged already has other coverage or has died. That claim does not hold up as cleanly for the September round. Independent observers who examined that announcement, including reporting on the task force's methodology, noted that the government has not disclosed how enrollments were selected for cancellation, only that some lacked a valid Social Security number or immigration documentation on file, categories that can just as easily reflect a data-entry problem as deliberate fraud.

"I think there's no question that somebody who was fraudulently enrolled should have their coverage canceled," said Cynthia Cox, a vice president at KFF who directs its ACA research program, of the September action. "I think the question is whether this was the appropriate process by which to identify fraudulent enrollees," she added, noting that the cancellations occurred outside the normal federal rulemaking process.

Ellen Montz, a former CMS official who oversaw the marketplace under the previous administration, made a similar point in reaction to that announcement: whatever legitimate fraud the task force caught, the rollout lacked the kind of detail that would let outside analysts or affected consumers understand why a specific enrollment was flagged. Representative Richard Neal of Massachusetts, the top Democrat on the House Ways and Means Committee, went further, arguing that the administration's broader approach to the marketplace, including its failure to extend the enhanced subsidies in the first place, had already created "the worst healthcare crisis" enrollees have faced under the law.

What should happen now

These two arguments are not actually in tension, and resolving them does not require choosing a side in the broader fight over the ACA's future. A government serious about marketplace integrity and a government serious about not harming lawful enrollees would do the same three things. First, CMS should publish the false-positive and successful-dispute rates from each matching round, the same way it publishes savings totals, so outside researchers and Congress can judge whether the net is being cast too widely. Second, an independent body such as GAO, whose own covert testing the administration has leaned on for justification, should be asked to audit a sample of this year's terminations rather than take the agency's self-reported numbers on faith. Third, and most immediately, CMS should stop timing enrollment-purge announcements to land in the two weeks before open enrollment, when the people affected have the least slack to absorb a mistake before new, more expensive plans lock in for the coming year.

None of this requires tolerating fraud. It requires treating the people on the other end of a termination notice as more than a line item in a savings announcement. An administration confident that its methodology is sound should welcome the scrutiny; one that resists it invites the suspicion that the number being optimized is the headline, not the integrity of the program.