Washington Purges 760,000 Obamacare Enrollees and Calls It Fraud

On Tuesday, September 22, 2026, officials at the Centers for Medicare and Medicaid Services stood at a podium and announced that 760,000 people had been stripped from Affordable Care Act marketplace coverage on suspicion of fraudulent enrollment. Another 420,000 are under active investigation. New agents and brokers face a six-month moratorium on selling marketplace policies, a freeze that runs straight through the open enrollment window beginning November 1. The Guardian reported the details on Friday. The administration calls it cleanup. Critics call it a purge conducted six weeks before the November 3 midterms. Both descriptions can be true at once, and the more interesting question is why a program this large produced this much rot in the first place.

The Numbers Behind the Purge

Start with scale. ACA marketplace enrollment reached roughly 24 million people in 2025, more than double the figure from 2020. That growth was not organic demand discovering a good product. It tracked almost exactly with the enhanced premium tax credits created by the American Rescue Plan in 2021 and extended through the Inflation Reduction Act, which pushed subsidies so high that millions of applicants qualified for plans with zero monthly premium. Federal spending on marketplace subsidies ran north of $100 billion a year at the peak.

Those enhanced credits expired on December 31, 2025. Congress did not renew them. The Congressional Budget Office had projected that letting them lapse would leave roughly 3.8 million more Americans uninsured over the following decade. Insurers responded the way anyone would expect. The Kaiser Family Foundation found that carriers requested a median premium increase near 18 percent for 2026 plan year rates, the steepest single-year jump since the exchanges opened in 2014.

Now layer this week's action on top. The 760,000 removals and 420,000 investigations together touch roughly 1.18 million enrollees, close to 5 percent of the marketplace. CMS has not published a state-by-state breakdown. Based on where zero-premium plans concentrated, the bulk almost certainly sits in Florida, Texas, and Georgia, the three largest states that declined Medicaid expansion and therefore pushed their low-income populations onto the exchanges instead.

How Phantom Enrollment Was Built

The fraud is real, and it was predictable. When a subsidy covers 100 percent of a premium, the enrollee has no financial stake in whether the policy exists, whether it is any good, or whether they signed up for it at all. The broker, meanwhile, collects a commission per head. That is not a market. That is a bounty system with federal money as the prize.

CMS itself logged roughly 274,000 complaints in the first half of 2024 from people whose plans had been switched without consent, or who discovered they had been enrolled by an agent they never contacted. The agency suspended more than 200 brokers and, in mid-2024, restricted access to the federal enrollment platform for agents not already associated with a consumer's account. The Paragon Health Institute, whose founder Brian Blase served on the first Trump administration's National Economic Council, estimated that four to six million marketplace enrollments were improper, costing taxpayers somewhere in the range of $20 billion annually. Paragon is an advocacy shop with an obvious interest in the answer. The figure was also never seriously rebutted with better data, because nobody in a position to produce better data wanted to.

The mechanism was mundane. Applicants self-attested to income between 100 and 150 percent of the federal poverty line, the band that triggers maximum subsidy. Self-attestation required no documentation in most cases. A broker with a list of names, birthdates, and Social Security numbers could enroll people who had no idea it happened, often people who already had employer coverage or Medicaid. The Treasury paid the carrier. The carrier kept the money. The phantom enrollee never filed a claim, which made them the most profitable customer an insurer could have.

The Case Against the Method

Defending the cleanup is not the same as defending how it was done, and here the administration's critics have a serious argument.

Removal is not a courtroom. CMS is acting on data-matching flags, not adjudicated findings. A flag can mean fraud. It can also mean a mismatched address, a name change after marriage, an income estimate that drifted, or a household that moved between states mid-year. The 760,000 are being terminated on suspicion. Consumer advocates point out that many will not learn they lost coverage until a pharmacy declines the card or a hospital sends a bill. The appeals process exists, but it is slow, and it is administered by the same agency that issued the termination.

The broker moratorium compounds the problem. Freezing new agents for six months, right across open enrollment, removes the main channel through which low-income and non-English-speaking applicants actually find plans. Honest brokers get punished alongside fraudulent ones, and the enrollment decline that follows will be reported as evidence that the fraud was even bigger than estimated. That is a measurement trap. You cannot distinguish a phantom enrollee who vanished from a real person who could not get through the door.

The administration's counter is straightforward. Every dollar routed to a ghost is a dollar not available to a sick person, and an agency that waits for perfect evidence before acting will wait forever while the fraud compounds. Officials also note that nothing stops a wrongly removed individual from re-enrolling with documentation. That is technically correct and practically hollow for a working-poor household with no fixed address and no reason to expect the federal government to be helpful.

Both sides are describing the same failure from opposite ends. One sees theft. The other sees people about to lose insulin. Neither is asking why a program needs a fraud sweep of this magnitude every few years to stay upright.

Subsidies as Price Signals

The Austrian objection to the ACA was never that health insurance is bad. It is that a subsidy is not a payment, it is an override of the information that payments carry.

A price is a signal compressed into a number. It tells a buyer what something costs society and tells a seller what buyers will bear. Insert a third party who pays the bill and the signal degrades. Insert a fourth party who prints the money the third party uses and the signal stops working entirely. American healthcare has all four layers. The patient does not pay. The insurer does not really pay, because the subsidy does. The subsidy comes from a Treasury running a deficit that has exceeded $1.8 trillion in recent years. The deficit is financed in a currency the central bank can create.

Look at what that produced. US health expenditure now sits near $5 trillion a year, roughly 17 percent of GDP, a share no other developed economy approaches. Since 2000, prices for medical care services have risen far faster than the general consumer price index, while prices in sectors where consumers actually pay directly, laser eye surgery, cosmetic procedures, veterinary care before it got insured, stayed flat or fell in real terms. The pattern is consistent. Where the buyer feels the price, the price behaves. Where the buyer feels nothing, the price climbs until something breaks.

Fraud is the terminal symptom of that design, not a deviation from it. When the product is free to the recipient and paid for by an entity that cannot run out of money, the only remaining constraint is the honesty of intermediaries. That is a thin reed. It always was. The 760,000 removals are the state discovering, again, that it cannot audit its way out of an incentive structure it built on purpose.

The Monetary Layer

Here is the part most coverage will skip. The reason Washington can run a program that leaks $20 billion a year to phantoms without immediate consequence is that it controls the unit of account. A household that overspends faces a creditor. A government that overspends faces a bond market it can influence and a central bank that has, since 2008, expanded its balance sheet by trillions to absorb what the market would not. The cost does not disappear. It is transmitted through the currency, and it lands on wage earners whose salaries lag the price level and savers whose dollars buy less each year. Medical inflation and monetary inflation are not separate stories. They are the same story told in different units.

This is the case for Bitcoin, and it is not a hedge. Bitcoin has a fixed supply of 21 million, of which roughly 19.95 million have been mined. The issuance rate is 3.125 coins per block and halves again in spring 2028. No administrator, no emergency session, no press conference can change that schedule. A system with a hard supply cap cannot fund a phantom enrollment scheme, because there is no mechanism to conjure the funding. Every dollar spent would have to be taken from someone visibly, by taxation, and taxation that visible does not survive contact with voters. Sound money does not make healthcare cheap. It makes the cost legible, and legible costs get argued about honestly instead of being laundered through a currency. The 1.18 million Americans caught in this sweep are collateral damage from a program that could only grow to this size because the money behind it was never scarce. That is the argument for monetary sovereignty stated in human terms, and it does not require believing anything about price charts.

What to Watch

Enrollment totals on December 15. Open enrollment for the 2027 plan year runs from November 1 through mid-December on the federal platform. Expect total marketplace enrollment to land well below the 24 million peak, likely in the 18 to 21 million range, driven by three compounding forces: the expired enhanced credits, the removals, and the broker freeze. The administration will attribute the entire drop to fraud elimination. Democrats will attribute all of it to sabotage. Neither claim will be separable from the data, which is convenient for both.

Insurer earnings in Q4 and Q1. Watch Centene, Oscar Health, and Elevance. Phantom enrollees generated premium revenue with near-zero claims, which flattered medical loss ratios. Remove them and the remaining risk pool is sicker on average. If carriers report deteriorating MLRs in the first quarter of 2027, that is proof the phantoms were subsidizing real patients, and it will feed directly into 2028 rate filings.

Litigation before Thanksgiving. State attorneys general in California, New York, and Illinois have standing to challenge the terminations on due process grounds, and the six-month broker moratorium is the softest target, because it sweeps in parties against whom no allegation exists. Expect at least one preliminary injunction motion filed before the end of October.

The subsidy fight returns in January. The expired enhanced credits will be back on the table in the next budget cycle, and this fraud sweep has just handed restoration opponents their strongest evidence. If the credits return without verification reform, the cycle repeats on a three-year lag. If they return with hard income documentation, enrollment falls further and the uninsured rate climbs toward pre-2021 levels.

The midterms, November 3. Healthcare polls as a top-three issue in every cycle where premiums rise. An 18 percent median rate increase plus a million coverage terminations six weeks out is a specific, datable event that campaigns can point at. Whether voters distinguish fraud enforcement from coverage loss is doubtful. They will feel the second one.

The deeper prediction is duller and more certain. No sweep fixes this. Remove 760,000 today and the same incentive structure regenerates the problem within three enrollment cycles, because nothing in this week's action changed the arithmetic that made phantom enrollment profitable. A system that pays intermediaries per signature with money nobody had to earn will keep producing signatures. The only durable fix is a budget constraint that binds, and the United States has spent fifty years engineering its money so that no constraint binds at all.

---

Go deeper: Inflation Is a Tax

Source: Guardian US