Who Catches the People Insurers Let Go

On Tuesday, October 27, two health care companies reported their summers a few hours apart. Neither mentioned the other. They were describing the same event from opposite ends.

From St. Louis, Centene, which sells health plans on the Affordable Care Act marketplaces, said its quarter was "better than expectations." Its chief executive, Sarah London, praised her teams for "delivering increased margin."

From Nashville, HCA Healthcare, the hospital company, reported something else. Its hospitals had seen "an increase in uninsured volume, primarily due to patients who lost coverage on the health insurance exchanges." HCA estimated that the shift cut its pre-tax income by about $400 million, in one quarter. Its chief executive, Sam Hazen, spoke of "the challenges brought about by federal policy changes this year."

One company let people go. The other company could not. That difference, between a business with an exit and a business without one, decides who pays when a subsidy ends.

The enhanced premium tax credits, the extra federal help that had made marketplace plans cheaper, expired at the end of 2025. People did the arithmetic. By February 2026, KFF counts, effectuated enrollment, the people who actually paid their premiums, was "down 12% (or 2.6 million people)" from a year earlier, to 19.2 million.

The insurers saw it coming and decided what kind of business they wanted to be. Centene's quarterly report explains its reasoning in one long sentence. The expiry, together with a federal rule, "led to a higher projected baseline of Marketplace morbidity." In plain words: the healthy would leave first and the sick would stay. So "we took corrective pricing actions for 2026 covering 95% of Marketplace membership."

Prices rose, and members left; Centene says the repricing contributed to the decline. It had 3,403,300 marketplace members on September 30, against 5,828,100 a year earlier, about 2.4 million fewer. Its commercial gross margin for the first nine months rose $659 million, which Centene attributes mainly to rate increases and a risk-adjustment settlement, even as the smaller book cut the third quarter's commercial gross margin by $158 million.

Molina was blunter. Its marketplace membership fell to 267,000, "a decrease of 446,000 members," and Molina called that "in line with our product and pricing strategy towards restoring our target margins." In the third quarter its marketplace business spent 92.5% of premiums on medical care, down from 95.6% a year earlier. And it is not finished. For 2027 it plans cuts it estimates at "a $1 billion decrease in Marketplace premiums."

None of this is a scandal. It is what a business does when it can choose its customers. An insurer prices for the people it expects to cover, and when the pool turns sicker, it raises the price or leaves. Some are leaving outright. KFF counts 11 carriers that have announced plans to exit at least one state for 2027, and Cigna "is leaving the individual market in all 11 states where it currently participates."

The people priced out did not stop getting sick.

Suppose you are one of them: a self-employed electrician whose premium jumped in January, who did the sums and let the policy lapse. Nothing about your heart or your appendix changed when you made that call. All that changed is who gets the bill when one of them fails. In the old arrangement it went to an insurer, which had priced for it. Now it goes to whichever hospital is nearest when the pain starts.

A hospital has no pricing action for the emergency room. In 1986 Congress passed the Emergency Medical Treatment and Labor Act, in the words of the federal agency that oversees it, "to ensure public access to emergency services regardless of ability to pay." A hospital that takes Medicare and runs an emergency department must examine whoever comes in, and "Hospitals are then required to provide stabilizing treatment." The insurer can say no. The emergency room cannot.

So the hospitals count the people instead. At HCA, same-facility uninsured admissions rose 15.5% in the first quarter of 2026 from a year earlier, 23.4% in the second and 27.4% in the third. Each quarter worse than the last. The year before shows how unusual that is. In the third quarter of 2025 the same measure fell 2.0%, and in the fourth it rose 7.1%. Uninsured patients are now 8% of HCA's admissions, against 6%. Texas and Florida alone account for 73% of the quarter's uninsured admissions.

HCA names more than the subsidy. The rise reflects "continued impacts from the expiration of the EPTCs at the end of 2025 and administrative reforms, as well as a decline in Medicaid conversions." A Medicaid conversion is the industry's term for signing an uninsured patient up for Medicaid, so fewer of them means more patients who stay uninsured. The subsidy is not the whole story at HCA. It is the cause the company names first.

Tenet Healthcare's numbers have the same shape. Its charity and uninsured admissions were 6,795 in the third quarter, against 5,394 a year earlier, up 26.0%. Its paying admissions, everyone else in the same hospitals, rose 2.2%. Tenet's own account: "we have experienced a decrease in admissions among patients with coverage through health insurance exchanges, resulting in an unfavorable shift in our payer mix." At Community Health Systems, self-pay revenue went from 0.8% of the total to 1.7%.

Admissions are not dollars, but here the filings give dollars too. HCA estimates its cost of uncompensated care, the charity care, uninsured discounts and bills it never expects to collect, at $1.477 billion for the quarter, against $1.188 billion a year earlier. That is about $289 million more in three months. Tenet puts its cost of caring for uninsured and charity patients at $181 million, against $144 million.

There is a fair case that the hospitals are only giving back a gift. The subsidy years were good to them. Tenet says the credits, along with longer enrollment windows and wider eligibility, "led to higher enrollment numbers," which meant more patients arriving with a card. HCA's own figures show how calm the uninsured line was in 2025. The expiry returned the hospitals to a bill they used to carry.

That is true, and it is the point. Both sides lost the same money. Only one could decide how much of the loss to keep. Centene repriced 95% of its marketplace book and accepted the members it lost. Molina shrank by design. HCA's emergency rooms could not reprice a heart attack. The cost of the policy did not vanish when the credits expired. It moved to the part of the system that had no exit.

The pattern is not peculiar to health care. It shows up wherever one party in a chain has a choice and another has a duty. A utility with an obligation to serve keeps the customers a competitor would never sign. A public school takes every child in its district, whatever the private school across town decides. A lender of last resort lends when everyone else has stopped. In each case the party with the duty ends up holding what the others declined, and its costs move with decisions it did not make.

When a subsidy ends, the customer does not disappear. Ask who in the chain can say no. The one that can exit reprices. The one that cannot inherits the bill.

Open enrollment for 2027 began on November 1. KFF finds that benchmark silver premiums are up 13% on average, and Molina has already said it will shrink again. So when the next quarter's reports arrive, the insurers' membership tables will tell you who chose to leave. The hospitals' uninsured admissions will tell you where those people went.