In an analysis published on 30 September, the Kaiser Family Foundation put one case this way: “A 60-year-old couple making $85,000 (or 402% FPL) would see yearly premium payments rise by over $22,600 in 2026.”
That sentence was written as a projection about a law that had not yet lapsed. The enhanced premium tax credits expired on 1 January, and the sentence is now a description of a bill that people are opening.
Before the ledger, the mechanism, because the size of that figure is not an accident or an error and it follows directly from how the subsidy is built.
How the credit is calculated
The premium tax credit does not work by paying a share of a premium. It works backward from income. The law names a benchmark plan, the second cheapest silver plan available to you, and then says that you should not have to pay more than a set percentage of your household income to buy it. The credit is whatever covers the gap between that percentage and the actual price.
Two consequences follow. The first is that the credit is larger where premiums are higher, because the gap is larger, and premiums are higher for older people: the Act permits insurers to charge an older adult up to three times what they charge a younger one. The second is that the credit is larger where income is lower, because the percentage applied is lower.
The enhanced version, in force from 2021 through 2025, changed two things. It reduced the applicable percentages across the board, to nothing at the bottom of the scale. And it removed the eligibility ceiling, so that households above four hundred percent of the federal poverty level, who had previously received no credit at all, became eligible for one capped at 8.5 percent of income.
Expiry restores both. The percentages go back up, and the ceiling comes back.
Why it is a cliff and not a slope
Eligibility above the line is not tapered. It is binary. A household at 399 percent of the poverty level receives a credit calculated on the formula above. A household at 401 percent receives nothing, and pays the unsubsidized price of the plan.
Put that together with age rating and the couple in the quotation stops being surprising. They are sixty years old, so their benchmark premium is near the top of the permitted band. They are two people rather than one, so it is charged twice. And at 402 percent of the poverty line they are two percentage points past the edge, which converts a credit worth most of the premium into no credit at all.
The $22,600 is not what their insurance costs. It is the part of what it costs that used to be paid by someone else and now is not.
The ledger
What it costs. Across subsidized enrollees, KFF put the average annual premium payment at $888 in 2025 and projected $1,904 for 2026, a rise of 114 percent. On the same analysis, extension would have saved the average subsidized enrollee $1,016 over the year. Those are averages over a population of more than twenty million people and they conceal the whole distribution, which is where the interesting part is.
Who pays. Not uniformly. A single adult earning $28,000, comfortably inside the old eligibility range, was projected to pay $1,238 more over the year: a real sum, absorbed with difficulty, at a price still below the unsubsidized one. The couple at 402 percent pays eighteen times that increase. The burden is not distributed along a gradient that tracks ability to pay. It concentrates on people near the line and on people who are old, and those two groups overlap.
Who collects. This is the part most often got wrong. Insurers do not gain. The credit was already paid to them on the enrollee’s behalf, so the gross premium they receive for a continuing customer is unchanged; only the payer has changed. What insurers lose is customers who stop buying. Separately, they filed 2026 rates proposing a median increase of about 18 percent, which reflects underlying cost growth and an expectation about who would remain.
Who saves. The federal government, which stops making the payments. That saving is real and it is also gross rather than net, because some people who leave the individual market do not become healthy and uninsured in a way that costs nothing. Some arrive later, sicker, at an emergency department, and that cost reappears as uncompensated care carried by hospitals and eventually by other payers. How much of the saving survives that is genuinely contested and I am not going to pretend the range is narrow.
What the numbers are and are not
Everything above except the rate filings is a projection, produced before the fact by people modeling a counterfactual. Projections of this kind have a decent record on direction and a patchier one on magnitude, and the honest position in January is that the direction is not in doubt and the magnitude is not yet measured.
The Congressional Budget Office had separately estimated a marketplace contraction of roughly a quarter following expiry. Whether that is close will be known from enrollment data, not from any model.
One more caution about the figure itself. A percentage increase is a ratio, and a ratio with a small denominator moves violently for reasons that have nothing to do with the size of the change. A premium going from twenty dollars to a hundred is a four hundred percent rise and an eighty dollar problem. The same percentage applied to a larger starting premium is a different problem entirely, and the two are not distinguishable from the headline.
The part that will take years
There is a second order effect that nobody can score properly, and it is the reason this is not simply a one year transfer.
When price rises sharply, the people who leave first are the ones who were least sure they needed the product. In insurance that means the healthier end of the pool, because a person managing a chronic condition will find the money and a person who has not seen a doctor in three years will decide to risk it. Each departure of that kind raises the average cost of everyone remaining, which raises next year’s premiums, which prompts the next marginal departure.
Whether that becomes a slow drift or something faster depends on the elasticity of a population that has never faced these prices, and the elasticity is not known. It is being measured now, involuntarily, by everyone in it.
Open enrollment on the federal marketplace closes tomorrow. After 15 January a person who let the deadline pass cannot buy a plan at any price until next winter, absent a qualifying life event. Which means the single most important figure in this ledger, the count of people who looked at the new number and decided not to pay it, does not exist yet, and by the time it does the decision will be unalterable for a year.
