The affordability story is well documented. Enhanced ACA subsidies expired at the end of 2025, out-of-pocket premiums jumped sharply, and enrollment fell from a peak above 22 million to about 19 million as of February 2026. For 2027, insurers are proposing a median increase of about 14%.
Less examined is a separate pathway by which people lose coverage, one that operates independently of price. It is worth understanding because it can be prevented, and because it catches people who are doing everything right.
The verification trap
Under a newer verification rule, a person's premium tax credit can be suspended when the details on their application don't match a government database. Fixing the mismatch can take months, and during that time the enrollee owes the full premium.
Consider what that means in practice. Someone whose income has not changed, whose eligibility is genuine, and who has done nothing wrong can have their subsidy suspended because a name is spelled differently in two federal systems, or an address is stale, or an employer's reported income lags reality. The subsidy stops. The bill arrives at full price, which for a subsidized enrollee can be several times what they were paying. And the resolution process takes months.
For most households on marketplace coverage, paying an unsubsidized premium for even two or three months is not feasible. So the coverage lapses, not because the person was ineligible, but because the system could not confirm they were eligible fast enough.
This is the same failure mode now appearing in Medicaid, where work requirements starting in January will disenroll people who are working but cannot document it within a 30-day window. In both cases the binding constraint is not eligibility. It is administrative throughput.
The practical implication is specific: if your marketplace application data is stale, a former address, a name that does not exactly match your Social Security record, an income estimate that has drifted, it is worth correcting before open enrollment rather than after a mismatch triggers a suspension.
Why the premium increases are not only about subsidies
For anyone deciding whether to keep coverage, it matters that the cost pressure has more than one source, because that determines whether waiting for a policy fix is a plausible strategy.
The main factor driving proposed 2027 increases, as in most years, is the rising cost and use of medical care, with growing demand for costly specialty medications and GLP-1 weight loss drugs. That portion of the increase would be happening regardless of what Congress does about subsidies.
The subsidy-related portion is real but smaller than the headline suggests. Insurers attribute about 4 percentage points of the proposed increases to lasting effects of the expiration of enhanced subsidies, through the risk-pool deterioration that follows when healthier people leave. UnitedHealthcare told New York regulators that the subsidy expiration together with new federal rules account for 12.7% of its requested rate change.
The disputed question about who left
There is a genuine disagreement about why enrollment fell, and it is worth stating both positions accurately rather than picking one.
KFF's Cynthia Cox and other policy analysts attribute the decline to cost, with a specific mechanism: it's likely that the people who dropped their coverage were also the healthier people, because sicker people were probably going to try to make it work however they could. The administration's position is different, framing the decline as improper enrollments being removed. These are testable claims that will be resolved over time by data on who left and what happened to them.
What actually reduces your cost
Two practical points from analysts are more useful than the political argument.
First, shopping is not optional this year. Enrollees may have to shop around when enrollment opens for 2027 coverage in October, according to Brookings senior fellow Matthew Fiedler. The credit is pegged to a benchmark plan, so if your plan's premium rises faster than the benchmark, your out-of-pocket cost rises even though your subsidy did not fall. Auto-renewal is the expensive default.
Second, the rate requests are proposals, not final. The 1% to 52% range across filings means state and plan variation is enormous, so a national median tells you little about your own renewal.
The decision people are actually facing
For someone genuinely priced out, the choice is between an unaffordable premium and the financial risk of going uninsured. CMS has relaxed what qualifies as an ACA-approved plan, which expands cheaper options and also expands the possibility of buying something that will not cover what you need.
Dropping coverage is also not easily reversible. Enrollment is generally limited to open enrollment or a qualifying life event, so someone who cancels in March and is diagnosed with something in June typically cannot buy coverage until the following January.
The gap between "I cannot pay this specific plan's premium" and "I must be uninsured" is often wider than it appears in October, and the people who close that gap are usually the ones who shopped rather than the ones who defaulted.