How Marketplace Subsidies Work: The FPL Band, the Benchmark Plan, and Tax Time

A premium tax credit is a federal subsidy that lowers what you pay for a Marketplace plan, calculated from your projected household income against the federal poverty level, and for plan year 2026 it runs from 100% to 400% of that level with a hard cliff at the top¹. The enhanced credits that softened those edges expired at the end of 2025 and are not in force for 2026, so the pre-2021 rules apply again² (checked 30 July 2026).

The first working week of January is when I learn what the year will be like. The phone starts, and it is never a policy question. It is somebody reading me an invoice: last year’s number, this year’s number, and a pause while they wait for me to say the higher one is a mistake. In January 2026 I had that call more times than in any January before it, and the honest answer was that neither the plan nor the household had changed. What changed was the credit behind the plan. That call is why this page dates every figure and names the plan year it is talking about.

Nothing here is an eligibility determination. Only the Marketplace can decide whether you qualify for a credit and how large it is, and only HHSC can decide a Medicaid or CHIP case. This is the published mechanism, its source, and the date it was checked.

Who can get a premium tax credit in Texas?

A household with projected income from 100% to 400% of the federal poverty level, that is not eligible for other qualifying coverage, and that files a federal tax return³ (checked 30 July 2026). Both ends of that band do real work in Texas, and the bottom end does more damage than the top.

Below 100% of the poverty level, HealthCare.gov’s wording is that “you probably won’t qualify for savings on a Marketplace plan, but you may qualify for Medicaid”³. In a state that expanded Medicaid, that points somewhere. In Texas it frequently points nowhere: Texas is one of ten states that have not adopted the ACA Medicaid expansion, while 41 states including the District of Columbia have⁴ (checked 30 July 2026). The result is about 605,000 poor uninsured Texan adults aged 19 to 64 in the coverage gap, roughly half the national total of about 1.2 million across the ten non-expansion states⁵ (KFF analysis published 27 July 2026, checked 30 July 2026).

That is the most misread situation on the Texas Marketplace. A subsidy result of zero at a low income is not a mistake on the form, and reapplying with the same income will not change it: see the Texas coverage gap, then getting care without insurance in Texas. Before concluding you are in it, check the categories that do exist, in Texas Medicaid eligibility and CHIP and Children’s Medicaid in Texas, because a household is not one eligibility case.

What changed for plan year 2026?

The enhanced premium tax credits expired at the end of 2025, they are not in force for plan year 2026, and the 400% FPL subsidy cliff is back. The enhancement came from the American Rescue Plan, was extended through 2025 by the Inflation Reduction Act, and removed the cliff while cutting the required contribution below it. It is the most volatile fact on this site.

The verified consequences, all from KFF’s analysis of 2026 Marketplace enrollment² and checked on 30 July 2026:

• Premium payments net of tax credits rose by 58% on average for people who signed up for 2026 coverage, which KFF elsewhere puts at about $780 more than the previous year. KFF names the cause as the return of the subsidy cliff as the enhanced tax credits expired.
• National enrollment fell for the first time in seven years.
• In Texas, sign-ups rose by about 206,000, a 5% increase, while effectuated enrollment fell by about 146,000, a 4% decrease. These two numbers are quoted together or not at all. The rise in sign-ups on its own reads as good news. The fall in effectuated enrollment is the one that reflects how many people could actually keep the plan once the first premium came due.

On plan year 2027, this site says only what it can verify. The enhanced credits expired at the end of 2025. No enacted extension or restoration had been found when this page was last checked on 30 July 2026, so the position for 2027 was still open at that date. That is an absence of evidence rather than a confirmed absence of legislation, and you will not read here that the credits will not return. The structure is re-checked in September, before any copy about the next open enrollment is written.

What is the benchmark plan?

The benchmark plan is the second lowest cost Silver plan available to your household in your area, and the size of your credit is calculated against its premium rather than against the plan you buy⁶ (checked 30 July 2026). Once the amount is set, you can apply it to a plan in any metal category⁷.

That single design decision explains most of what confuses people about subsidies.

• The credit is a dollar amount, not a percentage discount. Applied to a Bronze plan it can cover most of the premium; applied to a Gold plan it covers a smaller share of a larger number.
• It is local. The benchmark is drawn from the plans offered in your county, so your credit can change between plan years because the plan mix around you changed, even if your income did not move at all. That is the answer to the January call more often than anything the household did.
• It only exists inside the band. The whole calculation runs from 100% to 400% of the federal poverty level for 2026³ (checked 30 July 2026); above 400% the benchmark stops mattering, because there is no credit to calculate.
• Silver has a second reason to matter. Cost sharing reductions attach to Silver plans only, and they lower the deductible, copays, coinsurance and the out-of-pocket maximum rather than the premium⁸. A qualifying household that buys Bronze for the cheaper premium can spend more across the year than it would have on Silver.

Underneath the benchmark sits a table of required contribution percentages: the share of income a household is expected to put toward the benchmark premium before the credit fills the rest. This site publishes no figure from that table. It is indexed, and it changed materially when the enhancement expired, so a percentage carried over from an earlier plan year would be wrong in a way that reads as authoritative. The current version belongs to HealthCare.gov and to the application. What is here instead is the mechanism, which does not move: benchmark first, expected contribution second, credit is the difference. How the plans are built is in Marketplace plans in Texas, and the choice between them in picking a Marketplace plan.

Which poverty guidelines apply, and when?

Marketplace eligibility for a plan year uses the prior year’s poverty guidelines; Medicaid and CHIP use the current year’s. This lag rule is small, unglamorous and the source of a lot of avoidable confusion, because it means one household income can be measured against two different tables in the same month by two different offices.

The 2026 annual guidelines for the 48 contiguous states, with 2025 alongside for comparison³ (checked 30 July 2026):

Household size	2026	2025
1	$15,960	$15,650
2	$21,640	$21,150
3	$27,320	$26,650
4	$33,000	$32,150
5	$38,680	$37,650
6	$44,360	$43,150
7	$50,040	$48,650
8	$55,720	$54,150
each additional	add $5,680	add $5,500

On the Texas side, HHSC’s eligibility chart in the Texas Works Handbook is computed on the 2026 guidelines, effective 1 April 2026⁹ (checked 30 July 2026). A household applying in the same week can hear a Marketplace percentage worked against one year’s table and an HHSC percentage against another, and neither office is wrong.

So do not compute your own Marketplace percentage from a table and treat it as a decision: the application does it with the correct year and with household composition rules that are easy to get wrong. And never carry a prior-year guideline into a new year.

Advance payment, and the reconciliation at tax time

You can take the credit in advance, paid monthly to the insurance company to lower your bill, or take none in advance and claim the whole thing when you file; either way it is reconciled on your federal tax return at the end of the year¹ (checked 30 July 2026). Reconciliation is the step that catches people, and it catches them precisely because the advance option is the one almost everybody chooses.

The mechanism:

1. You give the Marketplace a projected household income for the coming plan year.
2. The Marketplace uses that projection to set an advance credit, paid to the insurer each month.
3. After the year ends, the Marketplace sends a statement of what was paid on your behalf.
4. Your federal return compares the advance against what you should have received on your actual income. Earn less than projected and you can be owed more; earn more, and some of the advance may be repaid.

One case is not a partial adjustment at all. With the 400% FPL cliff back for plan year 2026, a household that projected below that line and finishes above it is not eligible for the credit for that year, and the advance is reconciled on that basis² (checked 30 July 2026). Federal rules limit how much of an excess advance payment some households have to repay. This site publishes no repayment cap figure, because none has been read off an IRS or HealthCare.gov source here and dated, and a cap quoted at the wrong income band leads somebody to take a larger advance than they can safely carry. Ask a tax professional. The stable part is worth saying plainly: an advance credit is an estimate rather than a grant, and the return is where the estimate is settled.

The people this catches are not careless. Their income genuinely could not be predicted in November: overtime that appeared, a second job in the spring, a good quarter in a small business.

How do you estimate an income you cannot predict?

Estimate the whole plan year, including everyone in the tax household, and update the Marketplace during the year instead of waiting for the return to settle it. HealthCare.gov publishes what to include and what to leave out, and self-employment income is entered as expected net profit rather than gross receipts¹⁰ (checked 30 July 2026).

The habits that reduce the damage, from the counselor’s side of the desk:

• Report changes when they happen. A raise, a new job, a household member joining or leaving, a marriage: each changes the credit, and reporting mid-year adjusts the advance rather than leaving a lump to settle in April.
• Keep the Marketplace account contactable. A changed phone number or email is the most common reason somebody never sees the notice that would have prevented the problem.
• Do not round an estimate down to reach a threshold. Just over a band and honest beats just under it and reconciled.

Which income counts, and how a household is defined, is in what counts as income in Texas. If a job ends mid-year, the choice between continuation coverage and a Marketplace plan changes everything: coverage if you lose your job. Changes outside the enrollment window run through special enrollment periods; open enrollment itself runs 1 November to 15 January, with 15 December the date for coverage starting 1 January¹¹ (checked 30 July 2026).

What this page does not publish, and why

Three numbers a reader might expect on a subsidy page are deliberately absent, for the same reason each time. A figure slightly wrong in the generous direction leads somebody to take an advance they must repay. One slightly wrong the other way convinces somebody they are not eligible, so they never file, and nobody finds out that they were.

• The required contribution percentages behind the benchmark plan. They are indexed, they moved when the enhancement expired, and this site has not read the current table off a federal source and dated it. Source of record: HealthCare.gov and the Marketplace application⁶.
• The repayment caps that apply at reconciliation. Not verified here, and income banded, which makes a half-remembered version worse than useless. Source of record: the IRS.
• Cost sharing reduction income tiers and actuarial values. The mechanism is explained above; the numbers are not published here until each has been read off a HealthCare.gov or CMS page and stamped⁸. See cost sharing reductions.

For the same reason there is no Texas premium figure and no Texas plan-selection total anywhere on this site. The only Texas Marketplace enrollment figures published here are the two KFF change figures above, quoted as a pair. What this page can stand behind is the base the missing percentages are measured against: the 2026 poverty guidelines, $15,960 a year for a household of one and $27,320 for a household of three³ (checked 30 July 2026). Each missing figure will appear here as a percentage of those, once it has been read off a federal source and dated.

Where the decision is actually made

The Marketplace application decides your credit, and it is the only thing that can, because it prices your household against the current benchmark in your county using the correct year’s guidelines. This page exists so you recognize what it is doing while it does it.

Free, unbiased help beats an afternoon of searching. HealthCare.gov’s local help finder lists Navigators and certified application counselors by address¹² (checked 30 July 2026). A counselor may not charge you and may not be paid by an insurance company in connection with your enrollment, and neither Navigators nor counselors give tax advice, which matters here: reconciliation is a tax question. An agent or broker is a different role, licensed to sell and generally paid by the companies whose plans they represent.

If a result does not match the facts you gave, do not simply reapply: see appealing a denial in Texas, and check what the rest of the household qualifies for at applying through YourTexasBenefits. For where subsidies sit among the other routes to coverage here, start at health coverage in Texas.

This site is independent. No part of it is run by, funded by or affiliated with the State of Texas, the Texas Health and Human Services Commission, the Texas Department of Insurance or the Centers for Medicare and Medicaid Services, and it has no role in deciding what you are paid or what you pay.