ICHRA Affordability Safe Harbors: The 2026 Guide

An ICHRA — Individual Coverage Health Reimbursement Arrangement, the IRS-defined arrangement that lets an employer set a fixed monthly budget and let each employee buy their own individual health plan with it, tax-free — is only "affordable" if it clears one of three federal tests the IRS calls affordability safe harbors: the federal poverty line test, the rate of pay test, and the Form W-2 wages test. Each one multiplies a different number by the same annual percentage, 9.96% for the 2026 plan year, to produce a ceiling on what an employee can be asked to pay before your offer counts as unaffordable. This piece walks through all three, works the actual 2026 math for a real county, and shows how the same benchmark premium can require three different minimum contribution amounts depending on which safe harbor you pick. None of the three is automatically correct for every business; the right one depends on how your payroll is structured, and picking without running the numbers is the single most avoidable mistake in an otherwise straightforward ICHRA rollout.

Why guessing at a contribution number is risky

You have a headcount, a renewal deadline, and a target monthly budget in your head. Maybe it's $400 a person, because that felt round and defensible when you said it out loud. What you don't have, at the moment you pick that number, is any idea whether it clears the federal bar for "affordable" — and that single word decides whether your ICHRA does what you set it up to do.

Get it wrong in one direction and an employee who should have been able to use your ICHRA tax-free instead has to weigh giving up a premium tax credit they were entitled to keep. Get it wrong in the other direction, as an Applicable Large Employer, and you can end up owing a real federal penalty for an employee who declines your offer and gets subsidized marketplace coverage instead. Neither mistake announces itself immediately. Both surface later, usually when someone else — an employee's tax preparer, an IRS letter, a benefits attorney doing a year-two audit — asks a question you cannot answer with a number you can source.

The fix is not complicated, but it is precise: pick one of three federal formulas, run the real 2026 numbers, and set your contribution against the answer instead of a round figure that felt right.

This is also not a one-time task you finish at launch and forget. The required contribution percentage changes every year — it moved from 9.02% for 2025 to 9.96% for 2026, and it moves again to 10.22% for 2027 — and the federal poverty guideline behind one of the three safe harbors resets every January. A contribution that cleared the bar cleanly at launch can drift out of compliance a year later even if you never touched the dollar amount, simply because the bar itself moved and your county's benchmark premium moved with it. Treat this as an annual reset, on the same calendar as your renewal, not a decision you make once.

What "affordable" actually means here

"Affordable" is a defined term, not a feeling. Under Internal Revenue Code section 36B and the employer shared responsibility rules of section 4980H, an offer of coverage is affordable if the employee's required contribution for self-only coverage does not exceed a set percentage of their income — the required contribution percentage, 9.96% for plan years beginning in calendar year 2026, per IRS Revenue Procedure 2025-25. The same percentage decides two separate things at once: whether an Applicable Large Employer (ALE) — broadly, one that averaged 50 or more full-time-equivalent employees the prior year — risks a penalty for that employee, and whether that employee is allowed to keep a premium tax credit if they buy marketplace coverage instead of using your ICHRA.

Because an employer usually has no legal way to see an employee's actual household income, the IRS built three shortcuts, called affordability safe harbors, that substitute a number the employer can see — a national poverty figure, an hourly wage, or a W-2 — for household income. Pass one of them and the employer is protected from a penalty for that employee, full stop, regardless of what the employee's real household income turns out to be.

One more term matters specifically for an ICHRA, and it is where employers most often get confused: the lowest-cost silver plan (LCSP), also called the applicable benchmark plan. For a traditional group plan, affordability is tested against the group premium itself. For an ICHRA, it is tested against something narrower: the age-rated LCSP premium available in the employee's rating area (the geographic pricing zone insurers use, which does not always match county lines), minus the monthly ICHRA amount the employer actually makes available. The IRS and Treasury call that difference the required HRA contribution, and it is what every safe harbor formula below actually gets applied to — not the raw insurance premium (ICHRA final rule, Federal Register document 2019-12571).

The three IRS safe harbors, at a glance

All three safe harbors answer the identical question — "does this employee's required HRA contribution stay under 9.96% of something?" — but each one swaps in a different stand-in for household income. Here is the full 2026 comparison, every figure sourced independently.

The federal poverty line safe harbor

The FPL safe harbor swaps a household-income calculation no employer can verify for a single published number: the federal poverty guideline for a household of one. For a calendar-year plan beginning January 1, 2026, the rule lets an employer use the guideline "in effect within six months before the first day of the plan year" (26 CFR 54.4980H-5(e)(2)(iv)) — which means the 2025 guideline, published by HHS in the Federal Register on January 17, 2025, because the 2026 guideline would not exist yet at that six-month mark. That figure is $15,650 for one person in the 48 contiguous states and DC (90 Fed. Reg. 5917).

Multiply $15,650 by the 2026 required contribution percentage, 9.96%, and divide by twelve: $129.90 a month. That is the most any employee can be asked to pay, out of pocket, for the LCSP premium after your ICHRA contribution, for your offer to clear this specific safe harbor. It is the same number for every employee at every income level, which is exactly why it is the easiest of the three to administer — and, for most payrolls, the strictest, because it does not get more generous for a lower-earning employee the way the other two do.

The rate of pay safe harbor

The rate of pay safe harbor scales to what an employee is actually paid at the start of coverage. For an hourly employee, the rule multiplies 130 hours — a standardized full-time month, not the employee's actual hours worked — by the lower of their hourly rate at the start of the coverage period or their lowest hourly rate during the month, then applies the 9.96% ceiling (26 CFR 54.4980H-5(e)(2)(iii)). For a salaried employee, the same safe harbor applies 9.96% to their monthly salary as of the first day of the plan year instead.

Take an hourly employee earning $22 an hour. 130 hours × $22 = $2,860 in standardized monthly pay. 9.96% of that is $284.86 — more than double the flat FPL ceiling, because this employee's actual pay supports a higher affordable cost-share than the flat poverty-line number assumes. The tradeoff is administrative: this safe harbor has to be recalculated whenever a rate of pay changes, and hourly employers with wide pay bands end up running the formula separately for groups of employees rather than once for the whole class.

The Form W-2 wages safe harbor

The W-2 safe harbor uses the number your payroll system already produces every year: Box 1 wages on the employee's Form W-2. The rule applies the same 9.96% to those wages, divided by twelve (26 CFR 54.4980H-5(e)(2)(ii)). Unlike the other two, this safe harbor is tested after the year ends, against actual wages paid that year — which makes it the most precise of the three but also the hardest to use for setting a contribution amount in advance, since you are testing against a number you cannot know for certain until December.

Take a salaried employee earning $52,000 a year. $52,000 × 9.96% ÷ 12 = $431.60 a month — close to the rate of pay result for a comparably-paid hourly worker, because both methods scale to real earnings rather than a flat national figure. Most employers who use the W-2 safe harbor do so retroactively, as a confirmation check at year end, while using the FPL or rate of pay method to set the actual contribution going into the year.

Two things ICHRA adds on top

The three safe harbors above are general ACA employer-mandate rules; they existed before ICHRA did and apply to any employer coverage offer. The ICHRA final rule (Federal Register document 2019-12571) layers two more mechanisms on top, specific to how an ICHRA prices affordability against the individual market rather than a single group plan.

The location safe harbor. Because an ICHRA reimburses employees who may live in different rating areas, the rule lets an employer use either the employee's residence or their primary site of employment to identify the applicable LCSP premium — administratively useful for a company with a home office and a distributed workforce, since it means one consistent location can be used per employee rather than tracking moves in real time for affordability purposes specifically.

The look-back month safe harbor. Instead of pricing the LCSP fresh every month, a calendar-year ICHRA can use the January LCSP premium of the prior calendar year to set affordability for the whole upcoming plan year, applying the current year's age and location to that premium. This is what lets an employer set a single contribution amount before a plan year starts and know it will hold, rather than re-testing affordability every month as rates and ages shift.

One county, three safe harbors, three answers

Here is the whole mechanism worked through on one real county: Jackson County, Missouri, home to Kansas City. The 2026 individual-market benchmark premium there (age-50, silver tier) runs an estimated 27.6% below the small-group benchmark, a gap of about $2,800 a year per employee — which is exactly the kind of county this site's savings map exists to surface, alongside the rest of the state on our Missouri savings page. Set that county-level question aside for a moment and look only at affordability: the same $613-a-month LCSP benchmark premium produces three different minimum ICHRA contributions depending on which safe harbor you pick.

The spread is not a rounding error. Between the flat FPL method and the W-2 method for this $52,000-a-year employee, the minimum compliant contribution moves by $301.70 a month — $3,620 a year, on the identical benchmark premium, in the identical county. An employer using the FPL safe harbor across the board is not wrong; they are simply choosing the most conservative, least payroll-dependent method, and paying for that simplicity in a larger required contribution. An employer with a lower-earning workforce and the payroll infrastructure to run the other two formulas correctly can often set a smaller, equally compliant contribution instead.

Which safe harbor actually fits your payroll

Nothing stops an employer from running all three formulas for a representative sample of employees before choosing — in fact, that is the only reliable way to know which one actually produces the lowest required contribution for your specific workforce. The FPL safe harbor is not automatically the "safe" choice in a cost sense; it is only the simplest to compute.

How this interacts with employee classes

An ICHRA lets an employer split its workforce into permitted employee classes — full-time, part-time, salaried, hourly, seasonal, and several others defined in the final rule — and offer a different contribution amount to each class. Affordability is tested separately, class by class, which is exactly why the safe-harbor choice above matters more for a mixed workforce than a uniform one. A company with a salaried office staff and an hourly warehouse crew is not required to use the same safe harbor for both groups; it can run the W-2 method against the office class and the rate of pay method against the warehouse class, and set two different contribution floors that each reflect how that group is actually paid.

What it cannot do is cherry-pick the friendliest safe harbor for each individual employee inside the same class to minimize the contribution owed to any one person. The rule requires uniform treatment inside a class: whichever safe harbor an employer picks for a class has to apply to everyone in it, tested consistently. Decide the method at the class level, before you run anyone's individual numbers, and the rest of this article's arithmetic applies cleanly. Decide it employee by employee and you are no longer inside the safe harbor at all.

What getting it wrong actually costs

Two separate costs are in play, and they land on different employers.

For an Applicable Large Employer (50+ full-time-equivalent employees), an unaffordable offer that leads an employee to claim a marketplace premium tax credit can trigger a section 4980H(b) penalty: $5,010 per affected employee for the 2026 plan year, per IRS Revenue Procedure 2025-26. A related but separate penalty, 4980H(a), applies if an ALE fails to offer coverage broadly enough in the first place — $3,340 per full-time employee, minus the first 30, for 2026. These are indexed every year, and both went up from 2025 levels; do not assume last year's dollar figure still applies. See our guide to the 50-employee ALE rule for the full FTE counting method.

Put a number on it. A 60-person ALE sets a flat contribution using a safe harbor it never actually calculated, the offer turns out unaffordable for a subset of the lower-paid class, and five of those employees decline it and buy a subsidized marketplace plan instead. Five affected employees at $5,010 each is $25,050 in 4980H(b) exposure for the 2026 plan year — real money, for a mistake a single spreadsheet formula would have caught before the plan year even started.

For a smaller employer under the ALE threshold, there is no 4980H penalty exposure at all — but there is a quieter cost. An ICHRA that fails the affordability test for an employee does not stop being legal; it just stops accomplishing what most small employers actually want, because that employee has to choose between using the ICHRA and keeping a premium tax credit they may be relying on. An employer who set a contribution without running any of the three safe harbors has no way to know, in either direction, whether their offer is protecting employees' subsidy eligibility the way they assumed it was.

Either way, the tradeoff is the same and worth stating plainly: taking an ICHRA that is considered affordable generally means an employee waives their premium tax credit for the months that offer applies. Our ACA subsidy cliff guide covers the income side of that same interaction in more detail.

Setting your 2026 contribution, step by step

None of the arithmetic above is complicated once you have the right inputs in front of you. What trips employers up is skipping a step, not misunderstanding the math. Work through these in order and each one feeds the next.

1. Pull your actual county's LCSP benchmark premium, age-rated, for the rating area where each class of employees lives or works — not a national average.
2. Pick one safe harbor per employee class — FPL for simplicity, rate of pay for an hourly team, W-2 for a salaried team — and apply it uniformly inside that class.
3. Run the 2026 numbers, not last year's: 9.96% is this year's percentage, and the FPL figure resets every January.
4. Subtract the safe-harbor ceiling from the benchmark premium to get your minimum compliant contribution, the same arithmetic worked through above.
5. Document which safe harbor you used and why, in writing, before the plan year starts — this is the single easiest thing to produce later if anyone asks.
6. Send the required employee notice on time, generally at least 90 days before the plan year begins, explaining the ICHRA and the premium-tax-credit tradeoff.

Where ICHRA Savings fits

Affordability math tells you the floor your contribution has to clear. It does not tell you whether an ICHRA is the cheaper way to cover your team in the first place — that is a separate, county-specific question, and it is the one our county savings map is built to answer, using the same 2026 plan-year dataset behind the Jackson County numbers above. Before anything else, check whether your county is one where the individual market actually beats small group: https://ichrasavings.com/ichra-savings-map/. If the county math works and you want to see the full mechanics before talking to anyone, start here. If your renewal is already in hand and you are working against a real deadline, that is worth a conversation: https://ichrasavings.com/book.

What you get from getting this right

Get the safe harbor right and you get something concrete: an ICHRA contribution you can defend with a specific federal formula, employees who know exactly whether they are keeping or trading their subsidy eligibility, and — if you are an ALE — a documented reason a 4980H(b) penalty letter does not apply to a given employee. None of that requires guessing.

Questions employers actually ask