Best ACA affordability safe harbor for small employers

For most employers under 200 people the rate of pay safe harbor is the best choice. It is the only one of the three that hands you a firm dollar ceiling before the plan year begins, and it moves with what you actually pay, so your better paid employees can carry more of the premium. The Form W-2 harbor cannot price anything in advance because it settles after December 31. The federal poverty line harbor can, but it charges everyone at the floor.

The reason you need a safe harbor at all is that the statute defines affordability against the employee's household income, which no employer knows and none is entitled to ask for. So 26 CFR 54.4980H-5(e)(2) offers three substitutes you are allowed to measure instead. Get one of them right and section 4980H(b) cannot reach you for that employee, even if the coverage turns out to be unaffordable against their actual household income and they collect a premium tax credit anyway. That last part is the whole value of a safe harbor, and it is worth reading twice.

The three options, priced

For 2026 the affordability percentage is 9.96%, set by Rev. Proc. 2025-25 and up sharply from 9.02% in 2025. Here is what each harbor allows you to charge an employee per month for the lowest cost self only plan that provides minimum value.

The rate of pay figures are 9.96% of 130 hours times the hourly rate. The salaried figure is 9.96% of monthly salary, so $5,000 a month gives $498.00. Notice how much the ceiling moves with pay: the gap between an $18 and a $30 employee is $155.38 a month, or $1,864.56 a year, of premium you either can or cannot pass along.

Why the W-2 harbor is the wrong default

The Form W-2 safe harbor looks attractive because Box 1 wages are a real number you already have. The problem is when you have it. The regulation says application of this safe harbor is determined after the end of the calendar year and on an employee by employee basis. You set contributions in October for a plan year starting in January, and you find out in the following February whether those contributions were affordable.

That is backwards for anyone making a pricing decision, and it gets worse for the employees most likely to trigger a payment. Box 1 wages fall when someone takes unpaid leave, drops hours, or contributes more to a 401(k). Every one of those reduces the denominator and can push a contribution you set conservatively over the line, retroactively, with no opportunity to fix it. The harbor also requires the employee's contribution to stay a consistent amount or percentage of pay for the whole year, so you cannot adjust mid year to compensate.

The W-2 harbor earns its place in one specific situation: employees whose pay is substantially higher than their base rate suggests, because of commission or overtime. For a commissioned salesperson on a modest draw, Box 1 wages may be double the rate of pay figure, and the W-2 harbor gives you far more room. Use it for that category, uniformly, and use something else everywhere else.

The condition that voids the rate of pay harbor

There is one trap in the rate of pay harbor and it is not obvious from the summaries. For a non hourly employee, the regulation states that if the monthly salary is reduced, including because of a reduction in work hours, the safe harbor is not available. Not reduced proportionally, not recalculated: unavailable, for that employee, for that period. A salaried person who moves to a four day week in July takes their affordability protection with them, and you are back to the household income test you cannot measure.

Hourly employees are treated more gently but not exempt. You must use the lower of the rate on the first day of the coverage period or the lowest hourly rate during the calendar month. So a temporary rate cut, or moving someone from a shift that carries a shift differential onto one that does not, lowers the ceiling for that month even though their usual rate is higher. If you are near the line, a pay change is a benefits decision as well as a payroll one.

When the federal poverty line harbor is worth the cost

The federal poverty line harbor sets one flat ceiling for everybody: 9.96% of the poverty line for a single individual, divided by 12, using the poverty line for the state where the employee is employed. It is almost always the most expensive of the three, because it prices your whole workforce as though everyone earned at the floor.

It buys two things that the others do not. The first is certainty: an offer that meets the poverty line harbor is deemed affordable for premium tax credit purposes as well, so the employee cannot qualify for a credit through the Marketplace, and no credit means no 4980H(b) payment can be triggered at all. The second is simplicity, and for a small HR team that is not a trivial benefit. One number, applied to everyone, with nothing to recompute when someone changes hours.

For a workforce clustered near minimum wage the cost of that certainty is small, because the poverty line ceiling and the rate of pay ceiling are close together anyway. For a workforce with a wide pay range it is expensive, and the sensible pattern is a mixed approach: the poverty line harbor for your lowest paid category, the rate of pay harbor for everyone above it. That is explicitly permitted so long as the categories are reasonable and applied consistently.

Set the ceiling before you set the band

The practical consequence of all this is that affordability is downstream of pay design. Under the rate of pay harbor, each additional dollar of hourly rate buys exactly $12.95 a month of premium headroom in 2026. Which means the minimum of your lowest salary grade is not only a recruiting decision, it silently sets the maximum you can charge anyone in that grade for health coverage.

Employers discover this in the wrong order. They price the plan first, then find that the contribution they need is above the ceiling for their entry level roles, and end up subsidizing those employees more heavily than intended, or paying 4980H(b) for the ones who go to the Marketplace. Running it the other way around, so the band minimum and the contribution are set together, costs nothing and removes the problem. If you are building or refreshing grades this year, the salary structure work and the benefits budget belong in the same conversation.

One more thing to have in place before the plan year: documentation. A safe harbor is something you assert when the IRS proposes a payment in a Letter 226J, often two years after the fact, and the assertion is much stronger if the calculation was recorded when contributions were set rather than reconstructed afterwards. Keep the rate used, the category it applied to, and the resulting ceiling. Once you are running a group health plan you are also picking up the security and privacy controls that come with handling health plan data, so a single place to keep both sets of records tends to pay for itself.

The short version

Default to the rate of pay harbor. Add the W-2 harbor as a separate, uniformly applied category for commissioned or overtime heavy roles where Box 1 wages run well above base. Use the poverty line harbor for your lowest paid category if you want 4980H(b) risk removed entirely and the cost is small. Never rely on the rate of pay harbor for a salaried employee whose pay you might reduce mid year. And check the ceiling against your lowest band minimum before open enrollment, not after.

All of this only applies if you are an applicable large employer in the first place, which is decided by last year's average of full time employees plus full time equivalents. The calculator on that page prices the 2026 exposure and the affordability ceiling together.