76% Passed the Cost On. 60% Ate It. Both Numbers Are True.

Two numbers from the Federal Reserve's latest survey of small employers, printed a few lines apart.

Seventy-six percent of firms with foreign inputs passed at least some of the cost increase on to customers. Sixty percent absorbed at least some of it.

Those add to 136%, which is not an error, and the excess is the entire story. Most firms did both. They put the price up and took the margin hit, on the same inputs, in the same year, because passing all of it on would have cost them the customer and eating all of it would have cost them the business.

If you run something and you have spent the last eighteen months doing exactly that — nudging the price, quietly wearing the rest, telling yourself you'd fix it properly next quarter — this is the data saying you were not being indecisive. You were doing what nearly everybody did.

The overlap is the finding¶

Read the two percentages as a choice and you get a story about winners and losers: three-quarters of firms had pricing power, the rest didn't. That story is wrong, and the arithmetic says so. Seventy-six plus sixty is a hundred and thirty-six. At least thirty-six percent of firms are in both buckets, and given how these questions are asked — at least some — the real overlap is larger.

The behaviour underneath is the one every operator recognises. A supplier raises an input eleven percent. You do not raise your price eleven percent, because you know precisely which customers leave at that number. You raise it five, you wear six, and you tell yourself the six is temporary. Then the next increase arrives on top of a margin that is already thinner than it was.

That is not a pricing strategy. It is a slow transfer from the balance sheet to the customer, metered out at whatever rate the owner believes the market will tolerate, and it is what "firms absorbed the cost" actually looks like from the inside. It is also the mechanism sitting underneath a good share of the failure statistics — margin does not vanish in a single event, it gets given away in increments nobody logged.

Where it landed¶

Forty-eight percent of firms sourced at least some of their inputs from outside the United States in 2024. Fourteen percent sourced more than half. That second number is the one to hold on to, because it defines the group with no room to manoeuvre: a firm buying most of its inputs abroad cannot re-source its way out of a tariff in a single year, and the survey shows that it did not try.

The bottom three bars are the ones worth staring at. The stated purpose of a tariff is to change where things are made. In this survey, across 6,525 employer firms, three percent relocated production and thirteen percent found a domestic supplier. Twenty-one percent changed their sourcing at all; seventy-nine percent changed only their price or their margin.

That is not evidence that re-sourcing is impossible. It is evidence that within roughly a year, at small-firm scale, it is slower than the cost arrives. A supply chain has switching costs — qualification, tooling, minimum orders, the six months where the new supplier is worse — and a tariff has none. The cost lands in the next invoice.

The number that should worry you is not about tariffs at all¶

Buried past the cost questions are two indices about what owners expect next.

Both are the lowest since the 2020 survey — the one taken in the middle of a pandemic that had closed the doors of a large fraction of the respondents.

That comparison deserves care rather than drama. It does not mean conditions are as bad as 2020; it means owners' forward expectations are as low, which is a different and in some ways stranger thing. In 2020 the pessimism had an obvious cause and an obvious end. This is pessimism during a year in which sales, for a lot of operators, held up.

The most plausible reading is the one the pass-through numbers set up: firms have been absorbing cost for long enough that they have stopped believing the absorption is temporary, and they are planning hiring and investment on the assumption that the margin does not come back.

What this changes about what you do¶

Nothing about a tariff is within your control. Three things about your response are.

1. 1

   Find out what share of your inputs are actually exposed

   Not "do I import" — most owners answer that wrong in both directions. Go down the input list and mark which ones your supplier imports, including the ones you buy domestically from an importer. Forty-eight percent of firms had foreign exposure; a meaningful share of those learned it from a supplier's price letter rather than from their own records.

2. 2

   Write down the split you have actually been running

   You have been passing some fraction on and eating the rest. Most owners have never calculated which. Take the last four input increases, compare them to the last four price changes, and you have your real number. It is usually further toward "absorbed" than the owner believes. It also moves your taxable profit, which is worth knowing before it surprises you — the basics are here.

3. 3

   Decide the split deliberately, once, in advance

   The damage in this data is not that firms absorbed cost. It is that they absorbed it by default, increment by increment, without ever choosing a number. A stated rule — we pass through two-thirds, we review quarterly — is worse than a perfect answer and much better than drift.

4. 4

   Price the re-sourcing question honestly, then probably drop it

   Thirteen percent found a domestic supplier and three percent moved production. If you are not already deep into that work, the survey suggests it is unlikely to rescue this year's margin. It may still be the right multi-year call. It is not a cost fix.

The part worth being careful about¶

This survey is a convenience sample. Firms were recruited through partner organisations rather than drawn at random from a national frame, and the published results are weighted but not a probability projection of the roughly six million US employer firms.

That is not a reason to ignore it — it is the largest and most consistent regular read on small-employer conditions in the country, run by the Federal Reserve system, with 6,525 responses collected between 3 September and 14 November 2025. It is a reason to treat the direction and the relative figures as strong and the exact decimal as indicative. "Retail was hit harder than manufacturing, which was hit harder than everyone else" is a solid finding. "Exactly 69% of American retailers" is not a sentence this survey can support.

It's just business¶

The tidy version of a tariff is that it changes where things are made. The version in this data is that it changes who eats the difference, and the answer turned out to be everybody, a bit — the customer through the price, the owner through the margin, and the employee through a hiring plan that quietly got smaller.

Nobody in the survey decided that. Each firm made a reasonable local choice about how much to pass on, four in ten reported the cost as a challenge, and the aggregate came out as an expectations reading not seen since 2020.

If you are going to absorb part of it anyway — and the data says you will — the one thing worth doing is choosing the fraction on purpose. For the arithmetic of tracking that, bookkeeping is the twenty-minute-a-week version, and the broader picture of thinner traffic and fatter tickets is the same squeeze seen from the till.