What a 100bp rate rise costs: the median filing says $2.9m

The Federal Open Market Committee next meets on 15-16 September 2026. For two years the interesting direction was down. It is no longer obvious which way the next move goes, which makes a paragraph that most readers skip suddenly worth reading.

Somewhere in Item 3 of almost every quarterly report, a company answers this question about itself: if interest rates rose by one percentage point, what would it cost? The answer is a number the company calculated, on its own debt, as of a stated date. Nobody has to guess.

We read every one filed this summer. The median answer is $2.85 million a year.

The census

The population is every Form 10-Q filed between 1 July and 31 August 2026 whose text contains the phrase 100 basis point — the June quarter, in other words, filed after it closed. That is 393 companies. What they mean by the phrase is not the same thing.

Effect on their own interest expense, priced    93
Bank rate-shock and net-interest-income tables  46
A discount rate in an impairment test           76
Effect on interest income, priced                8
Uses the phrase without pricing any of these   170

That third row is the one to watch if you are searching for exposure. When Cable One says a 100 basis point increase in the discount rate would decrease fair value by $116 million, and Angi says the same move would produce approximately $35.0 million of additional impairment, neither is talking about the Fed. They are talking about the rate used to discount future cash flows in a goodwill test. The phrase is identical; the subject is not.

The 93 that priced their own debt

Among the companies that do answer the question about their interest bill, the striking thing is how small the answers are.

Fifty-five of the 70 annual figures are below $10 million. Fourteen are below $1 million — Steven Madden puts the effect on its term loan at approximately $1,300 a year. Las Vegas Sands, at the top, is 20 times the median.

The reason is structural, and it is the same reason we found that two years of Fed cuts never reached corporate interest expense. Large US borrowers fund themselves with fixed-rate bonds issued years ago. A policy move touches only the floating slice — the revolver, the term loan, the securitisation — and reaches the rest slowly, through refinancing. This disclosure measures the part that reprices immediately, and for most filers that part is small relative to the balance sheet.

Which is why the exceptions describe business models rather than balance sheets. AutoNation’s $40.6 million is floorplan interest expense — the cost of financing cars sitting on lots, which is floating by construction. Asbury’s $20.9 million is the same thing at a smaller dealer group. Las Vegas Sands is exposed through HIBOR and SORA, not SOFR, because its debt sits in Macao and Singapore. Xcel Energy’s figure is $25 million against $8 million disclosed for the same quarter a year earlier, which is what a utility shifting into short-term borrowing looks like in a single line.

The number is there. The unit often is not.

Twenty-three of the 93 filings state a dollar figure and never say over what period it applies. That is one in four, and it makes the raw figures dangerous to rank.

CoreWeave is the clearest case, because it says so explicitly. Its filing gives “for every 100-basis point increase or decrease in interest rates, our interest expense could increase or decrease by approximately $30 million and $61 million” — for the three and six months ended 30 June 2026. Taken at face value against the annual figures above, $30 million looks mid-table. On the company’s own six-month run rate it is closer to $122 million a year, which would be the largest number in this article by some distance. That is consistent with what we found when CoreWeave’s depreciation and interest exceeded its $1.5 billion of EBITDA.

Bunge’s $102 million is the largest single figure in the census and its filing attaches no period to it at all. We have not annualised it, because the filing does not say what to annualise.

Not everyone is on the same side

A smaller group discloses the mirror image. Paycom holds client payroll funds between collection and disbursement, and says a 100 basis point move would change interest earned on funds held for clients by approximately $25.7 million over the following twelve months — upward, if rates rise. Regions Financial, a bank, runs the question through a full rate-shock table rather than a sentence.

For companies with more cash than floating debt, the September meeting is an income event, and their Item 3 says so in the same language everyone else uses for a cost.

What this count is and is not

• It is a census of a phrase, not of exposure. A company with enormous floating-rate debt that writes “one percent” instead of “100 basis points” is absent by construction, and a company with none at all can appear because it used the phrase in a goodwill test.
• The figures are the companies’ own arithmetic, not ours. Each rests on that filer’s balance sheet at its own date, its own hedges, and its own assumption that the balance does not change. They are not forecasts and they are not comparable in the way a ranked list implies.
• No total is computed here. Adding 93 numbers measured over different periods, on different reference rates, against different debt stacks would produce a figure that means nothing. The median is reported instead, and only across the 70 that state an annual period.
• A hundred basis points is a large move. The disclosure is a standard stress test, not a prediction of what the FOMC will do in September, and nothing here says which way rates go.
• Item 3 has an exemption. Smaller reporting companies are not required to provide it at all, so the smallest filers are systematically thinner in this population.

Check it yourself

The population comes from one public query, linked in the sources: the phrase, restricted to Form 10-Q, restricted by filing date. Every figure quoted above sits in Item 3 of the filing beside it, usually as a single sentence, and the sentence almost always names the date the balance was measured.

One filer in the group that names the exposure without sizing it is nCino, which took on a $200 million floating-rate term loan during the quarter and spent $175.7 million buying its own shares against $115.6 million of operating cash flow. Its 10-Q says a 100 basis point change would not have been material; the sum it leaves to the reader is about $2 million a year.

The method is the one behind our other counts — read the whole population, then report what is missing from it as carefully as what is in it, the way we did when 422 filings named the Strait of Hormuz and two priced it. For what actually happened to interest bills rather than what might, the first half of 2026 is measured here.

No share prices appear in this article. The FOMC date is from the Federal Reserve’s published calendar. Every calculation above is written out in this article’s front matter and re-checked when the site builds.

If a figure here does not match a filing, tell us and it will be corrected on the article and on the corrections log, with the date.