Financial health
Median interest coverage ratio: 1.65x
Across 2,271 US companies holding an interest coverage ratio on 2026-09-07 the median was 1.65 times, below Damodaran's 3.0 investment-grade floor.
By Javier SanzPublished
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
Interest coverage is operating profit divided by the interest bill. On 2026-09-07, across 2,271 US companies with filed statements on record and an interest coverage figure among them, the median was 1.65 times.
Damodaran's synthetic-rating table maps coverage to a bond rating, and the edges this site uses come from it: below 1.5 is distress territory, 1.5 to 3.0 is speculative grade, 3.0 to 6.0 is investment grade, above 6.0 is comfortable. The median company in the cohort sits in the speculative band, 0.15 above the distress edge.
| Percentile | Interest coverage |
|---|---|
| 5th | -250.64 |
| 10th | -54.86 |
| 25th | -6.28 |
| 50th | 1.65 |
| 75th | 6.78 |
| 90th | 23.01 |
| 95th | 60.52 |
From GET /api/public/distribution/interest_coverage, build dated
2026-09-07. One deduplicated company per row. Percentiles are the endpoint's own values rounded to two decimals, halves away from zero.
A quarter of the cohort has a negative numerator
The 25th percentile is -6.28. A negative interest coverage ratio does not mean the interest is cheap. It means operating profit is negative, and dividing a loss by an interest bill produces a number that looks like a ratio and is really a statement that there was nothing to cover the interest with.
This is why the percentiles below the median are so violent. At the 10th percentile the figure is -54.86 and at the 5th it is -250.64, and neither is a measurement of leverage. They are measurements of the size of an operating loss relative to a small interest expense. The whole lower half of this distribution is a different question from the one the ratio is usually asked.
Read the upper half instead and the ratio behaves. At the 75th percentile operating profit covers interest 6.78 times, past the comfortable edge. At the 90th it is 23.01 times, which is a company with an interest bill that barely matters.
The sector that fails the test by design
| Sector | Companies | 25th | Median | 75th | 90th |
|---|---|---|---|---|---|
| Healthcare | 473 | -58.00 | -7.14 | 2.34 | 13.19 |
| Technology | 372 | -8.72 | 1.73 | 9.36 | 41.69 |
| Industrials | 370 | -2.14 | 2.69 | 8.14 | 19.60 |
| Consumer Cyclical | 248 | -1.81 | 2.83 | 10.01 | 33.90 |
| Energy | 165 | -0.41 | 2.56 | 5.27 | 12.12 |
| Basic Materials | 134 | -1.75 | 2.29 | 11.67 | 36.71 |
| Communication Services | 129 | -1.89 | 0.64 | 3.55 | 8.24 |
| Consumer Defensive | 128 | -1.06 | 3.41 | 9.80 | 41.65 |
| Real Estate | 88 | 0.43 | 1.62 | 3.26 | 4.98 |
| Financial Services | 82 | -8.36 | 4.53 | 13.21 | 40.70 |
| Utilities | 81 | 1.62 | 2.38 | 2.73 | 4.35 |
The utilities cohort is the interesting row. Its 25th percentile is 1.62 and its 75th is 2.73, so the middle half of the sector sits inside the speculative band and outside investment grade on this measure. Its 90th percentile, 4.35, is the first point in the sector where the figure reaches the investment-grade range at all.
Regulated utilities are among the most reliably investment-grade issuers in the market. The ratio is not wrong and the rating is not wrong. What is wrong is applying one boundary to a business whose entire model is heavy borrowing against a regulated and therefore predictable revenue stream. A lender looking at that business cares about the predictability of the numerator at least as much as its size, and the ratio carries no information about predictability at all.
The other direction of the same error is healthcare, whose median is -7.14. That is not a sector of distressed borrowers. It is a sector where a large share of the listings have no operating profit yet, and the ratio is reporting that fact rather than a debt problem.
What the figure is, precisely
Operating profit over interest expense, from the filings. Not EBITDA over interest. EBITDA coverage is a different and usually larger number because it adds back depreciation and amortisation, and a search for "EBITDA interest coverage" that lands here should know that the numerator here is the smaller one. Whichever is used, it is an accounting profit and not cash. A company can cover its interest on paper and not in the bank.
A company with no interest expense at all is withheld, not shown as infinity. A debt-free balance sheet is a different fact from good coverage, and printing an enormous number would merge the two.
Banks and mortgage REITs are excluded from the cohort. 359 companies were withheld on that rule at this build. For a lender, interest is both the main expense and the main revenue line, so the ratio is not measuring the burden of debt. It is measuring the shape of the loan book.
On 2026-09-07 the cohort was 2,271 companies and the coverage was 2,786 listings. Those are two different counts of two different things: the first is deduplicated companies at the verified tier that produced the percentiles, the second is listings holding the figure at all, out of a pool of 5,257 that day. Neither is the size of the market.
Both numbers move every night, so here they are as this page is being read: the pool holds 5,323 (counted ) listings, 1,450 (counted ) of them carry an interest coverage ratio, and the cohort behind the percentiles is 1,443 (built ) companies.
Graham's version of the same test
Graham asked, in the defensive-investor criteria of chapter 14, for average earnings to cover interest at least five times. He asked for the average rather than the year, which does most of the work: a single year's coverage on a cyclical business is a reading taken at whatever point of the cycle the filing happens to sit in.
Applied to this cohort, a five-times line sits between the 50th and the 75th percentile. That is a stricter test than the investment-grade floor, on purpose. Graham was not trying to rate the debt. He was trying to describe a company whose interest bill could not become a problem.
What this leaves out
These are cross-sectional percentiles from one build, dated 2026-09-07, on US listings only, at the verified tier, one company per row. They describe where a figure sat, not whether any company is worth owning.
The figure uses operating profit and not EBITDA and not cash from operations. Three different numerators are all called interest coverage in ordinary use, and they do not agree.
Every band edge quoted here comes from Damodaran's synthetic-rating table as cited in this product's own metric registry, and from Graham's chapter 14. A band is a published convention, not a boundary this product estimated. A company on one side of an edge is not thereby rated, and a rating agency's actual assessment uses far more than one ratio.
The percentiles are winsorised at the 1st and 99th percentile before they are computed, clamped to the observed edge rather than dropped, so the cohort size is the cohort that produced the numbers. Market-wide that clamp sat at -4,056.00 and 512.67, which is itself a description of how badly this ratio behaves at its tails.
One company was excluded market-wide for holding an interest coverage of exactly zero, on the rule that an exact zero on this figure is a missing value somewhere upstream rather than a filed one.