Three economic releases came out today, and each contains an internal contradiction. Not a disagreement among commentators - a contradiction between numbers in the same report, or between two reports measuring the same thing.
That sounds like a fault. It is not. When an index contradicts itself there is usually a structural reason, and finding it produces a better picture than either number would give on its own.
Errors or inaccuracies are possible. Spotted something that looks wrong? Write to me and I will correct it.
The First Contradiction: Two Indices for One Sector, Two Directions
Two purchasing managers' indices for US manufacturing were published today, for exactly the same month.
| August, US manufacturing | July | August | |
|---|---|---|---|
| ISM | 55.6 | 54.6 | fell a point, missing a 55.2 forecast |
| S&P Global | 53.9 | 53.9 | revised up from a 53.2 estimate |
One fell and missed. The other was revised upward. Both measure US manufacturing activity in August 2026.
The explanation is not that one of them is wrong. They are not the same instrument, and the differences between them are explicitly documented:
- Weighting of respondents. The ISM panel is "self-weighting" - every response counts equally, regardless of company size. S&P Global weights each response by company size and by the importance of the sub-sector within the economy.
- Weighting of components. ISM gives its five components equal weight. S&P Global assigns different weights: new orders 0.30, production 0.25, employment 0.20, supplier deliveries 0.15 and inventories 0.10.
- Sample size. About 1,200 manufacturing companies at S&P Global, against roughly 800 at ISM.
- Seasonal adjustment. S&P Global re-estimates seasonal factors every month; ISM sets the coming year's factors in advance.
- Timing. The S&P Global flash estimate is collected in mid-month, and the upward revision reflects responses gathered after it. The ISM survey covers the full month.
The component-weighting line is the one that bears directly on today's reading, and we come to it shortly.
The Second Contradiction, Inside ISM Itself
Breaking the ISM figure into its components shows that the headline conceals more than it reveals.
| Component | July | August | Change |
|---|---|---|---|
| New orders | 56.7 | 53.7 | -3.0 |
| Backlog of orders | 55.0 | 51.8 | -3.2 |
| Imports | 55.7 | 52.5 | -3.2 |
| Employment | 52.8 | 51.2 | -1.6 |
| Production | 58.5 | 58.3 | -0.2 |
| Supplier deliveries | 58.9 | 59.3 | +0.4 |
| Prices | 71.1 | 71.1 | unchanged |
Note the supplier deliveries line, because it teaches something about how the index is built.
In the ISM formula, slower delivery raises the headline index. The original logic was that when demand is strong, suppliers struggle to keep pace. But delivery can also slow for the opposite reason entirely - shortage, disruption or a bottleneck.
In other words, part of the number 54.6 rests on a deterioration counted as an improvement. The report itself lists electronic components in short supply for 18 months and electrical components for 14 months, alongside copper and memory chips. That is the context in which delivery is slowing.
And here the weighting difference returns: at ISM supplier deliveries carry a fifth of the index; at S&P Global only 0.15. The component that behaves misleadingly carries the greater weight in the index that fell.
And there is a further gap worth pausing on: production stands at 58.3, and new orders at 53.7. A factory producing faster than orders arrive is eating into its backlog. And that is exactly what the data show - the backlog of orders fell 3.2 points.
And the Third Contradiction, the Most Important of Them
This one sits between the two reports, and it concerns prices.
ISM reports a prices index of 71.1 - unchanged from July, and a 23rd consecutive month of price increases.
S&P Global reports that manufacturing input cost inflation moderated for a third consecutive month, and that input costs rose at their slowest pace since February.
Both can be true at once, and it is worth understanding how. The ISM prices index does not measure the size of the increase but the breadth of its spread - how many respondents report paying more. A reading of 71.1 says that a large majority is paying more, even if the increment itself is smaller than before.
So: the increases are less sharp, but they still touch almost everyone. Those are two different statements about the same reality, not a genuine contradiction - but a reader who sees only one of the two headlines gets a partial picture.
Exactly the Same Pattern, in Europe
The European release published today carries the same structure.
| Euro area, August | July | August |
|---|---|---|
| Headline inflation | 2.9% | 3.3% |
| Core inflation | 2.5% | 2.4% |
| Energy | 10.3% | 14.3% |
The headline rose to its highest level this year. The core fell. And the difference between them is explained almost entirely by the third line.
By country: Spain 4.5%, Italy 3.2%, Germany 2.9%, France 2.7%.
Markets fully price a European Central Bank rate rise to 2.50% on 10 September.
What the Three Contradictions Have in Common
Set side by side, they turn out to be not three phenomena but one.
In all three cases, the headline measure is being pushed up by prices while the demand measures underneath it soften.
In Europe this is explicit: energy at 14.3% lifts the headline, and the core falls.
In the US it is the same thing in different dress: the ISM prices index is locked at 71.1, while new orders, the backlog, imports and employment all fell.
And the common factor is familiar: energy. A barrel of Brent crossed $90 yesterday against the background of events in the Gulf. One respondent to the ISM survey put it this way: "The economy is annoying; it is getting in the way of otherwise good business" - attributing it to tariffs and the effect of the Middle East conflict on inflation.
And What the Data Do Say About Growth
So as not to leave a one-sided picture, the other side deserves to be stated in full.
ISM itself notes that a 54.6 reading is historically consistent with 2.4% annualised GDP growth. Manufacturing has expanded for an eighth consecutive month, and the overall economy for a 22nd.
At S&P Global the picture is more positive still. The flash estimate published on 21 August reported that US business activity is growing at its fastest pace in more than four years, and that survey data for the third quarter point to annualised growth approaching 3.0%, against 1.5% in the second quarter. It also stated that jobs growth revived in August, with employers gaining confidence as concerns over tariffs and the Middle East conflict fade.
That is hard to reconcile with the fall in ISM employment to 51.2, and it too is part of the picture.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
What I take from today is mainly a reminder of a working rule, not a conclusion about the economy.
When an index prints and its headline does not match what is happening underneath, the natural instinct is to pick a side - to decide that one measure is "right" and the other is noise. I think that is almost always the least useful choice. Both indices are correct within their own definitions, and the difference between them is the information.
And in today's case, the difference points entirely in one direction: what is lifting the headline numbers is not demand. New orders fell three points, the backlog fell 3.2, employment fell 1.6. What is rising is prices, and what is lifting prices sits mostly in the energy line.
That is also what makes the next two weeks unusually interesting. Two central banks are due to decide
- the European on 10 September and the American on the 15th and 16th - and both face the same uncomfortable question: inflation driven by energy does not respond well to interest rates, because rates do not produce barrels. But inflation that persists long enough stops being energy and starts being expectations.
What I will track is not the headline next time, but two specific components: ISM new orders, and the European core line. If orders keep falling while prices stay high, that is no longer a contradiction between indices - it is a single trend with a name.
(It is important to stress: this is my personal opinion only, and nothing herein constitutes a recommendation to take any action.)
