The US composite PMI rose to 56.0 in August, the highest level since April 2022. The consensus stood at 53.2.
And on that very same day, the 30-year yield returned to around 5.25%.
Errors or inaccuracies are possible. Spotted something that looks wrong? Write to me and I will correct it.
The Numbers
| Flash estimate, August 2026 | August | July | Expected |
|---|---|---|---|
| Composite | 56.0 | 54.5 | 53.2 |
| Services | 56.8 | 54.6 | 54.0 |
| Manufacturing | 53.2 | 53.9 | 53.9 |
Two indices beat expectations by a wide margin, and one missed. That is precisely the story.
What S&P Global writes in the release itself:
"US business activity growth accelerated sharply for a second successive month in August to reach the fastest since April 2022."
"A surge in service sector business activity helped offset a marked slowing of growth in the manufacturing sector" - a slowdown attributed in part to reduced inventory building and supply delays.
"Jobs were added at the fastest rate since the start of last year, as increasingly confident companies took on more staff to meet higher demand." And business growth expectations struck a nine-month high.
And Chris Williamson, Chief Business Economist at S&P Global Market Intelligence: "US business is booming... The survey data for the third quarter are currently pointing to annualized growth approaching 3.0%, up solidly from the 1.5% pace seen in the second quarter."
And What the Market Did With It
This is the reaction that defines the period, so it is worth pausing on.
| The 30-year yield, Friday | around 5.25% |
| The 10-year yield | around 4.70% |
| Gold | $4,585 an ounce, up 1.46% |
| S&P 500, previous session | -0.9% |
| Nasdaq 100 | -0.7% - a fifth straight decline |
A strong economic data point, and equities falling. That was not how it worked for fifteen years.
The pattern everyone grew used to after 2009 was simple: weak data → the market expects lower rates → multiples rise → equities rise. Strong data was sometimes bad news, because it postponed the relief.
In a period like this one it inverts: strong data → rates stay high, and may rise → the discount rate rises → multiples compress.
And above that sits a second layer: an accelerating economy generates more potential inflation, and therefore demands greater compensation for holding a 30-year bond. Which is why gold rose on the same day equities fell.
This is a structural change, not daily noise. It will not disappear because of one data point in the other direction.
How a Survey Number Becomes a Bond Price
And here it is worth taking the chain apart, because it is not intuitive - and this week you can watch it working on real numbers.
First, what the index actually measures. The PMI is a diffusion index: it counts how many respondents report an improvement, and adds half of those reporting no change. 56.0 does not mean the economy grew 56%, nor that it grew by any particular rate - it means the expansion is broad. So a jump from 54.5 to 56.0 is not a measure of intensity but of how much of the economy is taking part.
And that is precisely why the bond market responds to it: breadth is what predicts persistence.
And from here the chain splits in two, and this is the part that confuses most.
The first branch - the short end. The survey's input price and selling price components lead the consumer price index by one to two quarters. A company raising a price today creates inflation that will be measured tomorrow. The Fed responds to that, and the short end of the curve prices its response.
The second branch - the long end. And there sit three components, not one:
| The nominal 30-year yield | is made up of |
|---|---|
| The expected real rate | what the market thinks the price of money is over time |
| Expected inflation | how much that money will be eroded |
| The term premium | compensation for the uncertainty itself |
And the third component is the least understood. It is not compensation for expected inflation - it is compensation for the risk that the forecast turns out wrong, and for the quantity of paper that has to find a buyer.
And what is interesting this week is exactly where the movement happened. The Fed's minutes said explicitly that Treasury yields rose "driven by corresponding increases in real interest rates" - that is, not from inflation expectations.
And the strongest evidence for that sits in a single row of data:
The two-year yield stood at exactly 4.19% - on 17, 18, 19 and 20 August. Four consecutive trading sessions without a single basis point of movement.
And over those same four days, the long end travelled 12 basis points in both directions.
The practical meaning of that gap is larger than it looks.
Everything that happened this week - Treasury doubling its buybacks, the foreign holdings data, and the PMI itself - changed nothing in what the market thinks about the Fed's policy rate.
It changed what the market demands as compensation for time.
And why that bears directly on equities: the value of an asset is a discounted cash flow, so there are two variables - growth in the numerator, and the discount rate in the denominator.
When the discount rate rises because of growth, there is an offset: the numerator rises too. When it rises because of the term premium, there is no offset at all - only the denominator moves.
So you can see an excellent economic data point and equities falling on the same day, and it is entirely consistent. It is not that the market disbelieves the data - it is that the market is pricing what the data does to the cost of money.
And What Happened the Day Before Makes It Sharper
On 19 August the US Treasury announced it was doubling its long-end bond buybacks, and the 30-year yield fell from 5.28% to 5.19%.
The next day it closed at 5.23%. And on Friday, after the PMI, it is around 5.25%.
That is, everything the intervention achieved was erased within two days - and what erased it was a good economic data point.
And this is the point I want to stay with.
Treasury can buy paper. It cannot buy expectations.
When the economy grows at 3% rather than 1.5%, the market draws two conclusions at once: that the Fed will not rush to cut, and that inflation may return. Both push the long end higher.
And a technical operation in the secondary market changes neither of them.
Why This Bears Directly on Economic Policy
And here two numbers that do not appear in the same report need to be put together.
| The growth rate the survey points to for Q3 | about 3.0% |
| Employment | growing at its fastest since the start of 2025 |
| The federal deficit | about 6% of GDP |
A deficit of 6% of GDP in an economy growing at 3% and hiring is not supportive policy. It is pro-cyclical policy.
The logic of a large deficit is to cushion a recession - to inject demand while the private sector contracts. This is precisely the opposite of what is happening now.
And when the state borrows on that scale while the private sector is expanding, the two arms compete for the same money. The mathematical result is upward pressure on the real interest rate - and that is exactly what the Fed's minutes described this month: "Nominal Treasury yields rose 25 to 30 basis points, driven by corresponding increases in real interest rates."
So Friday's PMI is not only a data point about the economy. It is a data point about the price of the debt.
Three Things Inside the Index That Matter More Than the Headline
The headline is 56.0. These are the lines beneath it.
The first - the composition has flipped. In the second quarter manufacturing led and services lagged. Today it is reversed: services at 56.8, manufacturing at 53.2. And as Williamson puts it: "As reduced safety stock building and supply delays dampen factory production growth, the service sector is now playing a key role in driving a sustained US expansion, underscoring a dependency on consumer spending and financial services growth."
That is, the expansion now rests on the consumer. And that is a different base, and differently fragile, than an inventory cycle in manufacturing.
The second - delivery times. They lengthened again "to one of the greatest extents seen over the past four years", contributing to a build-up of uncompleted orders across both sectors. This is not noise: a physical bottleneck is constraining actual output - and according to the survey it is tied to shipping disruption, tariffs and thin inventories at suppliers.
And the third - prices. Price pressures moderated, especially in selling price inflation. But input costs remained elevated, mainly because of energy. And in the same week Brent traded close to $95 before easing to around $93.50.
Williamson says it explicitly: price pressures "remain elevated and prone to renewed upward pressures should energy prices rise again."
What Happens Next Week
| 27-29 August | Jackson Hole |
| 28 August | Kevin Warsh's first address as Fed Chair |
| The backdrop | The Fed held at 3.50%-3.75%, and three of its members wanted a rise |
| And what was added today | A survey pointing to 3% growth and recovering employment |
Warsh removed forward guidance from the committee's statements, which makes this address the main remaining instrument he has for communicating direction.
הזווית שלי
דעה אישית של אילן אברמוב - לא ייעוץ ולא המלצה
What I take from Friday is not the number 56.0, but what it did to equities.
An excellent economic data point, and a stock market falling for a fifth straight session. Anyone reading that through the pattern of the last decade will not understand it - and that is exactly the point. The pattern has changed.
And I am trying to formulate for myself what precisely changed. In a world of low inflation and zero rates, the economy and the markets pointed the same way: more growth, more profits, more value. In a world where inflation is alive and the long rate is 5.25%, they compete: more growth, more pressure on the price of money, a lower multiple.
And that changes something very practical. For years "good news" was a simple concept. Today you have to ask good for whom - for corporate profits, or for the discount rate that prices them. They pulled in opposite directions today, and the rate won.
And what troubles me more than the index itself is the combination it exposes: an economy at full employment, growing at 3%, with a deficit of 6% of GDP. We have not seen much of that in history, and on the occasions we have, it did not end quietly.
And here I want to be fair to the other side. There is an entirely reasonable reading that says all of this is temporary: the moderation in selling prices is real, the acceleration rests partly on one-off events, and if energy prices fall the pressure falls with them. On that reading the economy is simply strong, and that is a good thing. I do not dismiss it.
The real disagreement is a single question: is this growth coming from the private sector, or from the government injecting 6% of GDP into an economy that does not need it. The difference between those two answers is the difference between a sustainable expansion and one bought with debt issuance.
So what I will be looking for in Warsh's address on 28 August is not a hint about rates, but whether he talks about fiscal policy. A Fed chair who says explicitly that the deficit is a source of pressure on long rates - that is a sentence that moves the long end more than any data point.
(It is important to stress: this is my personal opinion only, and nothing herein constitutes a recommendation to take any action.)
