The 30-year Treasury is back at a level last seen in 2007, and the 10-year sits at 4.78%. This move is explained neither by the 25 basis points the Fed might add in September, nor by Brent at $95. It is explained by what sits between the two: the term premium, the price the market demands for locking up capital for thirty years in an economy whose fiscal rules it no longer knows.
There is a lazy way to read the yield curve: add up expected central bank hikes and declare the matter settled. It worked for fifteen years, back when forward guidance reduced the term structure to an arithmetic exercise on the Fed’s intentions. It no longer works. Since the meeting of 29 July, at the end of which the Federal Open Market Committee held rates in the 3.50% – 3.75% range by nine votes to three, short rates have eased slightly while long rates have risen. The 30-year reached its highest level since 2007. The 10-year stands at 4.78% as of 4 September, up 17 basis points on the month and 71 on the year.
A move of that shape is a message, and the message is not « the Fed is going to hike. » If the market were pricing a cyclical tightening alone, the curve would flatten: short rates up, long rates flat or lower, since a credible tightening pulls future inflation back to target. The opposite is happening. The curve is steepening from the long end. In bond language that has a name: the return of the term premium.
5.25%US 30-year TreasuryHighest since 2007
4.78%US 10-year Treasury, +71 bp on the year 4 September 2026
4.09%French 10-year OAT, 82 bp above the Bund 28 August 2026
$95Brent crude, against the EIA’s $85 quarterly forecast 3 September 2026
Supply shock or inflation? The old question, asked again
The immediate trigger is identified: renewed hostilities between Washington and Tehran. Brent traded around $95.26 on 3 September and WTI around $90.84, while the US Energy Information Administration was still projecting, in its 6 August outlook, an average of $85 for the third quarter. The Strait of Hormuz, through which roughly 20 million barrels a day passed before the conflict, had fallen to 2.7 million at the height of the March–May fighting. Traders today are not pricing a closure but the probability of one which is enough to move the futures curve.
The question a central bank must settle in the face of such a shock is theoretically simple and practically brutal. An oil price increase is, to a first approximation, a change in a relative price: energy becomes expensive relative to everything else, which impoverishes the net importer without creating inflation in the monetary sense. The orthodox response is therefore to « look through » the shock, because tightening against a supply shock means adding a recession to an impoverishment. That is the lesson drawn, in hindsight, from the first oil shock.
But it is also the opposite lesson drawn from the second. If the shock lasts long enough, agents stop treating it as temporary: wages re-index, contracts absorb the increase, and the relative price change turns into a drift in the general price level. This is precisely the warning Christine Lagarde issued after the Governing Council of 23 July: « the longer energy prices remain elevated, the greater the risk they will drive up general inflation through indirect effects and second-round impacts. »
A supply shock is a relative price until the day it becomes an expectation. Nobody knows in advance which day that is.
The central bankers’ dilemma, summer 2026
Two central banks, two real rates
Facing the same shock, the Federal Reserve and the European Central Bank do not start from the same place, and the chart below makes that visible. The Fed holds a range of 3.50% to 3.75% against US inflation measured at 3.4% in July: its real rate is slightly positive. The ECB holds its key rate at 2.25% against euro area inflation of 2.8%: its real rate is negative by roughly half a point.
Two central banks, one shock, opposite positions
Policy rate and inflation, United States and euro area, September 2026 (%)01234%3.50–3.75%3.4%United Statesreal rate ≈ +0.2 pt2.25%2.8%Euro areareal rate ≈ −0.55 pt Policy rateInflation, year on year.
| Region | Policy rate | Inflation |
|---|---|---|
| United States | 3.50 – 3.75% | 3.4% |
| Euro area | 2.25% | 2.8% |
Sources: Federal Reserve (meeting of 29 July 2026), ECB (Governing Council of 23 July 2026), consumer price indices. The US bar shows the midpoint of the official target range.
This asymmetry explains why the ECB, whose inflation is lower in level terms, signalled a September hike with an insistence the Fed did not put into its own communication. The ECB starts from a more accommodative position in real terms. It therefore has less room to ignore an energy shock. In the United States, the probability of a 25 basis-point hike in September implied by futures hovers around 52%, against roughly 65% in early August close to a coin toss, which is itself information.
Warsh, or the end of steering by speech
Fed Chair Kevin Warsh gave a press conference on 29 July notable for what it did not contain: forward guidance. Three committee members voted for an immediate hike. No path was promised. A number of commentators read this as a communication failure.
The opposite case can be made. Forward guidance rested on a rarely stated premise: that the committee knows the economy’s future path well enough to commit to it in advance. That premise has been contradicted systematically enough that it should no longer be treated as settled. There is a direct echo here of the Hayekian critique: an institution that announces a path is not merely communicating, it is claiming to hold information about the future state of the world that the price system is better equipped than it to aggregate. When the central bank falls silent, the yield curve returns to what it was meant to be, a discovery mechanism in which thousands of participants stake their own capital on their own estimate.
This reading has an unpleasant corollary for borrowers. A curve that becomes informative again is a more volatile curve, because it is no longer anchored by a promise. The price of monetary honesty is the term premium.
When the central bank stops promising, the market starts pricing again. That is more expensive, and it is more true.
The JTE view
What the long end is really pricing
A term premium is compensation for three distinct uncertainties: future inflation, the central bank’s reaction function, and the supply of paper. All three have deteriorated at once.
The third is the most underrated in public debate. A bond yield is a price, and a price forms where supply meets demand. A state that issues heavily pushes the price of its paper down and its yield up, independently of any inflation expectation. France offers the euro area’s clearest illustration. Its public debt reaches 117.5% of GDP, or €3,536 billion as of March 2026. Fitch confirmed the A+ rating with a stable outlook on 28 August, which relieved the market but raised its deficit projections to 5.2% of GDP in 2026, 5.5% in 2027 and 5.2% in 2028, against a government target of 4.9% for 2027. The gap, roughly 0.6 points, is some €18 billion a year. The 10-year OAT is back above 4% and the spread to the German Bund stands at 82 basis points, after a year-to-date high of 85 on 25 August.
Those 82 basis points have nothing to do with the Fed, nothing to do with Iran and nothing to do with euro area inflation, which is the same in Frankfurt as in Paris. They measure one thing: the extra yield demanded to hold French rather than German debt. It is a judgement on a fragmented political system’s capacity to deliver fiscal consolidation. Fitch said as much, citing « weaker growth, higher interest expenses, and additional defence commitments. »
Equities: dispersion is good news, concentration is not
The CAC 40 closed at 8,286 points on 3 September, up 0.07% on the day but down 1.65% on the week. Beneath that flat average, the spread between individual stocks is striking: Soitec jumped 10.27% in a single session and is up 422% year to date on demand for components destined for AI data centres; Publicis gained 4.4% after winning PepsiCo’s global advertising account, estimated at $1.7 billion; Voltalia fell 17.69% after withdrawing its 2026–2027 profitability targets and suspending dividend plans.
This dispersion is healthy. A market in which every stock rises together is a market driven by liquidity, in which capital no longer flows to the best projects but to the index. A market in which Soitec gains and Voltalia loses on the same day is a market doing its job allocating capital by judgement, company by company. Positive real rates help: when money has a price, projects have to justify it.
The concern lies elsewhere. Broadcom’s 6.6% fall, despite solid quarterly results but guidance judged disappointing, erased more than $100 billion of market value in one session. When a guidance revision at a single semiconductor firm destroys the equivalent of a mid-sized country’s annual GDP, that is no longer information about Broadcom: it is information about market concentration and the scale of index flows that mechanically track a handful of names.
The other case: why the term-premium reading may be wrong
1. It may simply be inflation. With Brent at $95 and US inflation at 3.4%, the rise in long rates may be no more than a revision of long-run inflation expectations, with no fiscal component. Term premium measures are model estimates, not observations.
2. Second-round effects justify acting. The case for looking through a supply shock assumes anchored expectations. After several years above target, that assumption is fragile, and the experience of 1979 suggests the cost of tightening late exceeds the cost of tightening early.
3. Forward guidance had genuine value. Central bank silence does not make markets wiser; it raises volatility and therefore the cost of capital for companies and states — including those managing their finances properly.
4. The French spread may be more technical than political. The Bund’s liquidity, prized as a reserve asset, mechanically compresses its yield. Part of those 82 basis points measures German scarcity as much as French risk.
What to watch
The September FOMC meeting will settle a debate three dissenters have already opened publicly. In Europe, the ECB has signalled a hike — the question is whether it frames it as a one-off adjustment or the start of a cycle, which will not have the same effect on the long end. For France, three rating reviews follow in sequence: Scope on 18 September, Moody’s on 23 October (Aa3, negative outlook), Standard & Poor’s on 27 November. Finally, $100 Brent remains conditional, according to traders, on a physical reduction in supply — restricted traffic through Hormuz, a confirmed strike on infrastructure, a production shutdown — rather than on perceived risk alone.
The useful conclusion for an investor is not a forecast. It is a change of framework. For a decade it was enough to follow the central bank. Now three independent variables have to be followed: the price of energy, the monetary reaction function, and the fiscal path. The US 30-year at 5.25% and the OAT 82 basis points above the Bund are saying the same thing in two different languages — that the third variable has stopped being free.
Sources
- J.P. Morgan Wealth Management, revised rate expectations after the 29 July 2026 meeting.
- Trading Economics, US 10-year and 30-year Treasury yields as of 4 September 2026.
- Daf-Mag, ECB decision of 23 July 2026 and Christine Lagarde’s remarks.
- Polyester Time, Brent and WTI prices and the Strait of Hormuz, 3 September 2026.
- France Épargne, Fitch review of 28 August 2026, OAT levels and the Bund spread.
- MoneyVox, market report, 3 September 2026.
The number
82 bp
The OAT–Bund spread. Neither the Fed, nor Iran, nor euro area inflation enters into it: it measures French fiscal and political risk alone.
Key takeaways
- The curve is steepening from the long endThe signature of a term premium, not a simple tightening.
- Two opposite real ratesSlightly positive in the US, negative in the euro area.
- Fed silenceFewer promises, more volatility — and a curve that is informative again.
- Equity dispersionSoitec +422% year to date, Voltalia −17.7% in one session.
