The 10-year US Treasury yield is near 4.80 per cent. Japan’s 10-year has reached 3 per cent for the first time since 1996. Brent is holding around $94-95 a barrel. After the Fed stopped offering rate guidance, Friday’s employment report will show whether the move is a passing scare - or whether long-term government debt is now permanently more expensive to fund.
2 September 2026 · Capital markets · Figures as of the afternoon of 2 September; the analysis to 31 August is unchanged
A sixteenth-century Venetian merchant did not buy oil and did not read central-bank minutes. He paid a premium on a ship that had not yet left harbour. The extra was not a weather forecast. It was the price of not knowing, on the Rialto, what waited in the eastern Mediterranean.
On 1 September 2026 the market struck a similar bargain. The 30-year US Treasury yield is near 5.25 per cent. European gas is the dearest it has been in three and a half years. The United States and Iran exchanged fire in the Strait of Hormuz for the first time in a month. The wires talk of oil, inflation and the Fed. That does not explain the rise in long-dated government yields. Part of the move is not a new inflation forecast. It is an extra premium investors now demand to hold long-term sovereign debt.
Bonds are already repricing time. Equities are still betting on distant profits from artificial intelligence. The two instruments are not speaking the same language.
What the term premium is - and why the 10-year yield is rising
The yield on a long-term government bond has two parts. The first is the expected path of short-term rates over the life of the bond. The second is extra compensation for the risks of holding duration: rates may move the wrong way; inflation may erode the real value of the coupon; the Treasury may issue more debt than investors wish to absorb; the Fed may change policy without prior guidance. That extra compensation is the term premium.
The term premium is not traded on an exchange. It is inferred from models. At the end of August the New York Fed’s Adrian-Crump-Moench (ACM) model split the 10-year yield into about 3.98 per cent of expected short rates and about 0.79 per cent of term premium. Other models put the premium a little higher. The estimates differ. The direction does not: the premium is back above zero.
In early 2025 the same measure stood near 26 basis points. By June it was around 80. From late June the nominal 10-year yield rose by about 36 basis points. Inflation expectations taken from Treasury inflation-protected securities (TIPS) rose by only nine. This is not simply higher expected inflation. Investors want more pay to hold long-term American debt.
History puts the level in context. The long-run ACM average is about 1.35 per cent. The peak was near 5 per cent in 1984. The trough was about minus 1.3 per cent in the summer of 2020. From 2016 to 2024 the premium stayed below zero for eight years - the longest such stretch in the series. Investors paid for the privilege of holding ten years of duration because the Fed was the largest silent buyer. That anomaly is over. The premium is above zero again. Against its own history it is not yet high. Against the past decade it is. What matters is not only the level but the speed of the return above zero.
The yield curve, real rates and inflation expectations
At the end of August the two-year yield was about 4.34 per cent, the 10-year about 4.75 per cent and the 30-year about 5.25 per cent. The 10-year/two-year spread was about 41 basis points; the 30-year/10-year spread about 50. The curve slopes up, but it is not steep. The front and the long end are rising together. Both Fed policy and the term premium are visible in the move.
TIPS give a sharper cut. Real 10-year yields at the end of August were about 2.34 per cent. Ten-year breakeven inflation was about 2.31 per cent. The 30-year real yield was near 3 per cent - the highest since 2010. Almost all of this year’s move in the long end has been in real yields, not in expected inflation. Every distant profit is now divided by a larger number.
Table 1. The curve, real yields and the premium.
| Indicator | Level | As of | What it says |
|---|---|---|---|
| 2-year nominal | 4.39% | 31 Aug | The Fed’s near-term path |
| 10-year nominal | 4.80% | 2 Sep | Forecast plus premium |
| 30-year nominal | 5.27% | 31 Aug | The long end and the deficit |
| 10s-2s spread | 41bp | 31 Aug | The curve is not inverted |
| 30s-10s spread | 47bp | 31 Aug | A modest lift |
| 10-year real (TIPS) | ≈ 2.44% | late Aug | The divisor of distant profit |
| 10-year breakevens | ≈ 2.36% | late Aug | The inflation half is quiet |
| 30-year real | ≈ 2.99% | 31 Aug | A high since 2010 |
| ACM 10-year premium | ≈ 0.73-0.79% | late Aug | The surcharge for time |
Who holds US debt - and why Japan matters
The term premium rises when the supply of government debt grows, when demand for it fades, or when both happen at once. The Fed no longer buys bonds as it did in the years of quantitative easing. The Treasury still funds the deficit with new issuance. Foreign central banks and sovereign funds are no longer a reliable bid.
In June Japan cut its holdings of US Treasuries by more than $26bn - then the largest monthly reduction by a single country in the available data. The sales were in support of the yen. The dollar is again near ¥160. The currency stays firm while others adjust to higher US yields.
A 3 per cent yield on Japanese 10-year government bonds matters beyond Tokyo. When the local risk-free rate is no longer pinned near zero, some of the capital that once hunted yield abroad stays at home. Supporting the yen also means selling US Treasuries. This year’s convulsions in Japanese bonds and foreign exchange have already lifted yields in the United States, Britain and Germany.
Credit spreads are not signalling a slowdown
When markets fear weak growth, premia on corporate debt widen. At the end of August they had not. The US high-yield option-adjusted spread was about 2.60 per cent - tight for that market, close to the cycle low. Investment-grade spreads were tight as well. CCC-rated paper is a different story: the spread there is about 10 per cent and is behaving unlike the rest of the market. Credit is calm. Its riskiest corner is not. What is visible is a repricing of money, assets and investment - not a wave of defaults.
Sectors tell the same tale. Energy stocks rose with oil. Utilities, which live on borrowed money at one end and regulated returns at the other, led the decline. Firms with solid foundations in the same sector did not.
Table 2. Credit, volatility and currencies, late August to 2 September 2026.
| Market | Level | Reading |
|---|---|---|
| HY OAS (ICE BofA) | ≈ 2.60% | Tight; not a growth shock |
| CCC OAS | ≈ 10.3% | The riskiest sleeve is diverging |
| VIX | ≈ 16.3 | Equities have paid a little; this is not panic |
| Brent | ≈ $94-95 | The oil price |
| EUR/USD | ≈ 1.16 | The euro is holding |
| USD/JPY | ≈ 160 | A channel into US debt |
Kevin Warsh, Jackson Hole and Friday’s report
At Jackson Hole Kevin Warsh did not only say he would talk less. He said inflation had not improved much, even though the summer prints were softer; that the labour market was consistent with full employment; and that financial conditions might not be cooling the economy.
The last of those matters most. The Fed’s target range is 3.50-3.75 per cent. The 10-year yield is near 4.80 per cent. Corporate credit is cheap. Equities are holding their prices. If demand is not cooling on those terms, the Fed has no reason to sit still merely because the market had marked down the odds of a rise. His point had force: the committee needs to be confident that underlying inflation is moving towards the target clearly and at sufficient speed. Otherwise there is work to do.
Friday’s employment report, on 4 September at 15:30 Bulgarian time, will settle the argument. In July the US economy lost 23,000 jobs after a revised gain of only 20,000 in June. Unemployment is 4.1 per cent. Initial claims sit near a 10-year low. The Fed chair speaks of full employment.
If Friday’s numbers are weak, the 15-16 September meeting may pass without a change despite Mr Warsh’s words. If they show a tight labour market with oil above $90, the chance of a rise will no longer live only in the futures strip. After the report the Fed enters its quiet period before the meeting. The inflation data will then do more to decide the call.
Stock-bond correlation: when 60/40 works, and when it does not
The correlation between equities and bonds does not protect a portfolio in every regime.
When the two assets are said to “move together”, what is usually meant is their prices. Equities fall. Bond prices fall as well. Total returns are negative. Told through yields the same fact looks different: prices fall, but bond yields rise, so the running loss is cushioned by income. These are not two scenarios. They are one fact, narrated through price or through the rate.
The classic 60/40 portfolio works in a particular sequence. Equities cheapen because earnings deteriorate. The central bank cuts. Long yields fall. Bond prices rise and offset some of the equity loss. That was, more or less, the world between 2000 and 2020: low inflation, room to ease, US Treasuries as a hedge. Those were good years for 60/40.
When the dominant fear is inflation, the hedge fails. The same shock hits both sleeves. In 2022 the broad US equity index lost about 18 per cent in total return. The broad bond index lost about 13 per cent. Long bonds did worse. Instead of protecting the book, duration added to the loss. The rolling 24-month correlation between equities and Treasury returns reached plus 0.50 in mid-2023.
In short: when the scare is growth, the bond works. When the scare is inflation, it mostly does not.
When core inflation sits lastingly above about 3 per cent, the two assets tend to suffer together. Below that threshold, growth fears return. Today TIPS breakevens are near 2.3 per cent. Printed inflation is higher. The term premium and the oil price lean towards an inflation regime. Softer summer price data lean the other way. The market is at a fork and has not chosen.
Maturity decides whether debt protects the portfolio - or taxes it
A short-dated bond almost never moves in lockstep with equity prices. That is why it is not used as a hedge. A long bond is another matter. If the market fears a shortfall of growth and the Fed is ready to cut, its price rises and trims the equity loss. If the dominant fear is inflation, the same duration works in reverse: yields rise, prices fall and the loss is doubled. The phrase “bonds no longer protect us” is empty until the maturity is named.
The term premium and the correlation measure different things. The premium is the extra an investor demands to hold a long government bond rather than a string of bills. The correlation shows whether that position pays when equities fall. Today the premium is back above zero: long debt is no longer cheap. The leading risks are the price of energy and uncertainty about the Fed’s next step. In that setting a long bond costs more as insurance and helps less if the blow comes from inflation rather than from growth.
The state funds itself dearly. Companies postpone the bill
On the same day the long end of the curve asks for a surcharge, the leading corporate story is different. The year is heading for a record in zero-coupon convertible bonds. Companies pay with a claim on the eventual rise in their own shares in order not to pay interest now. The state has no such exit. It cannot compensate its creditor with a share of future growth.
The large technology groups are still putting capital into chips, electric power and data centres. Nvidia has announced a $3.5bn investment in MediaTek. The wager on artificial intelligence is not over. What has changed is the denominator: high long-term and high real yields make every receipt that arrives far in the future more expensive. The convertible merely defers that bill.
Europe and the ECB: this is not 2011, and the address has moved
Long yields in France and Germany are at 15-year highs. For much of the summer France was more expensive to fund than Italy. The political premium is no longer confined to the periphery. It has appeared in the core: budgets, elections and fatigue with debt.
The European Central Bank may yet raise rates as soon as next week. Forward gas prices are higher than they were in winter. There is trouble, too, with supplies from Qatar. Europe is taking the blow on three lines: the price of gas, the passage of budgets, and a central bank that may tighten at the wrong moment.
Three older episodes
The picture resembles 1973. The mechanism resembles 2013. The risk is that the Fed surprises as it did in 1994, against an oil shock of the 2022 sort.
Table 3. Historical analogies are tools, not costumes.
| Episode | What rhymes | What does not |
|---|---|---|
| 2013 taper tantrum | The premium rose after a change in communication | There was no oil above $91 |
| 1994 Fed | The odds of a move had been marked down | A different funds-rate range; no energy shock |
| 1990 oil | A sharp rise that could fade | Today the Fed is not about to cut |
| 2022 energy | Energy lifts yields | Equities today have barely paid |
| 1973 | Only if Hormuz stays shut for long | Inflation is not in double digits |
| Japan 1996 | The 10-year JGB at 3 per cent | Then prices were falling; today they are rising |
In 2013 Ben Bernanke merely hinted that the Fed would slow its bond purchases. Front-end yields barely moved. Long yields rose because the term premium rose. The episode remains a lesson in how to speak to the market. Today Mr Warsh is withdrawing guidance. The difference is the backdrop. Then there was no oil above $91.
In 1994 the question was what the Fed thought. The economy was thought to be moving normally. Alan Greenspan raised rates faster than futures had implied. In August this year the probability of a September rise was about 35 per cent. After Jackson Hole it passed 60. The same species of surprise; another time. The target range is already above 3.50 per cent, and the higher coupon softens the mark-to-market hit on the bond.
In 1990 oil was the scarce good: a sharp rise in the price, then restored supply, then inflation that did not settle in. Today’s quotes only partly repeat that channel. In an oil shock, the length of the interruption weighs more on the market than the first jump in the price.
Japan is not a footnote. A 3 per cent 10-year JGB yield is an exit from a regime that had come to look permanent. In 1996 Tokyo was heading into falling prices. Today prices are rising again.
For a 60/40 book the lesson is not that the mix is dead. It is that the bond protects against a growth scare and not against inflation. The curve says this is not an ordinary summer.
Venetian merchants did not fail because of storms at sea. They failed when the premium no longer covered the next voyage and the chart still had unexplored blanks. Today’s premium on long debt is not yet of that kind. It is the first in years that is visible again - after a stretch in which the Fed, not the market, set the terms.
The question is not whether there is a storm in Hormuz. The question is how long it lasts, and by how much it will raise the cost of money.
Figures in this analysis run to 31 August and the morning of 1 September 2026. The commentary and charts are current as of the afternoon of 2 September and may be revised in the session. The term premium is a model estimate, not a traded price. Correlations depend on the window and on whether they are measured in prices or in yields. This is analysis, not a trade recommendation.
Frequently asked questions
What is the term premium, and how does it differ from the expected Fed funds rate?
The term premium is the extra compensation an investor demands to hold a long government bond rather than a sequence of short ones. It is not a forecast of what the Fed will do over the next two years. The 10-year yield is the sum of expected short rates and that premium. At the end of August 2026 the New York Fed’s ACM model put expected short rates at about 3.98 per cent and the premium at about 0.79 per cent. The premium is not quoted on an exchange. It is inferred. The estimates differ in size, not in direction.
Why is the 10-year yield rising if inflation expectations have barely moved?
Because most of the move since June is not in expected inflation. The nominal 10-year rose by about 36 basis points; TIPS breakevens by about nine. Investors want more pay for uncertainty about Fed policy after Jackson Hole, for oil risk at Hormuz, for the supply of US debt and for a Japan that is no longer an unconditional buyer. The 10-year real yield is about 2.34 per cent. That makes every profit after 2030 more expensive.
Does the 60/40 portfolio still work under this stock-bond correlation?
It works when the dominant fear is growth: equities fall, the Fed eases, bond prices rise. It works poorly when the dominant fear is inflation: the same shock knocks both. Since 2022 the one-year correlation of returns sat near plus 0.24 at the end of August. On Monday the sign was the same; the equity move was small. A short bond barely insures. A long bond insures hard in a growth shock and hurts hard in an inflation shock. Sixty-forty is not dead. Its insurance is conditional.
What should be watched on Friday, 4 September?
The US employment report at 15:30 Bulgarian time. In July the economy lost 23,000 jobs after plus 20,000 in June. Unemployment is 4.1 per cent. If the print is weak, oil falls and the term premium shrinks, the episode looks more like a reaction to the change in communication. If the print is weak, oil falls and the premium holds, the story is debt supply and Japan. If the labour market stays tight with Brent above $90, both halves of the yield can rise together before the Fed meeting on 15-16 September.
Important notice
This article is for information and education only. It is not financial advice, an investment recommendation or professional advice of any kind. The views expressed are the author’s personal opinions and are not recommendations to buy, sell or hold securities, assets or investments. Investing involves substantial risk, including the possible loss of all capital. Past performance is not a guarantee of future results. Each reader should conduct independent research, consult a qualified financial adviser and weigh financial circumstances, risk tolerance and investment objectives before taking any decision. The author and borsite.net accept no liability for any loss arising from the use of the information in this article.