Oil, Bond Yields and Private Credit: What Would Have to Change for Each to Get Better

• Three markets, and the rule that runs through all three
• Oil: the shortage lives in the barrels, not in the contracts
• Oil: four levers, in descending order
• Oil: what to watch, and the argument on the other side
• Bonds: the long end has stopped listening
• Bonds: three routes to a repair, and what to watch
• Private credit: both the alarming numbers and the reassuring ones are true
• Private credit: five requirements, and a tension that does not resolve
• Private credit: the indicators and the honest state of play
• The chain that connects the three, and eight markers

Three markets are doing badly at the same time — and the useful question is not what went wrong but what would have to happen first for each of them to heal, which indicator would move before the price did, and what the honest state of play actually is.

DeepDives · 2026 · An AI–human collaboration that weighs every source, mainstream or alternative, under strict rules of truth and logic.

§Three markets, and the rule that runs through all three

Three markets are doing badly at the same time, and they are the three this series has followed most closely: oil, government debt, and private credit. The usual episode about a moment like this asks what is wrong. This one asks what it would actually take for each of them to get better, and what would show up first if that began to happen. One principle runs through all three sections: the earliest evidence of improvement does not appear in the price, because prices move on sentiment and positioning while the plumbing moves only when facts change.

For oil, the plumbing is the insurance and freight market, because a barrel cannot move if no underwriter will cover the voyage. For government bonds, it is the term premium and the auction book, because those tell you what investors demand for tying up money for a decade and whether enough of them turned up. For private credit, it is the redemption queue and the marks on the loans, because those tell you whether the money is being returned and whether the valuations can be trusted.

§Oil: the shortage lives in the barrels, not in the contracts

Start with oil, and with the number that matters most, which is not the futures price. On the fourteenth of September Brent crude rose three dollars and twenty-one cents to a hundred and seven dollars and eighty-two cents, and West Texas Intermediate rose three dollars and seventeen cents to a hundred and three dollars and twenty-two. Both are roughly forty-five per cent above where they stood before the war.

The more informative number is the physical one. North Sea Dated, the benchmark for actual cargoes rather than paper contracts, averaged ninety-one dollars a barrel in August and surged to a hundred and thirteen dollars and forty-eight cents on the ninth of September. Physical crude trading well above the futures price is the market saying the shortage lives in the barrels, not in the contracts.

And the shortage is measurable. The International Energy Agency’s September oil market report, published five days before this recording, is the closest thing this market has to an audit. Global supply is set to fall by five point seven million barrels a day this year, to a hundred and one million barrels a day, with more than ten million barrels a day of Gulf production shut in. Oil demand is forecast to fall by two and a half million barrels a day in 2026, an outright decline rather than a slowdown, and the agency cut that forecast by another nine hundred and forty thousand barrels a day this month alone. Inventories have fallen by five hundred and seven million barrels since February, an average draw of two point eight million barrels a day.

Notice where the tightness sits. It is no longer crude alone. American diesel went above two hundred dollars a barrel in early September, ninety-four per cent above its pre-war level, and refining margins in the Atlantic basin reached record levels on diesel cracks. Gulf exports of diesel and gasoil averaged three hundred and ninety thousand barrels a day in August, just over a quarter of their pre-war level, and Gulf and Russian exports together are one point six million barrels a day below February levels, from two regions that supplied nearly forty-five per cent of the world’s seaborne trade in that fuel.

§Oil: four levers, in descending order

The first is restoring the route itself. The Strait of Hormuz remains effectively closed to normal commercial traffic. Ship tracking recorded six commodity vessels transiting on the ninth of September against a pre-crisis baseline of roughly eighty-five a day, and three hundred and thirty-six vessels held position outside the chokepoint rather than risking the passage. War risk insurance covered a transit at about a quarter of one per cent of hull value before the war and now quotes between three and ten per cent, which for a hundred million dollar tanker is a premium of three to ten million dollars for a single voyage.

The second is the bypass, and this is where the past week went backwards. Saudi Arabia’s East-West pipeline, twelve hundred kilometres from the Gulf coast to the Red Sea port of Yanbu, had become the kingdom’s main escape route, carrying between two point six and four million barrels a day since late August. On the tenth and eleventh of September it was hit by drone strikes launched from Iraqi territory and Saudi Arabia shut it down. Officials briefed on the repairs told the Associated Press the work could take three to five weeks. Four million barrels a day is about four per cent of world oil supply, and crude stocks at Yanbu may keep exports flowing for only five to seven days if the pipeline stays shut. In Europe, crude flows from Sidi Kerir to Poland are expected to fall this month to about two point one million barrels from six point six million in August.

The third is the buffer, and this is the part that should worry anyone who assumes the market can absorb another shock. The strategic petroleum reserve has been drawn by about one hundred and twenty-two million barrels since the strait closed, as part of a coordinated release of one hundred and seventy-two million barrels, and the announced releases have ended. OPEC’s effective spare capacity, by the agency’s own estimates, is two hundred and twenty thousand barrels a day across the whole group and just seventy thousand across the eight producers that have targets. The world’s spare capacity is effectively zero: Saudi production runs at just under six million barrels a day against a sustainable capacity of more than twelve million, and the difference is capacity behind the blockade.

Which leaves the fourth lever, the only one operating, and the one nobody wants. When supply cannot rise, price does the work, and price does its work by destroying demand. That is what the agency’s forecast of a decline of two and a half million barrels a day this year is: a description of high prices forcing refineries, factories and households to use less. It is a repair mechanism that shows up as falling output rather than recovery.

§Oil: what to watch, and the argument on the other side

Ranked by how early they move. First, the war risk insurance premium, because it is the gate every other fix has to pass: watch for quotes back below one per cent of hull value, then below half a per cent. Second, tanker economics: freight rates and the availability of ships willing to call at Gulf terminals. Third, the transit count, which needs to climb from single digits and the mid teens back toward several dozen a day. Fourth, the shape of the futures curve. Crude is in extreme backwardation, near-term barrels costing far more than later ones; when it flattens, the paper market is agreeing that relief is coming. Fifth, the inventory series, drawing at two point eight million barrels a day. Sixth, refining margins, where record diesel cracks mark the tightest link.

Measured against those indicators, the last fortnight moved the wrong way on every count. But the counterargument deserves its place, because this market has re-priced violently in both directions this year. Brent peaked at a hundred and twenty-six dollars in March, was at seventy-one dollars and fifty-seven cents by the first of July, back above a hundred by the twenty-third of July, and is at a hundred and eight now, and nothing about the conflict changed for the better across those swings. A single credible agreement, or a single week in which escorts demonstrably work, could take twenty dollars off the price faster than any forecast suggests. The level of the price is not evidence about the future, and the insurance market, not the oil price, is where the question is answered first.

§Bonds: the long end has stopped listening

The ten-year American Treasury yield was five point zero zero per cent on the sixteenth of September, having briefly touched five point zero one one per cent on the Monday. The last time it held above five per cent was October 2023, and the long end sits at levels several desks describe as a twenty-year high. Over the past month the ten-year has risen twenty-seven basis points; over the past year, ninety-one. Nor is this American alone: Britain paid a record five point eight one six eight per cent on a thirty-year gilt sale on the eighth of September, the highest its debt office has ever paid, German bunds are at three and a half per cent, a level last seen in 2011, and Japan’s ten-year reached three per cent in early September for the first time since 1996.

Here is why this confuses anyone who follows central banks. The Federal Reserve has not raised rates at all this year. It cut three times in the autumn of 2025, parked the target range at three point five to three point seven five per cent in December, and has held it since, and over that period the ten-year went from roughly four point two per cent to five. The long end has been ignoring the central bank, and the reason has a name: the term premium. Every long yield is the expected short-term rate over the bond’s life plus a premium for lending long rather than rolling over short bills. That premium was negative for most of the twenty tens, turned positive in late 2024, went above seven tenths of a percentage point in January, and sat near half a point in May, as compensation for supply and for inflation uncertainty. It is also why a rate cut would not automatically pull the long end down.

And the central bank is not cutting. Markets were pricing an eighty to ninety per cent chance of a quarter-point increase at the meeting concluding in the early hours of the seventeenth of September, so treat that as priced expectation rather than fact. What is fact is the inflation it is reacting to: consumer prices rose four tenths of one per cent in August, holding the annual rate at three point four per cent, with core inflation at two point four per cent over the year but three tenths higher over the month. Energy connects the first section of this examination to this one; a diesel price ninety-four per cent above its pre-war level is an inflation input, and the bond market is where it is priced.

§Bonds: three routes to a repair, and what to watch

Three routes could repair this, and only one is under anyone’s control. The first is policy addressing the causes: energy supply normalisation, which is the oil story, and fiscal tightening, which would mean a smaller deficit path. Neither is happening. The second is direct intervention. The Treasury announced on the nineteenth of August that it would at least double its buyback of longer-dated debt, from two billion dollars per operation to at least four billion, and yields fell on the announcement. But buybacks are more signal than substance: even doubled, they are absorbed into the flow of new supply. The powerful version is quantitative easing, and that is unlikely now, because the chairman of the Federal Reserve, Kevin Warsh, wants the balance sheet smaller. The third route is economic conditions: softer inflation or a weakening labour market would let the central bank ease.

Ranked by how early and how honestly they tell you the market is healing, the indicators begin with the term premium itself, published monthly by the Federal Reserve Bank of New York. It is slow, but it is the only measure that separates relief from the two very different reasons yields fall. Yields can fall because inflation expectations fall, which is a good repair, or because growth expectations collapse, which is a recession arriving, and the two look identical in the headline price. A rally led by the front end, with the two-year falling faster than the thirty-year, is the market pricing easier money; a rally led by the long end is the market pricing less inflation risk and less supply risk. The current move has been led by the long end in the wrong direction, and the repair would look like that reversing.

The third indicator is the one already giving a positive answer. The auction book. On the ninth of September the ten-year note auction cleared at a yield of four point eight three four per cent with a bid-to-cover ratio of two point seven one, and indirect bidders, mostly foreign institutions, took seventy-nine point two per cent of it. The thirty-year auction saw a bid-to-cover of two point six one two, its strongest reading since February. Set against the structural story of plateaued foreign buying, that is a genuine counterweight: at these yields, buyers are turning up, and in size. Then watch the Treasury’s issuance choices, because shifting the mix toward shorter maturities reduces the duration the market must absorb, and the mortgage market, where the thirty-year mortgage is around six point eight per cent and high long rates bite first by freezing transactions.

The reality is genuinely mixed. On one side the plumbing has held: auctions are clearing with strong demand, the Treasury is buying back long bonds, and a five per cent ten-year has not produced a funding accident. On the other, none of the structural drivers has improved. Supply is set to stay heavy and foreign demand is flat, the term premium is positive and has not fallen, and the central bank is expected to tighten into it rather than ease. As one veteran investor put it, five per cent does not break anything on the day it arrives; it breaks things twelve to eighteen months out, when the refinancing has to happen at the new rate. The maturity wall was moved, not removed, and the borrowers closest to it borrowed when money was cheap, in a market where the marks are set by the lender rather than by a price on a screen.

§Private credit: both the alarming numbers and the reassuring ones are true

Start with the disturbing numbers, all from the second quarter of this year. Fitch, applying a broad definition of default, put the American private credit default rate at six point one per cent in July, a record. The retail vehicles, the non-traded business development companies that ordinary investors can buy, received redemption requests above their five per cent quarterly caps at almost every large manager: Golub at eight and a half per cent of shares, Blackstone’s BCRED at around ten, BlackRock’s fund crossing its cap for the first time since it opened in 2022, Oaktree at six point eight, Morgan Stanley’s North Haven fund at ten point nine per cent, and the Cliffwater Corporate Lending Fund, a thirty-three billion dollar interval fund, at fourteen per cent against a seven per cent cap. On the third of September Blackstone capped redemptions at BCRED for a second consecutive quarter, and in the first quarter Blue Owl closed redemptions on one fund entirely, with a loan marked to near zero. New money has stopped arriving too: non-traded fundraising was two billion dollars in the second quarter, down eighty-two per cent from a year earlier.

Now the reassuring numbers. The funds paid. Non-traded business development companies returned about five point nine billion dollars to investors in the second quarter and more than twelve point seven billion in the year to date. The caps were honoured and the excess was prorated, which means the mechanism did what the offering documents always said it would: meet the limit, queue the rest. That is an orderly gate, not a frozen fund. Early third quarter data points the same way: the first three net asset value funds to report showed combined requests of four point six per cent of net asset value against seven point nine per cent in the second quarter, and Golub’s July tender met its requests in full.

One piece of analysis reconciles the enormous request figures with the modest real appetite to leave. When a fund fills only part of a request, the investor who wanted out gets half and stays in the queue for the rest, so the rational response is to ask for twice what you actually want next quarter. Multiply that across thousands of advisers and the headline request number inflates well beyond the genuine desire to exit. Bank of America’s analysts called this the proration reflex in April and predicted requests would peak in the second quarter and then fade. The single most useful private credit number in the third quarter is not the size of the requests but whether they fall below the caps.

§Private credit: five requirements, and a tension that does not resolve

The first requirement is honest marks. You cannot repair a price that is not being set. The loans are not traded daily, the valuations are produced by the manager, and regulators on both sides of the Atlantic have spent this year saying so. The Financial Stability Board’s May report flagged valuation opacity, inconsistent definitions of default, leverage at both fund and company level, and the mismatch between quarterly withdrawals and loans that cannot be sold quickly. The Office of the Comptroller of the Currency made private credit a discrete area of supervisory focus for the first time this spring. A write-down is not a failure of the system; it is the system producing information.

The second is that the queue clears, meaning requests below the caps for more than one quarter: one quarter is a report, two is a trend. The third is price discovery in the secondary market, because a discount is a market saying what it thinks the marks are worth. Publicly listed business development companies trade at roughly eighty per cent of net asset value, an average discount of about a fifth, and secondary buyers of private fund stakes have been offering twenty to thirty-five per cent below net asset value for immediate cash. When those discounts narrow, the marks are being validated.

The fourth is the return of new money, the slowest clock of the four, and it is not simply that fundraising fell but where the money went. Capital leaving these vehicles is not rotating into other private credit vehicles; it is going to hard assets, infrastructure and real estate, and the industry’s share of new fundraising for direct lending has fallen from more than half in 2024 to around a third. The fifth is the restoration of the premium, where the arithmetic has been unforgiving. Private credit historically paid about one point six percentage points more than comparable public loans as compensation for illiquidity, and one research house calculates that an investor now needs a commitment of seven to ten years to earn a two per cent premium over public debt markets.

That last point produces a tension that does not resolve neatly. The bond market’s repair requires lower inflation or weaker growth, and the central bank is expected to tighten rather than ease. But every borrower in a private credit portfolio pays a floating rate, so their interest burden falls when rates fall. The bond market wants higher for longer to break inflation; the private credit market needs lower to restore coverage, and both positions are held by the same investors.

§Private credit: the indicators and the honest state of play

Ranked by how early they move. First, the discount to net asset value on listed business development companies, a real-time market price rather than a manager’s opinion. Second, the quarterly tender results from the non-traded funds: the request level against the cap, and whether the cap is paid in full. Third, the marks in the quarterly net asset value reports, and the share of income coming as payment in kind rather than cash. At the largest exchange-traded business development companies that share fell to eight point two per cent of interest income in the first quarter, its lowest in two years, while across the nine largest non-traded funds payment-in-kind income rose by roughly forty-two per cent year on year. Fourth, non-accruals. Fifth, fundraising, the slowest. Sixth, the bank exposure underneath, where the six largest American banks hold an estimated three hundred to three hundred and twenty-two billion dollars of committed lending to private equity and private credit sponsors, against about ten billion in 2013.

Private credit is not collapsing, and the claim that the funds are refusing to return people’s money is half wrong: they pay the cap and queue the rest. But nothing that would constitute a repair has happened yet. Requests are still above the caps at the largest managers, fundraising is at a multi-year low, defaults are at a record on Fitch’s broad measure, payment-in-kind income is rising sharply in the very vehicles retail investors hold, and the structural mismatch between quarterly withdrawals and illiquid loans is untouched. The industry’s most quotable admission came from the chief executive of Carlyle, who said the business did itself a disservice by calling the vehicles semi-liquid, when they should have been called sometimes not liquid at all.

Then the counterargument, held seriously. Default rates depend on definitions, and the range this year is extraordinary: Fitch at six per cent on its broad measure, Moody’s somewhere between one point six and four point seven, and Proskauer’s index of senior secured loans at two point five one per cent. Much of the stress is in the wrapper rather than the underlying credit, and the International Monetary Fund’s assessment in April was that liquidity mismatches were largely confined to the semi-liquid vehicles while default rates in the two to three per cent range were manageable. The sectors genuinely under pressure are identifiable rather than universal: software and software-like businesses exposed to artificial intelligence, which in some portfolios are a quarter to nearly a third of all loans. The fair reading is a real credit cycle with real losses in identifiable places, wrapped in a structurally mismatched retail vehicle.

§The chain that connects the three, and eight markers

Bring the three together and the pattern is the answer to the question this examination started with. In all three, the price is the noisiest signal and the plumbing is the honest one: the insurance market for oil, the term premium and the auction book for bonds, the redemption queue and the valuations for credit. And the three are connected by one chain, not three separate stories. Energy prices feed inflation, inflation keeps the central bank tightening and the term premium elevated, and elevated long rates set the price at which debt raised in the cheap years must be refinanced, twelve to eighteen months out rather than today. The borrowers most exposed to that repricing are in private credit, which is why a repair there is unlikely before one in oil.

What this series will be checking. If the war risk insurance premium for a Hormuz transit is still above one per cent of hull value at the end of October, the diplomatic and military path is not working. If Brent is below ninety dollars at the end of October without a repaired East-West pipeline or transits back above twenty a day, the price is set by demand destruction rather than returning supply. If the ten-year is still above five per cent at the end of the year, the term premium has not repaired; below four and a half per cent with inflation expectations falling, it has. If the central bank raises rates at this meeting and the ten-year falls over the following fortnight, the long end is driven by supply and risk rather than policy expectations. If non-traded fundraising is still below five billion dollars in the fourth quarter, the demand side has not returned. If any large semi-liquid fund fails to pay its stated cap in full once requests exceed it, that is the first genuine failure of the mechanism rather than a test of it. If payment-in-kind income at the non-traded funds is still growing at forty per cent a year after another two quarters, credit quality is deteriorating faster than the marks admit. And if listed business development companies trade back above ninety per cent of net asset value, the market is validating the valuations rather than disputing them.

Eight markers, each one checkable, several of them dated within six weeks. That is the honest way to hold a view on three markets doing badly at once: not a forecast of when they get better, but a list of what has to happen first, and the willingness to be told you were wrong by the data.

This has been a deep dive into oil, bond yields and private credit, and what would have to change for each of them to get better.

Sources: this examination names its sources inline — the September oil market report and capacity estimates of the International Energy Agency; the ship-tracking, insurance and maritime position data for the Strait of Hormuz; the Associated Press reporting on the drone strikes against the East-West pipeline and the repair estimate that followed; the sovereign yield data for American Treasuries, British gilts, German bunds and Japanese government bonds, including the record thirty-year gilt sale; the Treasury buyback announcement of the nineteenth of August; the ten-year and thirty-year auction results of the ninth of September; the term premium series of the Federal Reserve Bank of New York; the default-rate measures of Fitch, Moody’s and Proskauer read side by side; the redemption request, cap and tender figures reported by Golub, Blackstone, BlackRock, Oaktree, Morgan Stanley, Cliffwater and Blue Owl; the fundraising and non-accrual data for non-traded business development companies; the May valuation findings of the Financial Stability Board and the supervisory expansions of the Office of the Comptroller of the Currency and the Federal Reserve; the April assessment of the International Monetary Fund; the proration-reflex argument of Bank of America’s analysts; and the remarks of the chief executive of Carlyle on what these vehicles should have been called. Every source is weighed with the same skepticism regardless of politics or business model. The audio version of this article, read by DeepDives, is available on Wavlake.