Credit Still Has Not Blinked
The gap
Treasuries keep repricing. The 10yr closed at 5.01% (as of 16 Sep), a 100th percentile level against its own 3yr history. The 5yr is at 4.86% (as of 16 Sep) and the 2yr at 4.74% (as of 16 Sep), up 7bp on the day.
High yield OAS is 2.76% (as of 15 Sep), 1 basis point wider than the prior session and unchanged in any meaningful sense. Investment grade OAS is 0.80% (as of 15 Sep) and has not moved in 2 sessions.
The all-in cost of money has risen sharply. The price of default risk has not.
What the listed market says
Credit ETFs are drifting, not breaking. HYG at 78.42 is down 0.56 over the week; JNK at 94.39 is down 0.65; LQD at 104.45 is down 0.86. Those are duration marks, not default marks: a credit market pricing distress falls faster and wider than this.
Equities are the ones carrying the stress. The S&P at 7,551.81 is down 33.92 on the session and 84.55 on the week, with the drawdown concentrated in the growth complex - QQQ down 11.59 on the week while the equal-weight economy has held up far better.
So the stress is being expressed in the equity risk premium and in the long end. It has not reached the corporate borrower.
Why the plumbing makes this unstable
Liquidity is the regime. The composite reads NEUTRAL at -0.266 with Liquidity the dominant sub-index at -2.247, the weakest of the five. Reserves are 2,991bn (as of 9 Sep), up 68bn from 2,894bn (as of 2 Sep). RRP is 5.4bn (as of 16 Sep).
The headline build in reserves sits alongside something that looks like its opposite. Treasury's balances at the Fed collapsed over the same fortnight: deposits other than the general account fell to 844bn (as of 9 Sep) from 944bn (as of 2 Sep), and the general account fell to 883bn (as of 9 Sep) from 968bn (as of 2 Sep). Both legs of the Treasury's cash at the Fed were drawn down by roughly 100bn in the same window in which reserves rose 68bn.
That is the mechanism, not a coincidence. Reserves are the mirror of the Treasury's Fed balance: when Treasury spends from its account, the money lands in the banking system and reserves rise. The build in reserves was funded by the drawdown in the Treasury's balance, and the Treasury's balance was drawn down by roughly 285bn over the fortnight - more than three times the reserve build it produced. The rest went to the public, where it does not show up as bank reserves.
There is no buffer underneath this. If credit does start to move, the market-making capacity that keeps it orderly is being funded into a system with almost nothing parked at the facility.
The rate-side read is mixed. MOVE at 80.73 is down 2.98 on the day but still 3.99 higher on the week. VIX at 17.71 is up 0.51 and sits in the 14.5th percentile. Rate volatility is the input cost of hedging a credit book, and it has not come down enough to make warehousing cheap.
The borrower side
Claims are low and not deteriorating. Initial claims 206k (as of 5 Sep), continuing claims 1,774k (as of 29 Aug), back to back flat weeks. Core CPI is 3.71% year on year (for August).
That is why credit has not cracked. Default risk is a function of cash flow and unemployment, and neither is signalling anything yet. The macro backdrop says refinancing is expensive, not impossible.
The uncomfortable part is what that means for anyone rolling debt. A borrower refinancing into a 5% 10yr with core inflation at 3.71% is paying a real premium for the privilege. That drag accumulates quietly before it shows up as a spread.
The UK just showed everyone the template
At noon London time the Bank of England held Bank Rate at 3.75% by a 6-3 vote, with Pill, Greene and Mann preferring 4.00%. That is the least important part of the announcement.
The Bank set out a multi-year path to run its monetary-policy gilt holdings to zero, unwinding the remaining stock at an average of 46bn a year to the end of 2034 through 20bn of annual sales alongside maturing gilts. The instrument was 489bn of gilts on 9 September - the same balance-sheet date as our US reserves print, which is a useful accident of timing. It had peaked at 895bn.
The design is the story. Active auctions are paused for 6 months while the Bank reviews a model for selling directly to the Government. Of the remaining stock, 222bn maturing before 2035 is retained to maturity, and 120bn of the longest-dated paper is retained to back banknote issuance. That leaves 146bn maturing between 2035 and 2049 as the pool the options face: the Bank Executive has engaged with the Treasury and the DMO on a model in which the DMO would purchase the gilts the Bank is selling, at market prices and on a pre-announced schedule.
Read that against the arithmetic of the asset. The Bank's year to September cut the stock by 70bn, of which only 21bn was active sales. The sales programme was never the volume driver; maturities were. What the announcement changes is not the pace of the unwind but who absorbs the duration, and at what price.
What the UK actually did
The Bank stopped being the marginal seller of the long end of its own curve. Into a market where 30yr gilt yields had touched 5.93% earlier in the week, the highest since March 1998, the supply of existing long-duration paper from the official sector goes to zero, and it goes to zero for good rather than for one meeting.
It is worth being precise about the direction of the effect. The Bank has ring-fenced 120bn of the longest paper permanently, and retained 222bn more to maturity. Neither number reaches the market. The 146bn in between does, but on a pre-announced calendar and to a single buyer whose instruction comes from the same government that issues the debt. For a pension fund deciding where to be in the 2040-2049 sector, the marginal official seller has been replaced by a price-insensitive official buyer with a published schedule.
The fiscal arithmetic runs the same way. Active sales were crystallising losses on a portfolio bought at prices the market has since repriced; the Bank's own position is that the pace of sales affects the timing of those losses rather than their total, which is true and also not reassuring. The Office for Budget Responsibility has projected that unwinding the whole portfolio costs roughly 100bn over 5 years, and long-dated sales have been estimated at 22bn of taxpayer cost since 2022.
This is what fiscal dominance looks like when it arrives politely. No yield cap, no emergency purchase, no announcement that anyone would call a monetary-financing event. A central bank deciding which maturities the market can absorb, and then arranging for the government to buy the ones it cannot.
The next 2 quarters
The near-term effect is mechanical and should be mildly supportive at the long end. Long-dated supply from the official sector stops for 6 months; the market was already positioned for it. The 30yr has rallied about 8bp from Tuesday's peak, and a halt that is expected still removes a seller that was not expected to stop permanently. Most of the relief is probably already in the price. The part that is not is the 8yr horizon: a long end that knows the official sector is a buyer of last resort until at least 2034 is a long end with a different term premium.
But the relief is a supply story, and supply was never the binding constraint. The DMO's 2026-27 Net Financing Requirement is 251.2bn, revised down 5.9bn, with gross gilt sales of 246.2bn. None of that changed. The Bank stopped one seller; the government did not stop borrowing. October and November carry the heaviest syndication calendar of the year, and the Autumn Budget lands on 28 October with the deficit, the energy shock and the indexation of the long end all still open.
So the honest 2-quarter shape is this. October and November: the long end stays supported into the Budget, with the 30yr likely held in a 5.6-5.9% range unless the Budget delivers a fiscal consolidation that removes the risk premium outright. December and into Q1: the support starts to look like accommodation. The Bank will have paused sales for half a year with inflation near 3.1% and three members voting to hike - a combination that is difficult to defend as a tightening stance, because it is not one. If the long end has not rallied materially by the time the pause is reviewed, the market has told you the supply story was never the story.
The inflation side is the part that cuts against all of this. A central bank that is simultaneously holding rates at 3.75% with a hawkish minority and refraining from selling long paper is signalling that it fears the fiscal reaction more than it fears inflation. That ordering is the thing to watch, because it is repeatable. If it holds, the long end grinds tighter on supply and steeper on risk, and sterling carries the adjustment.
The falsifiable markers for the 2 quarters are these. A 30yr gilt that cannot hold below 5.9% through the Budget means the halt bought nothing. A 30yr through 5.5% means the market believes the official sector has capped the long end, which is a much larger claim than the announcement made. And the 10yr gilt at 5.25% - already the highest borrowing cost in the G7 - against a Bank Rate of 3.75%, is a 150bp term premium that nothing in today's decision addresses.
Why every large economic actor will end up here
The conditions the UK just responded to are not British. A large balance sheet bought at prices the market has repriced; a long end whose natural buyers have thinned; a government borrowing 246bn in a year; and a central bank whose asset sales are visible, attributable, and politically costly in a way that interest-rate policy is not.
Every major central bank that bought duration holds the same structure of problem, and the ones that matter differ only in how far along the curve they are.
The Federal Reserve never sold. It lets paper mature and reinvests the rest, which is why it has never had to defend an active-sales programme and never had to price the political cost of one. The UK bought that cost by being the only major central bank selling actively, and is now the first to stop. Read the announcement as the ceiling on how much active sales any institution will tolerate, not as a UK story.
The European Central Bank is in the quiet part of the same cycle: not selling, running down reinvestments, with a balance sheet that peaked far above its pre-2015 level. It has never had to answer the question the Bank answered today, and it now has a worked example.
The Bank of Japan is the one place where the logic is furthest advanced, and the one place it gets worse. It is the marginal seller of its own duration into a curve at multi-decade highs, at a pace set by a taper plan announced in July 2024, while its domestic institutions have begun to find JGBs cheap enough to buy. It has the same three ingredients as the UK - official supply, thin long-end demand, and a fiscal authority with a large funding need - but it lacks the two valves that make the UK and US versions survivable.
If there is a transferable model in today's decision, it is the shape rather than the numbers: retain the paper, publish a long horizon, change the buyer rather than the pace. That is what a central bank does when it has concluded that the market can no longer absorb the duration it is selling. It is not a policy about inflation. It is a policy about absorption, dressed as one about pace, and it will be copied.
Japan is the same template with fewer valves
Japan is not a fourth example of the pattern. It is the same pattern with two of the three safety valves removed, and that changes the conclusion rather than the mechanism.
The structure matches almost exactly. The Bank of Japan's balance sheet is 6,446,620bn yen (as of Aug 2026). The 10yr JGB is 2.94% (as of Aug 2026), up from 2.79% (as of Jul 2026) and at its highest since 1996 on the daily prints. The taper runs under the July 2024 plan, which makes the Bank the marginal seller of its own duration. Add a domestic bid that has finally started to find JGBs attractive, a fiscal authority with a large funding need, and an official sector that owns an enormous share of the outstanding stock. Those are the UK's three ingredients and Japan has all of them.
The first missing valve is the currency. The UK sells gilts into a domestic buyer base and, when the long end gets expensive, can change the buyer to the DMO and call it a sales model. Sterling is not a constraint, because nobody is defending a level. For Japan the exchange rate is the adjustment mechanism, and it is already moving against them. That is part of why the Bank is expected to hike 25bp to 1.25% at the meeting ending tomorrow - the highest since 1995, and the shortest interval between increases since 1990. The Bank of Japan is tightening into a long end that is already at multi-decade highs, which is the exact opposite of what the Bank of England did today.
The second missing valve is the foreign bid. Japanese institutions are the marginal foreign buyer of US duration, and that is where this stops being a domestic Japanese story. Japan sold roughly 47bn dollars of Treasuries in March 2026, the largest sale of any foreign holder that month, and the fall in its foreign securities holdings in August roughly matched the size of the month's currency intervention. The mechanism is that hedged JGB returns now beat hedged Treasury returns, which creates a repatriation incentive that exists independently of any view on the dollar. If tomorrow's hike is delivered hawkishly, that incentive strengthens, and the marginal foreign bid for the US long end weakens at precisely the moment the Fed has stopped shedding duration and the Treasury is buying it back. Neither of those two can substitute for a real foreign buyer. They only change who holds the paper.
So the shift from arithmetic to politics is sharper in Japan, not softer. The arithmetic there has one answer - you cannot run a taper into a thin domestic bid while hiking rates and defending a currency - and no institution is taking it either. What decides the path is the press conference tomorrow, whether the Ministry of Finance intervenes again and how it funds that, and whether repatriation accelerates or stalls. Each of those is a decision rather than a calculation. The binding constraint is political in the same way as the US and UK, with one addition: in Japan, the politics and the currency are the same variable.
Where this breaks
A widening in HY OAS that accompanies a stabilising long end. That combination would mean the rate move has finally transmitted into default expectations, which is the point at which credit stops being a duration story and becomes a credit story.
Until then the honest read is that the front end, the long end and the corporate market are telling three different stories. Two of them are priced for stress. The third is not. And as of noon London time, the official sector just stopped adding to the bearish case at the long end - which is the one place the three stories had agreed.
There is one more thing to watch before the weekend, and it is not American. The Bank of Japan decides tomorrow. If it hikes and signals more, the marginal foreign bid for US duration gets more expensive on the same day the UK stopped selling its own. The long end is now being set by which sovereign is willing to absorb the duration, not by the price of money. That is a decision, not a calculation, and it is being made in three capitals this week.
| TENOR | LEVEL | 1D Δ | Z (3Y) | RANGE %ILE |
|---|---|---|---|---|
| 2Y | 4.610 | -0.060 | +1.39 | 78 |
| 5Y | 4.780 | -0.070 | +2.24 | 86 |
| 10Y | 4.980 | -0.070 | +2.51 | 84 |
| 30Y | 5.320 | -0.050 | +1.99 | 83 |
Update, 04:16 UTC
The Bank of Japan raised the policy rate to 1.25%, from 1.00%, by a 7-2 vote. The guideline now encourages the uncollateralized overnight call rate to remain at around 1.25 percent. Asada and Sato dissented - Asada preferring a hold, Sato arguing the timing was wrong. It is the highest policy rate since 1995. The decision was widely expected, so the hike itself carries little information.
What carries information is what did not move. No new JGB purchase plan accompanied the decision. The June 2026 meeting published one; September's are the rate change and the mechanical amendment to the Complementary Deposit Facility. The taper announced under the July 2024 plan stands unreduced. So the same institution, on the same day, raised the price of money and remained the marginal seller of its own duration into a 10yr JGB at 2.94%.
That is the template holding. Japan still has no third valve: it is tightening into a thin domestic bid with the currency as the adjustment mechanism rather than a lever it can pull. That is the structure the piece described, and the print did not resolve it.
The statement commits the Bank to continue raising the policy rate and adjust the degree of monetary accommodation, while considering the timing and pace of adjustment. Two dissenters is the only evidence so far that the pace is contested internally. Governor Ueda speaks around 06:30 UTC, and the pace guidance is the part that reprices the marginal foreign bid for US duration.