Two Hands on One Curve

Monetary policy · The long end · September 2026

Two Hands on One Curve

The story going around says Kevin Warsh and Scott Bessent are running a coordinated good-cop, bad-cop play against the diesel shock. The record says they are pulling the same yield curve in opposite directions — and on Thursday the bond market scored the contest.

• John Rector
• September 25, 2026
• Essay · about 11 minutes

Contents

1. The story going around
2. What Warsh actually did
3. What Bessent actually did
4. Where they part
5. The diesel underneath
6. Whose expectation is it?
7. The ledger
8. What I left out

01The story going around

There is a tidy account of the last five weeks circulating in newsletters and AI summaries, and it is worth stating fairly before taking it apart. It goes like this. Warsh and Bessent are both old market hands and old friends. They will not fight like bureaucrats. Instead they have split the work: Warsh plays bad cop, raising rates so no one doubts the Fed’s resolve on inflation, while Bessent plays good cop at the Treasury, buying back long bonds to keep the plumbing from bursting under Warsh’s hikes. Together they are holding the line until the energy shock passes and diesel falls on its own.

It is a satisfying story. It has two competent men, a shared enemy and a plan. It also asks you to believe that the chair of an independent central bank and the Treasury Secretary are coordinating monetary and debt policy, which, if it were true, would be a bigger story than the energy shock.

The record does not show a plan. It shows two men with opposite theories of what the long end of the yield curve is for, acting on those theories in the same five weeks, and a market that has had to price both at once.

02What Warsh actually did

At the press conference Warsh said inflation “is too high and has been for too long,” and that the summer’s softer readings did not convince him the underlying trend had improved. Sixteen of the eighteen officials who submitted projections now expect at least one more hike this year. Warsh, as in June, submitted no projection of his own.

That absence is the key to the man. Warsh has spent his first four months as chair taking forward guidance apart. He wants market prices as unfiltered as possible, and he wants the 10-year yield to be a reading the Fed takes, not a dial the Fed sets. At Jackson Hole on August 28 he described himself as “committed to a discipline, not to a decision.” In that same speech, nine days after the Treasury’s buyback announcement, he did not mention the buybacks at all.

A chair who wants the bond market to speak for itself does not need a partner at Treasury quieting it. He needs the opposite.

03What Bessent actually did

On August 18 the 30-year yield touched about 5.34 percent, a 19-year high. The next day Treasury announced it would at least double its liquidity-support buybacks in the 10-to-20-year and 20-to-30-year sectors, from $2 billion to at least $4 billion per operation, running September 9 through November 4. Yields fell hard on the news. Two days later Bessent went on CNBC to say the operations could run above $4 billion and that his department had a big toolkit, and argued that long yields did not reflect the underlying fundamentals of the Iran conflict.

Read that last claim carefully. It is not a plumbing argument. It is a statement that the market’s price is wrong and the Treasury intends to correct it. Treasury’s formal language was about liquidity. The Secretary’s own words were about the level.

The mechanism matters too, because the popular account gets it backwards. A buyback does not “force cash back into the financial system” in any net sense. Treasury pays for old long bonds with money it raises by selling new debt, usually shorter debt. As Citi’s head of USD swaps trading, Dan Gottlander, put it at the time, “It does not change deficits, obviously.” What it changes is the mix: less thirty-year paper in the market, more bills. That is a bet on the shape of the curve, placed by the borrower.

On September 9 Treasury sized its first enlarged operation at up to $6 billion, triple the old $2 billion maximum, and bought less than the maximum. It did not halt the selloff. By September 24 the long end was higher than the day the program was announced.

Figure 01 Five weeks of the long end

1. Aug 1830-year touches about 5.34%, a 19-year high.
2. Aug 19Treasury at least doubles long-end buybacks, $2B to $4B+ per operation. Yields fall. Public debt passes $40 trillion the same day.
3. Aug 20Bessent: operations could exceed $4B; yields do not reflect fundamentals.
4. Aug 28Warsh at Jackson Hole signals the Fed has work to do on inflation. Short yields jump. Buybacks go unmentioned.
5. Sep 9First enlarged buyback, up to $6B. The selloff continues.
6. Sep 16FOMC hikes 12-0 to 3.75-4.00%. 16 of 18 officials see another hike.
7. Sep 21Bessent declines to say whether he agrees with the hike.
8. Sep 2410-year touches 5.22% (highest since 2007); 30-year 5.50% (highest since 2004).

04Where they part

The strongest line in the tag-team story is Bessent’s appearance on CNBC’s Squawk Box on September 21, where he said he had great confidence in Warsh and that the president did too. The story reads this as political cover for Warsh’s independence. Here is the sentence that came first, from the transcript: “I’m not going to talk about whether I agree with it or not.”

Then he made the case against it. Core inflation, he argued, has been “very quiescent,” and the problem is all in the headline number. That is the textbook argument for a central bank to look through an energy shock rather than hike into it. The confidence he expressed was confidence that Warsh would balance the inflation mandate against the growth mandate — which is to say, confidence that the next decision might go the other way. It was a vote for the man and a quiet vote against the decision.

The Treasury Secretary did not provide cover for the hike. He declined to endorse it and then made the case for not doing it.

Nor was executive pressure neutralized. The president attacked the decision and told reporters Warsh had “a very tough board.” His trade adviser Peter Navarro called the hike a monumental mistake.

People who have sat in these chairs read it the same way. Eric Rosengren, the former Boston Fed president, told CNN “The Fed and Treasury are working at cross purposes.” Stanley Druckenmiller — Bessent’s own mentor — used a Wall Street Journal op-ed titled “Let the Bond Market Speak” to call the buybacks artificial yield suppression. Axios summed up the policy mix as contradictory: a Fed trying to guide markets less while Treasury guides them more.

Figure 02 Two theories of the same curve

Question	Warsh (Fed)	Bessent (Treasury)
What is the 10-year yield?	A reading. The market’s unfiltered verdict on growth, inflation and risk.	A problem. A level that does not reflect fundamentals and can be corrected.
Main tool this quarter	The policy rate: up a quarter point, with more signalled by the committee.	Debt management: buybacks of long bonds, funded by new issuance.
Diagnosis of inflation	Too high for too long; summer relief not convincing.	Core quiet; the trouble is energy in the headline.
Public stance on the other	Did not mention the buybacks at Jackson Hole.	Great confidence in the chair; declined to endorse the hike.
Guidance	As little as possible. No personal dot.	As much as possible. “Big toolkit,” amounts left open.

05The diesel underneath

Both men do agree on one thing, and it is the most useful thing in the original story: this is an energy supply shock. The August CPI makes the case in three numbers. Headline prices rose 3.4 percent from a year earlier. Excluding food and energy, 2.4 percent. Energy alone, 16.3 percent.

Figure 03 August CPI, year over year

Core (ex food & energy)2.4%

Headline3.4%

Energy16.3%

Diesel is the sharpest edge of it. The national average hit a record $6.51 a gallon on September 21, per AAA, against about $3.69 a year earlier. Diesel moves every truck, so it seeps into the price of nearly everything, slowly. The supply behind it is thin for reasons no central bank can reach: the war with Iran has cut Gulf refining, Russia banned diesel exports in July and has since extended the ban through year-end, and China spent the spring holding its own refined fuel at home. That last part is the subject of the follow-on piece.

So they share a diagnosis and split on the prescription. Bessent’s reading of a supply shock is: look through it, and lean on the long end so borrowing costs do not compound the damage. Warsh’s reading is: inflation has now run above target for five years, and a Fed that looks through a sixth shock teaches the market that it always will. Both are respectable positions. They cannot both be the plan.

06Whose expectation is it?

Here is where the Reality Equation earns its keep, because the whole dispute is about an expectation and who gets to move it.

Start by naming the Actualizer. It is not Warsh and it is not Bessent. It is the bond market — an institution with its own Reality, the ratio of what arrives against what it expected. Its Expectation is written into every yield: a forecast of inflation, of deficits, of who will still be buying thirty years from now. Neither man can touch that denominator. No one can. The only road to it runs through the record: an action leaves an artifact, the artifact becomes part of the Actual, and only Actuals retrain what the market expects.

Warsh’s method is built on that road. He acts, says as little as possible, and lets the artifacts — a hike, a unanimous vote, a dot plot leaning hawkish, and then the inflation prints that follow — do the teaching. It is slow, and it is the only method that has ever worked.

Bessent’s buybacks are also actions, and they also leave artifacts. But the stated purpose — to show that yields do not reflect fundamentals — is an attempt to move the market’s expectation by telling it what to expect. The market can only learn from what was actually left in the record, and what was left was a $4 billion purchase against a $40 trillion debt, followed by a Secretary promising more without saying how much. The lesson the market drew from that artifact was not “yields are too high.” It was closer to “Treasury is worried.” Five weeks later the long end is higher than when he started.

This is the same pattern I described in The Conservation of Surprise: you cannot remove surprise from a system by managing the expectation, only move it around in time. Suppress the long yield today and the surprise does not vanish. It waits for the next auction — which is roughly what happened on Wednesday, when a weak five-year sale helped send the 10-year to its biggest one-day jump since April 2025.

07The ledger

Already true

The Fed hiked unanimously and signalled more. Treasury expanded long-end buybacks and talked down yields. The Treasury Secretary publicly declined to endorse the hike and argued the inflation was in energy. Long yields are at multi-decade highs anyway.

What the tag-team story needs

Private coordination that neither man has disclosed, and a bond market that calms once the October hike and the remaining buybacks land. If the 10-year falls back well under 5 percent by early November with the Fed still tightening, the good-cop, bad-cop reading earns a point.

Where I am probably wrong

I cannot see private conversations, and these two have known each other for decades. Warsh has also called for a new accord with Treasury that coordinates balance-sheet policy, so a formal division of labour may be coming. If one is announced, I will have described the opening weeks of a partnership as a quarrel. I would still say the first five weeks, as played in public, were at cross purposes — and that the market priced them that way.

For anyone outside a trading desk, the practical lesson is simpler than any of this. Watch the 10-year, not the speeches. Mortgage rates and business credit follow it. On Thursday it said neither man is in control of it, and that is the honest state of play.

08What I left out

Sources

1. Federal Reserve — FOMC press conference transcript, Sept 16, 2026
2. Federal Reserve — Warsh, Jackson Hole remarks, Aug 28, 2026
3. U.S. Treasury — buyback operation announcement, Aug 19, 2026
4. CNBC — Fed rate decision, Sept 16, 2026
5. The Hill — Bessent says he has great confidence in Warsh
6. Grabien — Bessent Squawk Box transcript, Sept 21, 2026
7. Benzinga — Bessent on core vs headline inflation
8. Reuters via Yahoo Finance — Treasury doubles long-bond buybacks
9. CNBC — Bessent: buybacks could exceed $4 billion
10. Bloomberg — Upsized buybacks hit by bond market reality
11. CNN — Kevin Warsh has a plan for the Fed. Scott Bessent is getting in the way
12. Axios — Fed and Treasury appear at odds
13. CNBC — Warsh warns on inflation at Jackson Hole
14. BLS — Consumer Price Index, August 2026
15. WTTW — Diesel hits all-time high, Sept 21, 2026
16. The Conversation — Iran war pushes diesel to record highs
17. CNBC — 30-year yield hits highest since 2004, Sept 24, 2026
18. Yahoo Finance — Trump criticizes the Fed hike