The Fed Went Quiet. The Treasury Did Not.
Chair Warsh used Jackson Hole to stop narrating the path of policy. The same week, Treasury was already managing the long end by buying it.
The Fed did not lose its voice on Friday. It declined to keep using that voice as a market instrument. Chair Kevin Warsh’s first Jackson Hole speech refused both forward guidance and a published reaction function, a bid for what he called a “quieter Fed” after a decade in which policy rhetoric was itself a form of liquidity. That silence arrived in a week the long end was still digesting Treasury’s August 19 decision to double “liquidity support” buybacks in the 10-to-30-year sector from $2 billion to at least $4 billion per operation, effective September 9 through November 4. Markets did not lose a narrator. They changed who they listen to.
Silence is a liquidity instrument being put away
Jackson Hole used to be the set piece in which the Fed authored the term structure: a path, a reaction function, sometimes a new framework. Warsh used the same stage to retire the genre. “You can call it an outline. You can call it a trail map,” he said. “Just don’t call it forward guidance.” He would not supply a rule from which a rate path could be read off if the data came in hot or cold. Knowledge, he argued, does not extend that far.
For most of the 2010s and the pandemic years, words from the chair compressed term premium by promising that short rates would stay put. Rhetoric did the work of duration. It changed the price of time without changing the size of anyone’s book. A quieter Fed is an attempt to give that instrument back. If the attempt holds, the two-year and the thirty-year stop being chapters in the same official story.
The market heard the refusal as tightness, which is the honest first reading. The two-year yield rose nearly 8 basis points on Friday to 4.31 percent, its highest since late July, according to CNBC. CME FedWatch put the odds of a September hike at about 55.7 percent, up roughly 20 percentage points on the day. Warsh declined to treat the summer’s better inflation prints as evidence that underlying trends had improved, and he named artificial intelligence “a new factor of production,” a supply story with a long duration attached. None of that is dovish. It is also not a map. He did not mention the buybacks.
Treasury is already in the yield-management business
Nine days earlier, the fiscal authority did the opposite of going quiet. Off the quarterly refunding calendar, Treasury said buybacks in the 10-to-20-year and 20-to-30-year nominal sectors would at least double, from a $2 billion maximum to at least $4 billion per operation. The 30-year had just printed around a 19-year high. Reuters reported the long bond down almost 10 basis points on the announcement, then a retracement. BNY’s iFlow Short Thoughts, written ahead of Jackson Hole, put the round trip more cleanly: 5.31 percent on the Tuesday close, 5.19 percent on Wednesday, then back above 5.20 percent, with most of the term-premium dip given back.
A regular program that mops up off-the-run paper is debt management. An unscheduled doubling, two weeks after a refunding, aimed at the sector that had just made a 19-year high, is a reaction function. It does not have to be yield-curve control to change who the market thinks will lean against the long end. It only has to show that someone in an official building is willing to do so between scheduled announcements.
BNY’s Treasury desk, using in-house depth of book in 10-year equivalents, found no meaningful pre-announcement collapse in liquidity, and neither did Bloomberg’s government-liquidity index. If the book was not thin, “liquidity support” is a story the market is being asked to tell itself. The thing being supported looks more like a yield. Treasury has reserved the right to make that decision for the long end of its own curve, with a cash balance BNY puts well over $900 billion and, eventually, with bills.
The term structure is no longer a single-agent problem. It is a contest between a fiscal authority that will buy duration and a monetary authority that will not narrate it.
The small-size objection is the right one to take seriously
The counter-case is the note a careful rates desk would write, and parts of it are already in BNY’s.
The checks are small. Even doubled, the operations add a low-tens-of-billions impulse against a market measured in the tens of trillions. Reuters put the incremental capacity this quarter on the order of an extra $14 billion. That is not quantitative easing and not a cap. Drawn from the Treasury General Account, the cash buffer is large enough that the operation would barely show. Funded by bills, as BNY thinks they ultimately must be if Treasury sticks to an $850 billion year-end TGA target, the government is shortening its own duration, not shrinking the stock of debt. Either way, the flow is too modest to pin the 30-year against a deficit that still has to be placed.
Warsh’s silence already did work the buybacks cannot. Hike odds jumped the same morning. The front end sold off. A chair who refuses to pre-commit, while calling inflation trends unimproved, is not a hidden cut. Markets that spent a decade fading the Fed’s words now have to underwrite the data instead.
And the long end was never going to be set by who speaks. Term premium spent the spring rising without a policy event, as we argued in July. Deficits, a cooler official foreign bid, and AI capex still compete for the same long-duration dollar. Warsh himself called AI a new factor of production. On this view, the buybacks are a rounding error, the quiet is already in the two-year, and the 30-year will still be written by supply and the build-out.
All three points are true as far as they go. They do not settle the institutional question. A backstop does not have to be large on day one to change the game. It has to be understood as scalable. Treasury dated the new maximum through November 4, the next refunding, and left the phrase “at least” on the $4 billion. That is an option, not a size. The Fed, in the same window, declined to publish the function from which a size could be inferred. Traders can believe the checks are small and still have to decide which building they think will flinch first.
The residual claimant is a private balance sheet
The earlier repricing of term premium asked what long money costs when no one at the Fed asked it to cost more. The question now is who stands behind that cost when the two official books pull in opposite directions. One still sets overnight rates and is trying not to narrate the rest of the curve. The other issues the paper, holds a cash account well above $900 billion, and has just demonstrated it will buy duration off-cycle under a liquidity label. Neither book is the residual claimant. Insurers matching long-tailed liabilities, pensions that still need thirty-year product, and the dealers who warehouse auctions between those two official postures are.
That is a re-underwriting problem, not a forecast of the next print. A single-agent reaction function at least told a treasurer which speech to wait for. A two-agent curve does not. The spread business that assumed cheap, stable term premium has already had to mark a higher cost of time. It now has to mark a second thing: official duration supply and official duration demand are no longer run from the same theory of the market. Liquidity, we have argued, is a story told in common. The long end is becoming a story told in two buildings, and only one of them is still talking.