The Only Disciplinarian Left

Set aside whether he's right about the deficit. He is describing a mechanism, and the mechanism has a service history.

In 1942 the Federal Reserve pegged the yield curve. Treasury bills at three-eighths of a percent, long bonds at two and a half. It was a war-finance measure and it worked exactly as designed. The problem, as always, was afterward. By 1950 the Korean War had brought inflation with it and the Fed wanted out. Treasury Secretary John Snyder did not: he had debt to roll and a price he liked. In January 1951 Harry Truman summoned the entire Federal Open Market Committee to the White House — the only time a president has ever called in the whole body — and explained that maintaining the peg was a patriotic obligation. The committee listened, went home, and disagreed. On March 4 the two agencies issued a joint statement announcing they had reached "full accord with respect to debt management and monetary policies" in order to "minimize monetization of the public debt."

The word accord is carrying an enormous load in that sentence. It was a surrender document. Treasury gave up the ability to set the price of its own borrowing, permanently, and the man it had sent to negotiate the terms — William McChesney Martin — was promptly installed as Fed chairman, where he spent the next nineteen years enforcing the thing he had been dispatched to prevent. Nobody has held the job longer.

The comparison has to survive an objection here, and unanswered the objection is fatal. What the Fed did in 1942 was monetization: a central bank creating money to buy bonds, defending a price it can always defend because it prints the thing it pays with. A Treasury buyback is not that and cannot become that. Treasury retires old, off-the-run long bonds and funds the purchase by issuing new debt, mostly bills. It has no printing press, no unlimited firepower, and no capacity to add a dollar to the system. What it has is a maturity swap.

The parallel survives on that swap, which is precisely the thing the 1951 negotiators never thought to fence. Every long bond Treasury retires and replaces with bills is duration the private market no longer has to hold, and term premium is simply the price of holding duration. Shorten the average maturity of the public debt and the long end comes down — without a dollar being created, and without asking the Fed for anything. That is price management by the other route: the one available to a fiscal authority that doesn't have a central bank's balance sheet and, as it turns out, doesn't need to borrow one.

Which brings us to the tell, and it recurs with the reliability of a reflex: the peg is never called a peg. It is called orderly markets. It is called liquidity support. It is called plumbing. The 1942 operation was framed as wartime necessity, which it was; the 1947–51 continuation was framed as the same necessity, which it was not. The vocabulary does not change when the purpose does, because the vocabulary is what conceals that the purpose changed. Treasury's stated reason this month is a specimen worth keeping. It cited "consistent strong sponsorship from market participants" — it is buying more long bonds because dealers keep offering it long bonds. That is a reason to run the program. It is not a reason to double it in the week the 30-year hit a nineteen-year high.

This should be falsifiable, and the obvious test turns out to be worthless — which is worth showing rather than quietly replacing. The obvious test is a sunset. In September 2022 the Bank of England bought long gilts on an emergency footing after the Truss mini-budget detonated the liability-driven investment complex and pension funds started facing margin calls they could not meet. That was a genuine stability operation, and the tell was structural: announced with an end date, ran thirteen business days, closed on October 14 as scheduled, and the Bank subsequently sold what it had bought. Intervention is not automatically a peg.

But Treasury's announcement has an end date — November 4, 2026 — and it means nothing. That is the last day of the refunding quarter, the boundary at which every issuance parameter is reset by default. A date that arrives and rolls forward is a calendar, not a sunset, and a test an operation can pass by filing paperwork is not a test.

So here is one it could fail. The stated purpose is liquidity, and liquidity is measurable: bid-ask spreads in the off-the-run long sector, the on-the-run premium, dealer balance-sheet capacity. If operation sizes track those, the stated purpose is the real one. If they track the 30-year yield, it isn't. And the cleanest disconfirming event is a decline — if the long end falls substantially and the buybacks shrink along with it, I am wrong and this is plumbing. A liquidity program has no reason to contract when yields drop. A price-support program has every reason.

The other counter-example runs the opposite way. The Bank of Japan ran explicit yield curve control from 2016 to 2024 and had the unusual decency to name it accurately — it said out loud that it was setting the price of ten-year money, and then did so for eight years. What ended it was not the bond market. It was the yen, which delivered the same information through a different instrument at a worse exchange rate.

That is the whole of it. A suppressed price does not stop reporting; it reports somewhere else. Cap the long end and the fiscal news migrates into the currency, into the gold bid, into the inflation print, into the term premium that reappears the instant you stop buying — and it arrives late, compounded, and attributed to whatever is in the newspaper that week. Which makes disciplinarian the wrong word, even in a headline: a system doesn't need a punisher, it needs an undistorted signal, and the punishment is downstream and automatic either way. The disciplinarian is the arithmetic. The yield was only ever the room's cheapest instrument for reading it.

And the arithmetic is now being read through an instrument nobody thought to fence. The 1951 Accord fenced monetization. It is a good fence, and it has held for seventy-five years, and it says nothing whatever about duration. Since 2023 the composition of Treasury issuance — how much bill, how much bond — has functioned as a monetary lever operated entirely from the fiscal side: no Fed cooperation, no White House summons, no joint statement, nobody to negotiate with at all. Truman needed a room full of governors and couldn't get them. The modern version needs a refunding calendar and a press release.

Which is worth setting beside what happened in June, when the Supreme Court preserved a Federal Reserve governor's seat and the thing that appeared to do the work was not a reading of Article II but the cost of the alternative — quoted live, on a screen, in the same long end now being smoothed. Seventy-five years is a respectable run for a fence nobody remembers building. It is a shorter run than it looks if the gap beside it was never measured, and shorter still if the instrument standing guard is the one being managed.