Buying trouble: the limits of the Bessent put

America's Treasury is borrowing short to bid down its own long-term debt. The arithmetic will not cooperate.

SCOTT BESSENT has long treated the ten-year Treasury yield as the scoreboard for his economic record; he calls himself America's "top bond salesman". As long-term borrowing costs climbed through a jittery summer, he reached for the one lever a treasury secretary actually owns. On August 19th the Treasury said it would at least double its purchases of long-dated government bonds, from $2bn to $4bn per operation; the next day he told CNBC that the pace could end up "more than $4 billion". The scale of the gesture is modest. The significance of what it signals is not.

Start with the size. Against a $30trn stock of marketable Treasuries, $4bn of purchases is what Wall Street called, with unusual unanimity, "a drop in the bucket". The ten-year yield, which had jumped nearly 70 basis points to about 4.7% this summer, dipped on the news and then clawed its way back up. The thirty-year bond, having touched its highest level since 2007, gave back a few basis points before resuming its climb. Investors priced in the futility before a single extra bond had been bought.

The boldness is not in the arithmetic but in what the Treasury has chosen to do with one of its most basic tools. Debt management is supposed to be plumbing: boring, predictable and strategically inert, its only job to fund the government at the lowest average cost without telling the world anything about policy. Buying back "off-the-run" securities — older bonds that have faded from the thick of trading — is a legitimate part of that mission. It improves liquidity, and the Treasury notes that its buyback operations are chronically oversubscribed. In that guise, the bigger purchases are not monetisation: unlike the Federal Reserve, the Treasury creates no reserves, and swapping one kind of debt for another changes the mix of the stock, not its size. To be sure, a modest programme of this kind can steady nerves; the thirty-year duly recorded its largest one-day fall since late June when the announcement landed. But a debt manager who doubles the schedule in the middle of a panic and promises more is no longer doing plumbing. He is bidding against time for his own liabilities.

The deeper irony is that Mr Bessent spent 2024 denouncing this very manoeuvre when someone else performed it. As a candidate he accused Janet Yellen, then treasury secretary, of "taking control of monetary policy" by funding the government with short-term bills so as to hold down long-term yields — of, as he put it, distorting the Treasury market. Now he presides over a variant of the same contrivance. The arithmetic is both the temptation and the trap. To buy a $4bn long bond, the Treasury must sell $4bn of something else; in practice that something is bills, the shortest and cheapest debt on offer. Cheapness has a price, though. Bills must be rolled over constantly, and they already make up 22.2% of outstanding federal debt, above the rough 20% ceiling recommended by the Treasury's own advisory committee. A bill-heavy debt is hostage to whatever the Fed does next, and the hostage is valuable: the government paid $963bn in net interest in the first ten months of this fiscal year, roughly 15% of federal spending. Borrowing short to make long rates look tame is a decision about when the interest pain lands, not whether it does.

The real product of the announcement was a signal. A few billion of purchases cannot move a market of tens of trillions; a treasury secretary promising "more to come" and brandishing a "big toolkit" can. The message was that the administration treats the long end of the yield curve as a level to be managed rather than a price to be observed — that it has identified a pain threshold beyond which it will intervene. Investors are right to trade on a promise of that kind. They now hold what might be called a Bessent put: the belief that the Treasury will step in whenever yields rise uncomfortably. The trouble with a put that everyone can see is that it invites the market to squeeze it.

Each spike becomes a test of nerve, and investors demand compensation for the risk that debt management has been subordinated to politics. The compensation is a tax on the very thing Mr Bessent wants to cheapen. The "term premium" — the extra yield investors demand for the uncertainties of holding long bonds — properly prices the risks of inflation and fiscal recklessness; add the risk that the price is being managed, and the premium rises, not falls. The historical record agrees. Strategists at UBS note that similar interventions in Japan and Britain did not permanently lower borrowing costs so long as fiscal and inflation fundamentals stayed unfavourable. Meanwhile the market's most patient buyers are stepping back: foreign central banks, once the most price-insensitive bid for Treasuries, are buying fewer, so the marginal price of American debt is set by investors who must be bribed to hold it.

The heaviest toll may fall not on the bond market but on the Federal Reserve. Kevin Warsh, the new Fed chairman, faces his first Jackson Hole speech, due on August 28th, caught between a treasury secretary asking for lower long yields and his own convictions. He has called changes to the central bank's balance-sheet "partially fiscal policy in disguise" and has argued for revisiting the 1951 Treasury-Fed accord, the post-war pact that freed the central bank from financing the government at pegged rates, to give the treasury secretary more say over its holdings. Yet his own preference — to shrink the Fed's portfolio and skew the remainder toward short maturities — would push long yields up, the opposite of Mr Bessent's goal. One of them will give way. The greater danger is that the price signal central to modern central banking is jammed in the process. Bond yields are the unfiltered message through which fiscal improvidence is disciplined; the more the treasury secretary tinkers with the messenger, the less the message is worth. As Joseph Brusuelas of RSM US puts it, the logic of populism is to insist that the central bank support fiscal objectives. Mr Warsh is the man who must prove that logic wrong.

The deepest truth is that no quantity of buybacks can outbid the supply. America's national debt has crossed $40trn and the budget deficit is running at about 6.4% of GDP, roughly $2.1trn this year; the government must still sell about $550bn of bonds this quarter, into a market already flooded by hyperscalers borrowing to build AI data centres. War in the Gulf has pushed Brent crude above $90 a barrel, reviving the inflationary pressure that the Fed's preferred price measure, still at 3.7%, had begun to shed. Against that, Mr Bessent's assertion that "yields don't reflect the underlying fundamentals" reads less like a diagnosis than a hope. The bond market's answer is that the yields reflect exactly that: the fundamentals are the problem. As Gennadiy Goldberg of TD Securities notes, restraining the deficit is Congress's job, not the Treasury's — which is why the Treasury is left holding this lever instead of a better one.

The choice before Mr Bessent is the one his entire career was spent exposing in others, and it will not be resolved by borrowing more. Money raised at the short end to suppress a long-term rate is not a discount; it is a rollover, a promise to face the arithmetic again, sooner, at a price set by other people. When the promise is strained, the strain will surface somewhere. The dollar fell sharply on the day of the announcement, a reminder that a currency can absorb what a yield curve will not — and that a weaker dollar imports inflation, the last thing a man trying to push long yields down needs. A treasury secretary cannot have cheap long-term borrowing, an independent central bank and a shrinking deficit at once; he can promise all three, but he will be forced to choose. The market will keep testing the promise until Congress delivers the only one of the three that lowers the term premium. American yields will fall when American deficits do, and not before.