Who Buys the Debt?

Last week we argued that artificial intelligence is a genuine tailwind for productivity, that it is not on its own enough to fix the fiscal trajectory of the United States, and that the bond market — rather than the ballot box — is the more likely forcing mechanism for an eventual correction. This issue takes up the question that was left standing. If the bond market is to force anything, who is sitting on the other side of the trade? A forcing mechanism needs a buyer as much as a seller, and the identity of that buyer tells us a great deal about how orderly the correction is likely to be.

The Trajectory Is Not in Dispute

The growth in US government debt over the next three decades is not a forecast in the way a GDP or inflation print is. It is, to a first approximation, arithmetic — and the Congressional Budget Office has already done it. CBO's February 2026 baseline puts federal debt held by the public at 101 per cent of GDP this fiscal year, closing in on the post-war record of 106 per cent set in 1946. That record is expected to fall by 2030, with debt reaching 120 per cent of GDP by 2036 and 175 per cent by 2056 — more than three times the fifty-year average. In dollar terms, debt held by the public grows from close to USD 31 trillion today to roughly USD 56 trillion by 2036.

The deficit path behind this is widening too. CBO has the fiscal 2026 deficit at USD 1.9 trillion, or 5.8 per cent of GDP, rising to USD 3.1 trillion, or 6.7 per cent, by 2036 — USD 23.1 trillion cumulatively over the decade. Net interest, the least politically negotiable line in the budget, passes USD 1 trillion this year and doubles to around USD 2.1 trillion by 2036. None of this requires a recession, a war, or a policy accident. It is the baseline.

The Supply Wall

Treasury's own advisers are the most useful witness, because they have to place the paper. The Treasury Borrowing Advisory Committee — primary dealers and buy-side investors who meet quarterly with Treasury's debt managers — has been candid about the gap opening between today's auction sizes and the borrowing that will actually be required. At the Committee's most recent meeting, Treasury's Office of Debt Management noted that dealers still judge the government slightly overfunded in FY2026 at current auction sizes. On the same sizes, however, the median dealer forecast implies a funding shortfall of roughly USD 1.1 trillion across FY2027 and FY2028. Dealers broadly expect coupon auction sizes to rise again in 2027 — the first increase since the large expansion of 2023. That shortfall is the supply wall in concrete form — a scheduled event that dealers, obliged to bid at every auction, are already planning around.

The Traditional Buyers Are Stepping Back

For most of the post-crisis period the marginal buyer of Treasury debt was either a foreign official reserve manager or the Federal Reserve. Neither was price sensitive. A reserve manager in Beijing or Tokyo bought Treasuries as a by-product of currency management, not because the 10-year looked cheap. The Fed, running quantitative easing, bought to hit a policy objective, not a return target. Both have been retreating, for reasons that predate the current fiscal debate.

China's holdings have fallen by more than two-fifths from their 2013 peak, almost entirely because Beijing stopped accumulating dollar reserves once its currency priorities shifted from holding the renminbi down to preventing capital flight. That is a story about China's exchange rate regime, not a verdict on US creditworthiness — but the effect on the buyer base is the same either way. Japan's holdings have fallen more modestly, partly offset by a rotation into US agency debt and partly a function of the rising cost of hedging dollars back into yen. Meanwhile the Federal Reserve has been running down its own holdings through quantitative tightening, removing a second price-insensitive buyer at the same time. As State Street Global Advisors have observed, domestic private investors — commercial banks prominent among them — have already become the default marginal buyer by process of elimination.

Enter the Basis Trade

The most significant new entrant of the past decade has not been a pension fund, an insurer, or a sovereign wealth fund. It has been the leveraged hedge fund community, operating mainly through the cash-futures basis trade. A fund buys a cash Treasury security, financed almost entirely with short-term repo borrowing, and simultaneously sells the matching Treasury futures contract, capturing the small spread between the two prices as they converge at expiry. At leverage ratios the Fed's own researchers describe as fifty to one or higher, a spread that looks trivially small becomes a business.

Fed researchers estimate that hedge funds' long Treasury exposure grew from roughly USD 600 billion in 2014 to about USD 2.4 trillion by September 2025. The basis trade alone reached some USD 830 billion of gross notional — nearly double its early-2020 peak, and around 35 per cent of total long positioning. By early 2026 hedge funds held close to USD 2 trillion of Treasuries outright: a record 7 per cent of the marketable market, more than double their share five years earlier.

Markets that lived through March 2020 will recognise the risk. It was the forced unwinding of these same leveraged positions, as funding markets seized in the early pandemic, that turned a liquidity event into a near-disorderly episode in what is supposed to be the world's deepest market. Reports through mid-2026 suggest hedge funds have pulled back more than USD 200 billion of leveraged basis positions as the spread has compressed to the point where even extreme leverage cannot make the trade pay. If that pullback is durable, it shrinks precisely the buyer that has been quietly absorbing new supply — just as the supply wall begins to bite.

Will Banks Be Made to Hold More?

US regulators finalised a recalibration of the enhanced supplementary leverage ratio in late November 2025, effective 1 April 2026. The change replaces the flat 6 per cent enhanced standard for the largest globally systemic banks with a lower, more risk-sensitive calculation. The rationale is that a leverage ratio designed as a backstop had become the binding constraint on how much Treasury inventory the largest dealers could hold. The reform raises the ceiling on Treasury exposure before leverage constraints bind. It does not create new bank capital or fresh commercial appetite for holding low-yielding government paper. Separately, central clearing of Treasury repo should free further dealer balance sheet — the Brookings Institution estimates netting benefits of up to USD 1.3 trillion if all dealer repo were centrally cleared. Both reforms point the same way: regulators are widening the pipe through which banks can absorb Treasury supply, precisely because the alternatives are proving less reliable.

New Buyers, New Volatility

Over roughly a decade, the Treasury market's marginal buyer has moved from price-insensitive official and central bank demand toward a mix of fast-moving leveraged hedge fund capital and a banking system whose capacity is being expanded by regulatory design rather than by organic appetite for duration risk. Each of these buyers behaves differently under stress than the one it replaces. The New York Fed's term premium gauge, dormant or negative for most of the post-crisis period, turned decisively positive in 2023 and has stayed elevated; BBVA Research noted in August 2026 that the year was on track for a third consecutive annual rise. A rising term premium is the market's own admission that it now requires more compensation to hold long-dated government debt. The sharper risk is not the level of the term premium but its behaviour in genuine stress: a fund running a basis trade at fifty to one has neither the balance sheet nor the mandate to hold through a drawdown the way a reserve manager could.

Three Scenarios

We are not in a position to assign confident probabilities to these paths, and we would treat anyone who claims to be with some scepticism. What we can say is that nothing in the data reviewed here points toward scenario three arriving on its own, without a deliberate and sustained policy choice.