The Safest Market in the World Is Now Run on 50:1 Leverage

Rising yields made US pension funds whole for the first time in decades, so they stopped buying the bonds that made them whole. The buyer who replaced them earns 0.1% on a trade and borrows 98 cents of every dollar to do it. That is a solvency story dressed as a demand story.

The US Treasury market is roughly $40 trillion of paper that the rest of the world prices its own borrowing against. A mortgage in New York, a corporate loan in London and a sovereign bond in Brazil all carry a spread over the same anchor. For forty years that anchor held because the people holding it did not care what it was worth tomorrow.

That has changed, and the change is not mainly about foreign governments selling. It is about who replaced them, and what that replacement needs in order to keep showing up.

Official foreign holders — the central banks and finance ministries of other countries — have fallen from 42% of all Treasuries in 2008 to 13% today, according to the holdings data in the Edge Funds transcript that prompted this piece. Foreign private holders have stayed flat at 19%. The Federal Reserve has stayed at 13%. Domestic investors went from 25% to 55%, and now hold more than half the market. The marginal buyer of the global risk-free asset is American.

Two American institutions swapped places inside that number. Understanding why requires no model, only a trucking company.

A pension is a warehouse of promises

Imagine a trucking firm with 400 drivers, each promised $2,000 a month from the day they retire until the day they die. The payments arrive on a fixed date, in a fixed amount, with no defined end. One driver draws for ten years; another for thirty. The firm cannot know which.

The obvious answer is to buy an asset that pays a fixed amount on a fixed date for decades with no default risk. Long-dated Treasuries do exactly that. You line the coupon and principal up against the payroll you owe and let the two cancel. This is liability-driven investing, or LDI, and its defining feature is indifference: the bond was purchased to settle an invoice, not to be sold. It can lose half its market value tomorrow and the invoice still gets paid.

Scale that firm to thousands of employers and you have a pension fund. In 2008 US pensions held about $400bn of Treasuries. Fifteen straight years of buying took that to $1.2 trillion by 2023 — nearly 5% of the entire market, more than the whole of Japan.

Then the buying stopped.

Why a rise in yields made the buyer disappear

In 2008 the aggregate funded ratio — assets divided by the promises outstanding — sat near 80%. For every dollar owed, eighty cents was set aside. That gap is what kept the bid coming: every spare dollar went into duration to close it.

Now price the liability. A pension that owes $100,000 in 30 years needs $55,207 set aside today if it can earn 2%. At 5% it needs $23,138. The same promise, 58% cheaper to fund, by MacroTape's calculation from the transcript's rates.

Nothing about the promise changed. The discount rate did. And a rise in the discount rate shrinks a liability the way a rise in mortgage rates shrinks the house you can afford.

So US pensions moved from roughly 80% funded in 2008 to about 105% this year, on the transcript's figures — the first sustained surplus in a generation. Once the invoice is covered, another long bond does nothing for you. It locks in safety you already have, at a yield you have already captured. The trade that was hired to close a gap has finished the job.

Pensions began trimming the surplus out of Treasuries and moving it into private credit, equities and other growth assets. The most price-insensitive buyer in the market became a price-sensitive one, and it became so precisely because yields rose.

The Federal Reserve described the consequence in the meeting notes it released in July 2026: ownership of Treasuries has shifted from relatively price-insensitive holders to more price-sensitive investors. Translated out of central-bank language, the buyers who did not care about price are being replaced by buyers who do.

There is a loop inside that sentence. Higher yields make pensions healthier; healthier pensions need fewer long bonds; fewer pension buyers force the government to pay more to attract the price-sensitive buyer; the higher yield makes pensions healthier again. The UK ran a fast, violent version of it in September 2022, when the Truss mini-budget pushed long gilt yields up more than 1.5% in a month and forced the Bank of England into an emergency intervention to stop the selling from feeding itself. Liz Truss resigned within weeks.

The US has run the first half of that loop without the second. Yields are up; pensions have stepped back. The market did not seize, because a new buyer arrived.

The two doors to the same house

A Treasury can be bought in two ways. You can buy the bond, or you can buy a futures contract — an agreement to take delivery of that bond at a set price on a set date. Futures exist so large investors can move size instantly, to speculate or to hedge, and the demand for that convenience is so strong that a contract often trades slightly rich to the bond it references.

That gap is the basis, and it is the entire trade. Buy the cash bond, sell the future against it, wait. At delivery the two prices must meet. The convergence is arithmetic, not opinion — no view on the direction of rates is required, only a view that the contract will settle.

Work the numbers from the transcript. A hedge fund buys $100m of Treasuries and simultaneously sells a future on the same $100m at $100.1m. At expiry the gap closes and $100,000 is realised. On $100m that is 0.1%, which is not a business.

Scale is the only way the trade pays: at $10bn, 0.1% is $10m. And scale is bought with borrowed money.

Two cents on the dollar

The hedge fund does not need $10bn. It pledges the bonds it just bought as collateral in the repo market — repo being the secured lending market where cash is swapped for securities overnight, the plumbing's pawnshop — and borrows against them. Market estimates cited in the transcript put the haircut as low as 2 cents on the dollar: the fund posts $200m of its own capital and borrows the other $9.8bn.

The same 0.1% gap then earns $10m on $200m of equity, a 5% return, which is a real business. Fifty times leverage is what turns a rounding error into a product.

The pile has grown accordingly. Repo borrowing against Treasuries by hedge funds went from about $1trn in 2023 to $4trn today. That is roughly twice the size of the yen carry trade that the market spent the summer worrying could take down the global financial system. Hedge fund ownership of Treasuries went from 4% in 2023 to 9% — one bond in eleven — and the pace is accelerating: more than $60bn bought in the second quarter of 2026 against $26bn in the first.

Read the two facts together. The marginal holder of the global pricing anchor now runs a convergence trade, funded overnight, marked daily, and sized at 50 times its own capital.

What a haircut does to a 50:1 position

The vulnerability is not the trade. It is the financing.

Suppose a lender raises the haircut from 2 cents to 6 cents, because volatility picked up or a regulator cleared its throat. The fund holding a $10bn position now needs $400m of cash immediately. It does not have $400m; that is why it was levered 50 times. So it sells the most liquid asset it owns, which is the Treasury.

That sale pushes the price down. The next fund, running the same trade with the same collateral, gets a margin call of its own. It sells too. The loop is called a liquidity spiral, and its distinguishing feature is that every participant is behaving rationally and the aggregate is catastrophic.

The Federal Reserve flagged the mechanism in its May 2026 financial stability report, saying explicitly that high leverage can produce spillovers if these funds suddenly lose access to funding. The IMF raised similar concerns this month.

Whether a spiral starts is a question about cash, and cash in this system means bank reserves — the deposits commercial banks hold at the Fed, which are the raw material of the repo market. Reserves have fallen from 17% of GDP in 2021 to about 9% today. Fed research cited in the transcript puts 10% at the line where reserves stop being plentiful and become merely adequate, and 8% as the level where liquidity stress begins to show.

The system has been through this door before. In September 2019 reserves fell below 8%, cash vanished from the repo market, and the overnight rate jumped from about 2% to as high as 10% in a single day. Lenders with cash refused to lend it. The basis trade unwound violently, and the Fed injected tens of billions to stop it.

The basis trade is now roughly five times larger than it was then. Reserves are already under 10% and drifting toward the same 8%.

The codependency

This is the structure the Financial Times called a toxic codependency, and the description is not rhetorical. The government needs the hedge fund to hold the paper that the pension fund no longer wants. The hedge fund needs the repo market to stay open, the haircut to stay low and the basis to stay positive. The repo market needs reserves to stay plentiful. And every one of those conditions is set by a Fed that is trying to run policy for an economy, not a balance sheet.

Reuters has argued hedge funds now pose a greater threat to Treasuries than China does. The Wall Street Journal called them the bond market's wild card. Both are describing the same asymmetry: a holder that will not sell is a source of stability at any price, and a holder that must sell is a source of instability at the worst moment. Pensions were the first kind. This buyer is the second.

The 2019 episode is the useful precedent because of its size, not its severity. The Fed contained it in days with tens of billions. Today the trade is five times larger and reserves are closer to the line, in the one market every other price on earth is built on. The question is no longer whether the Fed would intervene. It is the size of the cheque.

Call and invalidation

**Call.** The US Treasury market has swapped a price-insensitive holder for a leveraged one, and the swap is structural rather than cyclical: 105% funding removes the pension bid at precisely the yields that restore it. That makes long-end yields a function of repo plumbing in a way they were not in 2019, when the basis trade was a fifth of today's size and reserves had room to fall. The risk is not a gradual repricing. It is a funding shock that arrives as a Treasury selloff and is diagnosed, wrongly, as a fiscal event.

**Invalidation.** Three things would break the thesis. Pension funded ratios falling back below 100%, which restores the structural bid as yields rise. A sustained rise in reserves above 12% of GDP, which removes the funding constraint regardless of position size. Or a regulatory change that forces central clearing and higher haircuts onto the basis trade quickly enough to shrink it without a disorderly unwind — the outcome the Fed's 2026 report argues for, and the one that would make this piece's risk scenario obsolete before it fires.

---

*Figures are drawn from the Edge Funds transcript "The FINAL Domino in the US Debt Crisis" (October 2026), the Federal Reserve's July 2026 meeting notes and May 2026 financial stability report, and Treasury par-yield closes for 19 October 2026: 2yr 4.77%, 5yr 4.94%, 10yr 5.15%, 30yr 5.48%. The 30yr present-value figures are MacroTape's own calculation.*