Where the Leverage Hides
In the sovereign-bond market, the leverage that matters often hides inside a trade designed to exploit a small price difference. This chapter follows how a hedge fund can turn that difference into a position far larger than its own capital, how repo and derivatives financing make the position possible, and why a change in funding terms can force sales even when the underlying government bond remains a high-quality asset. The practical judgment is to distinguish the size of an exposure, the amount of capital supporting it and the cash that must be found when markets move.
The strategy is not mysterious once the balance sheet is visible. A fund brings some equity, borrows against securities and uses futures, spot bonds or interest-rate derivatives to express a relative-value view. The return is generated by the price relationship between two instruments, not necessarily by a large directional bet on whether government bonds will rise or fall. The risk, however, is carried by the whole financed position and by the contracts that keep it funded.
How a small spread becomes a large position
A relative-value trade begins with an apparent inconsistency. Two instruments that should have closely related prices—such as a government bond and a related futures contract, or two positions with similar interest-rate exposure—temporarily diverge. If the difference narrows, the trader can earn a small spread. The expected gain on each unit may be modest, so the strategy becomes more attractive when the fund can hold many units with a limited amount of its own capital.
Consider a deliberately simplified example. A fund contributes $100 of equity and finances a $1,000 position. A 1 percent loss on the position is $10, equal to 10 percent of the fund’s capital. If the trade contains offsetting positions, the net directional exposure may be smaller than $1,000, but the financing, collateral and margin obligations can still be based on large gross positions. That is why leverage changes the meaning of a small price move: the relevant question is not only how far the bond moved, but how much balance sheet was built around the move.
In real markets, the trade can involve a cash Treasury position on one side and a futures or swap position on the other. The instruments may be economically related without being identical. Their prices can diverge because of funding costs, delivery options, dealer balance sheets, liquidity or a change in demand for one instrument. The hedge reduces some risks, but it does not remove the need to finance the positions or to meet margin requirements on derivatives.
This is the setting in which the Federal Reserve reported near-record hedge-fund leverage. Its measure of gross notional leverage is gross notional exposure divided by net asset value, or NAV. NAV is the fund’s capital after accounting for its assets and liabilities; gross notional exposure counts the face amount of positions before the reader treats them as a single net bet. The distinction matters. A high gross number signals extensive market activity and potential funding needs, but it does not by itself reveal the fund’s net loss if prices move, the quality of its hedges or the order in which positions would be sold.
The Federal Reserve’s latest comprehensive Form PF data in the report cover the third quarter of 2025. They show gross notional leverage near all-time highs, concentrated among larger funds, and support positions in Treasury securities, interest-rate derivatives and equities. A March 2026 survey of dealers found no broad change in hedge-fund leverage since the end of 2025. Together, those observations support a claim about sensitivity, not a forecast. A leveraged strategy can remain profitable and orderly for a long time when its relative prices, financing and risk limits cooperate.
The hidden feature is the dependence between return and funding. A small spread may be the visible source of return, but cheap and plentiful financing determines whether the fund can hold enough of the position for that return to matter. If the financing disappears before the spread closes, the trade’s economic logic becomes secondary to the need for cash.
Repo makes collateral part of the mechanism
Repurchase agreements, or repos, are one of the main ways a market participant raises short-term cash against securities. The borrower transfers or pledges collateral and receives cash, agreeing to reverse the transaction later. The lender is protected by the collateral, while the borrower obtains financing without permanently selling the asset. Government bonds are especially useful in this role because they are widely accepted, actively priced and generally viewed as liquid.
A repo haircut is the difference between the market value of the collateral and the amount of cash advanced against it. If a $100 bond is subject to a 2 percent haircut, the borrower receives about $98. The gap is a buffer against a price decline, the cost of liquidating the bond or uncertainty about its value. A zero haircut allows borrowing against essentially the full market value. It may look efficient while prices are stable, but it leaves less room for the lender if the collateral must be sold quickly.
Haircuts therefore influence leverage before any crisis begins. A low haircut lets a fund finance a larger position with the same equity. The position can earn a higher return on equity if the expected spread is realized, but the fund also has less protection against a change in the bond’s price or in the lender’s willingness to finance it. The risk is not simply that the collateral loses credit quality. A highly liquid asset can still become difficult to sell at the expected price when many financed holders need cash together.
The BIS reports that around 70 percent of bilateral US-dollar repos and more than 50 percent of bilateral euro repos with hedge funds were transacted at zero haircuts. Those figures describe the covered bilateral transactions, not every repo in every currency. The report also says that the most favorable terms were concentrated among the largest funds. That concentration matters because a funding change affecting a large borrower can produce a larger sale, and because several funds may rely on the same prime brokers or dealers.
A lender does not need to believe that a government will default to change a haircut. The lender may become more concerned about price volatility, the time needed to liquidate collateral, the fund’s other positions or the amount of risk already carried on the lender’s balance sheet. A dealer may also face its own internal value-at-risk limit or a higher capital cost for warehousing positions. Funding can tighten because the intermediary’s capacity changes, even if the original relative-value relationship has not changed.
This is what makes repo more than a neutral pipe between cash and bonds. It is a contract that determines how much balance sheet can be built on a given asset and how quickly that balance sheet must shrink when conditions deteriorate. The same government bond can support more leverage when haircuts are low and less leverage when haircuts rise. The market’s apparent depth is partly a product of those terms.
The path from funding pressure to forced selling
The stress sequence begins with a price or volatility move. A rise in yields lowers the market price of an existing fixed-rate bond. A change in the relationship between a bond and its hedge can create a loss on one leg of a relative-value trade even if the position is expected to converge later. A derivatives counterparty may ask for additional margin, or a repo lender may increase the haircut. The fund then has to supply cash or reduce the position.
If the fund has spare liquidity, the first pressure may be absorbed. If not, selling becomes a way to meet the demand. The sale pushes down the price of the security and can make the market less liquid. Dealers that normally buy inventory may become more cautious because volatility has increased, their risk limits are closer or their balance sheets are already full. The fund receives less cash than expected from each sale, which can require more selling. A small initial move has become a balance-sheet feedback loop.
The sequence is easier to follow when the participants and contracts are kept together:
The diagram is deliberately simplified. A fund may use several dealers, a clearing intermediary or more than one source of secured finance; the repo lender may not face the fund directly. The important relationship is the sequence: a position supported by collateral is exposed to terms set by intermediaries, and those terms can change faster than the underlying investment thesis can play out.
The BIS uses the March 2020 margin calls and the April 2025 unwind of swap-spread trades as examples of this mechanism. The lesson is not that the episodes were identical or that every relative-value trade must unwind violently. It is that funding terms can turn a price relationship into a source of market-wide selling. In calm conditions, low haircuts and active dealers help positions coexist. Under stress, the same arrangements can synchronize decisions that would otherwise be made separately.
Concentration increases the sensitivity. The largest funds received the most favorable repo terms in the BIS evidence, while the Federal Reserve reports that leverage is skewed toward larger funds. A lender may regard a large fund as sophisticated and diversified, and the fund may have better access to several markets. But a large position also creates a larger amount of collateral to sell if financing is withdrawn. If multiple funds use similar strategies, their supposedly independent trades can respond to the same margin signal.
The vulnerability is not limited to the fund’s own loss. A rapid sale can affect the price used to value other positions, the collateral accepted in other repos and the risk limits of dealers serving other clients. That is how leverage in a trade tied to government bonds can alter the liquidity and yields of the broader sovereign-bond market.
What the leverage numbers can—and cannot—tell us
It is tempting to turn a leverage statistic into a simple danger ranking. That would be a mistake. Gross notional leverage, balance-sheet leverage, repo borrowing and net economic exposure answer different questions. Gross notional exposure shows the scale of positions before netting. Balance-sheet leverage relates assets to equity. Repo borrowing shows dependence on secured funding. None of these measures alone reveals the fund’s complete hedge, the maturity of every liability, the collateral held by each counterparty or the price at which a forced sale would occur.
The data also arrive with different levels of timeliness. The Federal Reserve’s detailed hedge-fund measure is based on the third quarter of 2025, while its dealer survey provides a later but less granular indication. The BIS combines data on repo terms, sovereign holdings and trading activity from several sources. The reports are strong evidence that leverage and funding dependence are material vulnerabilities; they are not a complete, real-time map of every position.
This distinction keeps the conclusion properly bounded. High leverage raises the potential for spillovers if a fund loses access to funding. It does not prove that a spillover will occur, identify the next seller or quantify the eventual loss. Current market conditions can remain orderly because funds have liquidity, dealers continue to intermediate, price moves remain within risk limits or the shock never arrives. The vulnerability becomes consequential when those buffers and relationships are tested together.
The position is also only one node in the sovereign-bond system. Its effect depends on the prime broker that finances it, the dealer that makes a market, the other institutions holding the same collateral and the funds that may need to meet redemptions. Once the balance-sheet arithmetic is visible, the next question is no longer just how one hedge fund borrows. It is how that borrowing connects to a network of banks, dealers, insurers, pension funds and other non-bank investors—and whether those connections absorb a shock or pass it on.