The Leverage You Can See Is Not the Leverage That Matters

Primary dealers held about 436 billion dollars of net outright Treasury positions in the week to 19 August 2026. Around the same time, roughly 5.4 trillion dollars of repo was changing hands every day. The first number is the one usually quoted as dealer balance-sheet risk. The second is about thirteen times larger, and it is the one that determines who has to sell when funding turns.

Figure 1 plots both on a single axis, which is the point of drawing it that way. The inventory line sits along the bottom of a chart whose vertical extent is set by the financing that levers positions across the system.

Section 01Two rungs, and the ones below them

Leverage is not one quantity. Assets divided by net asset value measures balance-sheet gearing. Gross notional exposure adds derivatives, often with delta or duration adjustments. Neither measures the cash a position will demand tomorrow if prices move, which is the quantity that decides whether a holder stays in the trade.

The two series in Figure 1 are the two rungs public daily data can reach. Dealer net inventory is what is owned, net of what is shorted, and it is published weekly. Repo volume, collected by the Office of Financial Research across the delivery-versus-payment, tri-party and general collateral finance segments, is the secured borrowing through which positions are financed. Total daily repo volume has risen 221 per cent since May 2018, from about 1.7 to about 5.4 trillion dollars.

The ratio between them has been remarkably stable: a median of 13.1 times over the sample, ranging from 8.1 to 56.7. That stability is itself informative. The financing system has not been growing relative to inventory; it has always been an order of magnitude larger, and a discussion of leverage anchored on inventory has always been anchored on the smaller quantity.

Section 02The leverage visibility ladder

Figure 2 stacks the questions rather than ranking instruments by visibility. Balance-sheet assets show funded positions. Repo adds maturity, counterparty and haircut terms. Futures and swaps translate a small current accounting exposure into a large economic one. Below all of these sit the collateral schedule and the available cash, which is where a position is actually won or lost in a stress.

No rung is the true leverage measure. Each is a projection built for a different purpose and each misleads outside it. A gross asset ratio treats a fully hedged book and a directional one alike. A net exposure measure can show almost nothing for a position carrying enormous gross two-way risk. A defensible assessment reconciles gross assets, net and gross derivative exposure, financing terms, and the contingent cash calls each of them can generate.

balance-sheet leverage    =  assets / NAV
gross notional exposure   =  sum of |adjusted notionals| / NAV

neither answers:
    cash_due_tomorrow  =  variation_margin(price move)
                        + haircut_increase * financed_position
                        + repo_rollover_gap

measured, from public daily and weekly sources:

    dealer net Treasury inventory       0.44 tn      (weekly, NY Fed)
    US repo volume, all segments        5.40 tn/day  (daily, OFR)
    ratio                              12.4x   median over sample 13.1x

what neither source contains:
    derivative notionals, bilateral margin terms, haircut schedules

Section 03Why the hidden rungs decide it

The lower rungs matter because they are contingent and procyclical. Margin requirements and haircuts rise when volatility rises, so a quiet, fully collateralised position becomes a demand for cash at the moment cash is dearest. That conversion is not visible in any leverage ratio computed from a balance sheet, because the balance sheet records the position and not the schedule of calls attached to it.

The Federal Reserve's Financial Stability Report has recorded hedge fund leverage at or near record highs in recent comprehensive readings, drawn from Form PF, which arrives with a substantial lag and covers only registered advisers above reporting thresholds. That the most authoritative measure of the rung that matters most is quarterly, delayed, and partial is not incidental to this article's argument. It is the argument.

What a balance sheet omits is precisely what a crisis finds.

The title is deliberately paradoxical and the resolution is narrow. Visible leverage matters and answers one question. Figure 1 establishes that the financing rung is roughly thirteen times the size of the inventory rung and that this ratio has been stable for eight years. It cannot establish what sits below, because derivative notionals and bilateral margin terms appear in neither series and in no public daily source. On the rung that most determines outcomes, the honest position is that the public record does not permit a measurement, and any confident statement about it is coming from somewhere other than data.

This research is analysis and commentary for general information. It is not investment advice, an offer, or a solicitation, and it contains no price forecasts. Quantities in the text are the author's own calculations from the official sources cited.