Clearing Rewires the Treasury Market

The migration of Treasury repo into central clearing is visible, but it is not a smooth conversion of one market into another. Sponsored DVP settled an average of 0.59 trillion dollars of transactions a day in May 2023 and 1.59 trillion in August 2026, an increase of 169 per cent. Over the same span the rest of FICC DVP rose 41 per cent and non-centrally cleared tri-party volume rose 54 per cent.

Reporting dates supply a second result. At a quarter-end, tri-party volume falls 4.3 percentage points more than it does at an ordinary month-end. That difference survives adjustment for four tests. The public cleared series rise, but none supplies comparably strong evidence that the activity leaving tri-party moves directly into clearing.

Section 01The channel doing the work

Central clearing changes the legal and operational path of a repo. A central counterparty becomes buyer to every seller and seller to every buyer, collects margin and default resources, and may replace several gross bilateral obligations with a smaller net obligation. That can reduce counterparty exposures and balance-sheet use when offsetting trades meet the accounting conditions for netting. It also concentrates risk management and liquidity demands at the clearing house. Clearing redistributes and standardises risk; it does not make the financing or its leverage disappear.

Figure 1 follows three public venue series from the Office of Financial Research and the sponsored service published by the Fixed Income Clearing Corporation. FICC clears Treasury repo through two services, the DVP Service and GCF Repo; FICC cleared volume here is the sum of their daily transaction volume. The first hypothesis was that growth since 2023 would be concentrated in the sponsored channel, through which an FICC member submits a client's trade for clearing. A similar rise in non-sponsored DVP would have rejected it.

The result supports concentration, with one qualification. Sponsored DVP rose by 169 per cent from May 2023, against 41 per cent for DVP outside that channel. Sponsored general-collateral activity also expanded sharply. The migration is therefore better described as broader client access to FICC than as uniform growth across an old interdealer venue.

It is not monotonic. On the narrow reading in Figure 1, FICC DVP and GCF together reached 60.5 per cent of transaction volume across the publicly observed venues in November 2025, and every month since has been lower, ending at 56.2 per cent in August 2026. The level remains far above the 39.3 per cent trough in August 2018, but a rule deadline and a long-run direction do not imply a straight line.

Section 02The denominator that is missing

The phrase "cleared share" needs a denominator. The narrow series divides FICC's DVP and GCF transaction volume by those services plus BNY tri-party volume. It is a share of the venues for which the OFR publishes comparable daily histories. It is not the centrally cleared share of the entire Treasury repo market.

The omitted segment is non-centrally cleared bilateral repo. The OFR's 2025 Annual Report sized average daily outstanding positions across the US repo market at about 12.6 trillion dollars in the third quarter of 2025, including about 5.0 trillion in that bilateral segment. Outstanding position is a stock and transaction volume is a flow, so that market-wide figure cannot simply be appended to the daily denominator used here.

A second boundary concerns Sponsored GC. It is centrally cleared but settles on a tri-party basis. Crediting it to the cleared numerator gives the wider reading: a peak of 73.6 per cent and 69.7 per cent in August 2026. The narrow and wider values bracket an operational classification choice inside the published venues. Neither recovers the missing bilateral market.

daily public series, converted to USD trillions:

    FICC cleared       = DVP transaction volume + GCF transaction volume
    sponsored activity = Sponsored DVP + Sponsored GC

    narrow cleared share = FICC cleared / (FICC cleared + tri-party)

    wider reading credits Sponsored GC to the cleared numerator
    but still excludes non-centrally cleared bilateral repo

sample: 2018-05-07 to 2026-08-31, 2072 sessions
monthly observations are means of daily transaction volume,
taken over complete calendar months only

Section 03A reporting-date experiment

The balance-sheet account of clearing makes a testable prediction. A dealer reports its balance sheet at a period end. Repos that qualify for balance-sheet netting should be less expensive to carry into that date than otherwise similar gross obligations. If business migrates towards nettable infrastructure at the turn, non-cleared activity should contract more at quarter-ends than at ordinary month-ends, while cleared activity should expand by more.

For each month, Figure 2 compares the period-end observation with the mean of the preceding ten sessions. March, June, September and December form the quarter-end sample; the other eight months are the control. The comparison is made separately for tri-party volume, DVP volume, GCF volume and DVP outstanding. Five-session and twenty-session baselines test whether ten sessions manufacture the sign.

The leaving side is unambiguous. Tri-party volume averages a 4.2 per cent fall at quarter-end and is essentially unchanged at an ordinary month-end. The difference is minus 4.3 points, with a 95 per cent interval from -5.4 to -3.2 and p = 2.6e-10. It survives the four-test Bonferroni threshold of 0.0125, and keeps its sign with all three baseline lengths.

The arriving side is weaker. DVP transaction volume rises at both kinds of period end. Its quarter-end increment is 3.1 points larger, with p = 0.020, but that does not survive the adjusted threshold. GCF volume and DVP outstanding differ by only 0.9 and 0.7 points, with intervals that cross zero. The data show a reporting-date contraction in non-cleared tri-party. They do not identify where those missing trades go, whether they are delayed, netted before execution, shifted into bilateral channels, or submitted to clearing in a series this design cannot isolate.

Section 04Netting is conditional

The contrast does not identify a causal release of dealer balance-sheet capacity. Quarter-end differs from other month-ends in more than reporting intensity, the composition of borrowers can change, and the venue series are aggregates rather than matched trades. Even the DVP result, whose unadjusted p value falls below 0.05, could be a false positive among four tests.

The accounting mechanism is conditional too. Central clearing creates a common counterparty, but balance-sheet offset generally requires more than a common name: enforceable netting rights, the same settlement date, and an intention and ability to settle net or simultaneously all matter. Margin then replaces part of the bilateral exposure with a prefunded liquidity demand. The correct risk question is therefore which exposures become nettable, which new margin calls appear, and where default management is concentrated.

That distinction separates this issue from Issues 002 and 011. Issue 002 follows the cash bond, futures hedge and repo borrowing that make a Treasury basis position possible. Issue 011 compares dealer inventory with the much larger financing stock behind it. Both are about positions and leverage. This issue measures the infrastructure through which an increasing part of that financing passes.

Section 05The rule is a deadline, not an observation

The Securities and Exchange Commission adopted its Treasury clearing rule on 13 December 2023, requiring a covered clearing agency for Treasury securities to ensure that its direct participants submit eligible cash and repo transactions for clearing. Release 34-102487 then extended the compliance date for eligible cash-market transactions to 31 December 2026 and for eligible repo transactions to 30 June 2027, under Rule 17ad-22(e)(18)(iv)(A) and (B).

Both dates sit after most of the sample. They are marked on Figure 1 as regulatory events, not treated as causes. Voluntary sponsored clearing grew before the rule was adopted; balance-sheet incentives, counterparty access, product changes and preparation for compliance all overlap. A time-series line cannot allocate the growth among them.

As at this revision, the SEC was considering requests for further exemptive relief and had reopened the comment periods in August 2026. The Commission had not replaced the compliance dates stated in Release 34-102487. Pending applications are evidence of implementation questions, not a change in law.

This research is analysis and commentary for general information. It is not investment advice, an offer, or a solicitation, and it contains no view on the price of a security. Figures 1 and 2 are the author's calculations from the primary public series cited. Regulatory dates are verified against the SEC release but should be checked again after this revision date.