Perli: Supplying Ample Reserves

Introduction

It is a pleasure to offer closing remarks at this year's U.S. Treasury Market Conference.¹ For over a decade now, this conference has been a unique and important forum to discuss developments in, and the evolving structure of, the Treasury market.

There is a lot of work involved with organizing this event, and I want to thank all the staff from the Joint Member Agencies who contributed to making the conference run so smoothly today.²

Both the many participants in this conference and the broader public have a common interest in a well-functioning Treasury market. Among other things, a well-functioning Treasury market minimizes the cost of debt to taxpayers, supports the broader financial system, and helps to facilitate effective implementation and transmission of monetary policy.³

For this latter reason, the Treasury market and monetary policy implementation are closely connected. Today, I'll first discuss the Open Market Trading Desk's approach to maintaining ample reserves by increasing SOMA portfolio holdings, when appropriate, through reserve management purchases (RMPs) of Treasury bills, and factors influencing recent decisions to temporarily pause RMPs.^4,5 I'll then provide observations on how the Federal Reserve's approach to supplying ample reserves compares to approaches adopted at some other advanced-economy central banks.

But before I go further, I will give the usual disclaimer that these views are my own, and not necessarily those of the Federal Reserve Bank of New York or the Federal Reserve System.

Maintaining Ample Reserves

I'll begin with the Desk's approach to RMPs.

As I've discussed previously, our decisions on the size of RMPs are informed by three considerations, all equally important: our assessment of reserve demand, our forecast for reserve supply, and both current and expected money market conditions.⁶ Each of these played an important part in our decision to pause RMPs for the past two purchase periods.

Based on our outreach and analysis, we assessed that reserve demand was little changed over that time. In contrast, our forecast for reserve supply did change materially as we updated it to reflect Treasury's guidance for the end-of-December level of the Treasury General Account (TGA), which was lower than we had previously assumed.⁷ The result was a shift higher in our estimate for the supply of reserves.⁸ This upward revision to reserve supply, combined with a low likelihood of very large swings in that supply (as seen, for example, around April tax season), limited the risks that reserves might fall below ample levels later in our forecast horizon.

A third important consideration in the Desk's RMP decision process is money market conditions. Various signals from money markets confirmed that reserves were ample and likely to remain so in the near term. Money market rates, including the effective federal funds rate and Treasury repo rates, continued to trade close to the interest rate on reserve balances (IORB) (Panel 1). To be precise, overnight money market rates, on average, were a bit below IORB, which suggests that reserves were likely in the higher part of the ample range. Other indicators that we monitor regularly sent the same message.⁹ Market pricing was also consistent with expectations for relatively benign conditions to continue in the near term.

It was notable that in July and August a substantial amount of Treasury bill issuance was absorbed without much effect on money market conditions (Panel 2). This was a key watch point for the Desk. In the weeks leading up to our August RMP decision, cumulative net bill issuance was very large, at about $400 billion, but resulted in only very modest upward pressure on repo rates. This contrasted with conditions in the fourth quarter of last year, when repo rates rose substantially and were quite sensitive to bill issuance. Of course, the Federal Open Market Committee (FOMC) doesn't target repo rates. However, upward pressure on repo rates can translate into upward pressure on the federal funds rate, which is the FOMC's target, as happened late last year.¹⁰

Overall, the totality of the information at our disposal suggested that reserves were likely to remain within the ample range in the near term even without RMPs. Accordingly, we felt it appropriate to reduce our RMPs to zero since mid-August (Panel 3).¹¹ As I have said in the past, RMPs are never on a preset course.¹² The recent decisions to set them at zero are no different in spirit than any of the other decisions the Desk has made since RMPs started in December of last year. We have adjusted RMP amounts several times in response to evolving conditions, and we stand ready to adjust them again in the future to fulfill the FOMC's policy to keep reserves within the ample range. For example, we will monitor how the market responds if, as observers expect, another round of significant net bill issuance occurs in October. We recently sent out the latest Senior Financial Officer Survey (SFOS), and we will be interested in any indication that reserve demand might have changed.¹³ We will also be attentive to any signs of pressures in the repo market.

Before moving on, I'd like to comment briefly on the Desk's reserves forecast. The Desk's process for forecasting reserves is robust and generally quite accurate. Forecast misses over the past four years represent a very small fraction of total reserves supply and are easily accommodated by our ample reserves framework. Of course, like any forecast, ours is imperfect, and the market occasionally might throw us a curveball, resulting in reality deviating substantially from our projections. A recent example stemmed from the operational issue that occurred at DTCC at the end of June.¹⁴ In short, a faulty trade submission prevented DTCC from settling transactions and returning cash to member banks at the end of the day. These higher deposits held by DTCC at the Fed led to an unexpected, short-lived reduction in reserve balances, which undershot our forecast by a large amount. While I would be cautious in drawing too many conclusions about market resiliency given some of the idiosyncratic aspects of the event, it's worth noting that the ample reserves framework was also able to accommodate this surprise and avoid broader disruption.

Alternative Approaches to Supplying Reserves

Overall, I would say that the Fed's approach to monetary policy implementation that I just described has been working well (Slide 4). We have maintained very strong interest rate control, we have kept reserves within the ample range, and our Treasury bill purchases have run smoothly.

Of course, this is not the only way of implementing monetary policy, and other operational approaches can work equally well, as seen at other central banks. Consider, for example, the operational approaches adopted by the Bank of England and the European Central Bank (ECB), among others.^15,16 From my perspective, similar to the Fed, these central banks supply all the reserves their financial systems need to operate smoothly and efficiently. However, they supply marginal reserves through repo operations (known as a "repo-led" approach), whereas the Fed has been doing so primarily through asset purchases (which I'll refer to as a "securities-led" approach).¹⁷

Certainly, repo operations are an integral part of the Fed's operating system, designed to support rate control by helping to provide a ceiling on money market rates.¹⁸ But the positioning of the Fed's standing repo operations (SRPs) as a ceiling tool stands in some contrast to repo operations at both the Bank of England and the ECB, which are designed to incentivize usage as part of routine bank liquidity management.¹⁹

The choice of one operational approach over the other can be influenced by a variety of factors. Local realities, including institutional details and the structure of the banking system and financial markets, are some of them. For example, one consideration for the Bank of England, given its institutional arrangements, was minimizing interest rate risk on its balance sheet, which informed its longer-run objective of providing reserves primarily through repo.²⁰ In addition, the sovereign bill market is comparatively smaller in the U.K., which could conceivably be an obstacle to supplying reserves through short-dated securities holdings.²¹ In the same vein, the ECB's repo-led approach was motivated by heterogeneity in the euro area banking sector and the uneven distribution of excess liquidity across member countries-considerations that are less relevant in the U.S.²²

Of course, policy considerations and policy preferences are also important. A few prominent ones that come to mind are the desired size of central banks' balance sheets, their composition and duration, and preferences around broadly and directly supplying liquidity to banks versus relying on market intermediation.

Under both securities-led and repo-led approaches, central banks would provide the amount of reserves that the financial system demands. However, reserve demand under a repo-led approach might be somewhat smaller than it would otherwise be as banks, knowing that central bank liquidity would be available at close to market prices, may require less of a reserves buffer than in the alternative approach. This may lessen the need for the central bank to provision additional reserves ex ante to accommodate exogenous shocks to supply or demand. This point was also demonstrated analytically by President Williams and colleagues in recent years.²³

The choice of implementation approach also affects the composition of central banks' balance sheets and therefore their duration. In general, operating under a securities-led approach would tend to result in a longer-duration balance sheet, depending on the securities held, whereas even a partial substitution of some securities with a repo portfolio would shorten that duration. A repo-driven approach, therefore, can also be thought of as an alternative way (other than changing the composition of the securities portfolio) for policymakers to achieve a certain desired portfolio duration.

Because a repo-led framework inherently relies on central bank liquidity operations, there are trade-offs to consider between directly provisioning reserves and incentivizing private market intermediation. In a repo-led approach, policymakers must decide how directly the central bank should provide liquidity to the system. The central bank could use broad-based repo operations with many counterparties at close to market prices and against a wide variety of collateral to provide reserves directly, though it might weigh these considerations against risks of creating too much reliance on central bank operations. Alternatively, it could choose to accept a narrower set of collateral and rely on a smaller set of counterparties to intermediate liquidity to the rest of the system, which is closer to the Fed's current operating framework.

In our current securities-led approach, reserves primarily enter the banking system via our RMP transactions with primary dealers or via repo operations with that same set of dealers and a relatively small group of banks; repo collateral is limited to Treasury, agency, and agency mortgage-backed securities (MBS).²⁴ A repo-led framework with a smaller set of counterparties would more heavily rely on counterparties' ability to channel liquidity to the rest of the system; therefore, issues such as dealer intermediation capacity and the robustness of interbank funding markets become more important. This point about intermediation capacity brings me to the topic of central clearing, which has been discussed in detail today.

I have said in the past that, strictly from a monetary policy implementation perspective, offering a centrally cleared version of our standing repo operations would offer some clear benefits.²⁵ Those benefits would become more important in an operational approach with greater reliance on repo because it would enhance our counterparties' ability to intermediate liquidity through the system. Of course, like all the trade-offs inherent in any implementation framework, the potential benefits associated with centrally clearing Fed repo operations would need to be weighed against other policy considerations, as I've previously discussed.²⁶

To conclude, I would say that neither of the operational approaches I discussed (securities-led vs. repo-led) is inherently superior to the other. In fact, I would say that they are very similar in principle-they both aim to supply the amount of reserves that the financial system needs-but differ in the details of how reserves are supplied. Moreover, both of them satisfy the key principles of monetary policy implementation-effective interest rate control, low opportunity cost, and elasticity in reserve supply-outlined earlier by President Williams.²⁷ It's up to policymakers to choose the one they think is most suitable to their jurisdictions, based on their own policy considerations.

With that, the conference comes to an end. Let me thank again all the staff involved in the organization as well as the other speakers and participants for contributing to the success of this 12th U.S. Treasury Market Conference.

Presentation

¹ I would like to thank Brian Gowen and Eric LeSueur for their assistance in preparing these remarks, and Amulya Natchukuri and Navya Sharma for their assistance with the presentation.

² The Joint Member Agencies are the U.S. Department of the Treasury, the Board of Governors of the Federal Reserve System, the Federal Reserve Bank of New York, the U.S. Securities and Exchange Commission, and the U.S. Commodity Futures Trading Commission. In particular, I would like to acknowledge those that supported the logistics of today's event, including Veronika Batista, Ellen Correia Golay, Trevor Graney, Julie Hennighausen, and Hannah Ueshiro.

³ See discussion of the objectives that the Joint Member Agencies set for the Treasury market in Recent Disruptions and Potential Reforms in the U.S. Treasury Market: A Staff Progress Report, November 8, 2021.

⁴ Specifically, the Federal Open Market Committee (FOMC) has directed the Desk to "[w]hen appropriate, increase the System Open Market Account holdings of securities through purchases of Treasury bills and, if needed, other Treasury securities with remaining maturities of 3 years or less to maintain an ample level of reserves." See, for example, Board of Governors of the Federal Reserve System, Implementation Note issued September 16, 2026.

⁵ "Ample" refers to that range of reserves that makes the federal funds rate only modestly sensitive to changes in reserves. See, for example, Roberto Perli, Reflections on the Early Days of Reserve Management Purchases and the Maintenance of Ample Reserves, remarks before the Money Marketeers of New York University, New York City, March 26, 2026, and Julie Remache, Monetary Policy Implementation in an Ample Reserves Regime, remarks at the Fixed Income Analysts Society, Inc., Women in Fixed Income Conference, Federal Reserve Bank of New York, New York City, February 12, 2026.

⁶ See, for example, Roberto Perli, Repo Market Structure and Monetary Policy Implementation, remarks at The Future of Market Liquidity and Functioning Workshop, Federal Reserve Bank of New York, New York City, July 9, 2026, and Perli (March 2026).

⁷ See U.S. Department of the Treasury, Treasury Announces Marketable Borrowing Estimates, August 3, 2026.

⁸ Recall that, all else equal, changes in the Fed's non-reserve liabilities, which include the TGA, have an equal and opposite effect on reserve balances.

⁹ I have discussed these indicators in past speeches, for example: Roberto Perli, Balance Sheet Reduction: Progress to Date and a Look Ahead, remarks at the 2024 Annual Primary Dealer Meeting, Federal Reserve Bank of New York, New York City, May 8, 2024.

¹⁰ See Perli (March 2026).

¹¹ Per the FOMC's directive, the Desk continues to reinvest principal payments from agency mortgage-backed securities (MBS) into Treasury bills each month. So even though the Fed is not currently growing the SOMA portfolio, the Desk is still active in the Treasury market via these purchases.

¹² Roberto Perli, Reserve Management and the Fed's System Open Market Account: Recent Experience and Insights from Surveys, remarks at the Federal Reserve Bank of Atlanta 2026 Financial Markets Conference, Fernandina Beach, Florida, May 19, 2026.

¹³ See summaries of past surveys at Board of Governors of the Federal Reserve System, Senior Financial Officer Survey.

¹⁴ See Depository Trust & Clearing Corporation, Notice of Actions Taken Under DTC Rule 38 - Market Disruption and Force Majeure and NSCC Rule 60, June 30, 2026.

¹⁵ See Bank of England, Transitioning to a repo-led operating framework, December 9, 2024, and European Central Bank, Changes to the operational framework for implementing monetary policy, March 13, 2024.

¹⁶ Other central banks also operate similar systems, including the Bank of Canada, the Reserve Bank of Australia, and the Reserve Bank of New Zealand.

¹⁷ Sometimes these alternative operational approaches are referred to as "demand-driven," whereas the Fed's approach is referred to as "supply-driven." This distinction is a bit artificial since meeting reserve demand is critical in any ample reserves framework, regardless of operational approach.

¹⁸ See discussion of the Fed's SRPs in Christian Cabanilla, Roberto Perli, and Julie Remache, Standing Repo Operations in the Federal Reserve's Monetary Policy Implementation Framework, The Teller Window, December 23, 2025.

¹⁹ Both the Bank of England and ECB are in the midst of transitioning their operating frameworks, and thus repo operations still comprise a relatively small portion of their overall balance sheets. As securities holdings continue to decline, both central banks expect their repo operations to grow in size and share. See additional information on their balance sheets at Bank of England balance sheet and weekly report and Consolidated financial statement of the Eurosystem.

²⁰ Unlike the Federal Reserve, the Bank of England marks its securities holdings to market on a quarterly basis, and losses are indemnified by the U.K. Treasury and are therefore borne by the public close to real time. See Financial relationship between HM Treasury and the Bank of England: memorandum of understanding, February 2025, as well as Andrew Bailey, The importance of central bank reserves, lecture at London School of Economics, May 21, 2024, and Victoria Saporta, Let's get ready to repo!, remarks at the Association for Financial Markets in Europe, London, July 22, 2024.

²¹ As of August 2026, outstanding U.K. Treasury bills represented around 4 percent of all gilt securities outstanding. For comparison, U.S. Treasury bills currently represent around 22 percent of all Treasury securities outstanding. Note that earlier this year, the U.K. Treasury and Debt Management Office ran a consultation on expanding and deepening the U.K. T-bill market. See The UK Treasury bill market: Consultation response, May 5, 2026.

²² See Isabel Schnabel, The Eurosystem's operational framework, remarks at the Money Market Contact Group meeting, March 14, 2024.

²³ See John C. Williams, On the Optimal Supply of Reserves, remarks at the New York Fed - Columbia SIPA Monetary Policy Implementation Workshop, Federal Reserve Bank of New York, New York City, May 22, 2025, and Gara Afonso, Gabriele La Spada, Thomas M. Mertens, and John C. Williams, The Optimal Supply of Central Bank Reserves under Uncertainty, Federal Reserve Bank of New York Staff Reports, no. 1077, November 2023, revised December 2025.

²⁴ There are 47 bank counterparties for our SRPs, and a much broader set of banks eligible to borrow from the discount window. As of June 2026, bank SRP counterparties account for around 60 percent of reserves and about 80 percent of SRP-eligible collateral held by the banking system.

²⁵ See, for example, Perli (July 2026)

²⁶ See, for example, Perli (May 2026).

²⁷ See John C. Williams, Do You Remember?, remarks at 2026 U.S. Treasury Market Conference, Federal Reserve Bank of New York, New York City, September 22, 2026.