It is a pleasure to join today’s workshop and this panel on monetary policy implementation around the world.1 I would like to thank everyone at the New York Fed and Columbia University’s School of International and Public Affairs (SIPA) who helped organize today’s event.
As the System Open Market Account (SOMA) Manager, it is my responsibility to support the implementation of monetary policy at the direction of the Federal Open Market Committee (FOMC). In my remarks today, I will provide my perspective on the Federal Reserve’s monetary policy implementation framework and some observations related to the importance of making our facilities—specifically, the Standing Repo Facility (SRF)—more effective. I will conclude by highlighting both some differences in major central banks’ frameworks and their ultimate similarities.
Before going further, I will make the usual disclaimer that these remarks reflect my own views and do not necessarily represent those of the New York Fed or the Federal Reserve System.
Background on the Federal Reserve’s Operational Framework
I’ll begin with some background on the Federal Reserve’s operational framework and its performance.
In January 2019, after a series of discussions around long-run frameworks for monetary policy implementation, the FOMC communicated its intention to maintain an ample supply of reserves, affirming the framework that, de facto, had been in place since the Global Financial Crisis (GFC).2,3
Ample reserves frameworks are often referred to as floor systems because, when reserve supply is large enough, the main influence on short-term market rates is the floor established under them by the central bank’s administered rates, and the active management of reserves is not needed. This contrasts with corridor systems or scarce-reserves frameworks, like the one used by the Federal Reserve before the GFC, which involve controlling the policy rate through active management of the supply of reserves based on short-term forecasts of reserve demand and so-called autonomous factors.4
An ample reserves framework has several advantages over a scarce reserves framework in the U.S. context.5 One of these is solid control of the policy rate—in fact, since the formal adoption of the current ample reserves framework in January 2019, the effective federal funds rate has been outside of the target range on only one day, in September 2019 (Panel 1). A second advantage of an ample reserves framework is that it promotes the efficient transmission of the policy rate to other money market rates and the stability of short-term funding markets, which in turn is essential to the smooth functioning of the Treasury market and the financial system in general. And a third advantage is that it is robust to changes in economic and financial market conditions—for example, rate control has been very strong despite major shocks like the onset of the COVID-19 pandemic in 2020, the banking sector stress in 2023, and, to stick close to the present, the broad market volatility we all witnessed last month. Importantly, this robustness applies also to times when economic or market conditions necessitate large expansions of the Fed’s balance sheet.
By contrast, the FOMC concluded that a scarce reserves system, which had served the Federal Reserve well in the past, was likely to present notable disadvantages for the U.S. in the post-GFC environment. Specifically, because of structural changes, the level and variability of reserve demand and supply became much larger, and forecasts of those variables consequently would be prone to correspondingly larger errors. Maintaining strong rate control in this environment would be very difficult and would require large and frequent open market operations, and we would likely experience substantial variability in the effective federal funds rate and other money market rates, thus adding uncertainty in funding markets.
The reality is that reserve balances are higher and more variable than they used to be, and while the minimum baseline level of demand remains uncertain, banks tell us in outreach and in surveys (such as the Senior Financial Officer Survey, or SFOS) that their reserve demand has increased materially in recent years, reflecting prudent liquidity management, growing intraday liquidity needs, and regulatory and supervisory considerations.6
Meanwhile, other non-reserve liabilities have also expanded a lot and in some cases have become subject to considerable fluctuations for different reasons. Currency has increased over time, reflecting nominal economic growth and international demand for U.S. dollars, among other factors.7 The Treasury General Account (TGA) has also grown substantially—in line with the expansion of U.S. debt outstanding and the U.S. Treasury’s stated target for the TGA to cover about one week of net cash outlays and maturing debt—and especially has become more variable.8 By way of example, we are currently in the midst of a debt ceiling episode, which—if past experience is a guide—could see the TGA fall further before rapidly increasing by half a trillion dollars or more over the course of a few months. Maintaining an ample, rather than scarce, level of reserves is a proven and effective way to accommodate such large and sometimes unpredictable shifts in these items.9
There is no doubt that reserves rose to an abundant level—or well above ample—in the post-pandemic period. This was driven by the large asset purchases that were necessary to restore market functioning in Treasury and agency MBS markets and support the economic recovery while the federal funds rate was constrained by the effective lower bound.
Since June 2022, the FOMC has instructed the Open Market Trading Desk at the New York Fed (the Desk) to reduce the size of the balance sheet by letting securities run off at a predictable pace, subject to monthly redemption caps (Panel 2 and 3).10 Over this period, the SOMA portfolio has been reduced by $2.1 trillion thus far. While all indications are that reserves remain abundant and the Federal Reserve’s balance sheet can continue to shrink, there are limits.
On the liabilities side, while I expect the overnight reverse repo (ON RRP) facility to dwindle to minimal levels once the debt limit situation is addressed, other liabilities are likely to remain sizeable. Currency stands today at nearly $2.4 trillion. Late last year, before the debt limit was reached, the TGA fluctuated between $700 and $800 billion. Deposits from designated financial market utilities, government-sponsored enterprises, and other entities, plus reverse repurchase agreements with foreign international and monetary authorities, at present total around $600 billion.
As I said, it’s hard to estimate exactly what constitutes an ample level of reserves. We know that number is lower than the current $3.2 trillion, since market indicators still point to reserves remaining abundant. But we also know it is substantially larger than the $1.5 trillion level to which reserves dipped in September 2019, when they evidently became less than ample. Since then, the U.S. economy, bank assets, and the size of money markets have all grown a lot, and SFOS respondents tell us that banks have increased their desired reserve levels and buffers materially since that time. In addition, the Fed will need to maintain an operational buffer to account for uncertainty and ensure that the supply of reserves remains ample.
Given this, under the current FOMC plans, the balance sheet will eventually cease to shrink, and when reserves reach the ample level, it will at some point start growing again to meet the likely growing demand for Federal Reserve liabilities. In other words, the longer-run size of the Federal Reserve’s balance sheet will be determined by the demand for reserves, currency, and deposits by the Treasury and other entities, all of which are not under the Federal Reserve’s direct control. Even so, some details related to our implementation tools can make a difference to reserve demand.
Implementation Tools
Under the ample reserves framework, the Federal Reserve achieves rate control via administered rates that work together to maintain the fed funds rate within the target range indicated by the FOMC. Two administered rates set the floor for the fed funds rate. The interest paid on reserve balances (IORB) establishes a benchmark against which banks assess short-term lending and borrowing opportunities. The overnight reverse repo facility (ON RRP) allows certain counterparties—most notably money funds—to place cash at the Fed at the bottom of the target range, reinforcing the floor across a broader set of market participants.
The Federal Reserve also administers standing liquidity facilities that, as a counterpart to the ON RRP, help to set a ceiling on the fed funds rate by allowing counterparties to access overnight funding.11 A key facility, which works alongside the discount window to help ensure good rate control, is the SRF. The SRF dampens upward pressure on money market rates and supports money market functioning by allowing primary dealers and certain depository institutions to obtain liquidity from the Federal Reserve.12
So far, the floor tools—IORB and the ON RRP facility—have been very effective in enabling the Federal Reserve to maintain control of the federal funds rate and in enabling good transmission to money market rates.13 The SRF is a comparatively newer tool, introduced in 2021, and it has not been used in meaningful size because repo rates in the tri-party segment of the market, where the facility operates, have for the most part remained below the SRF minimum bid rate. An important reason for that is that reserves have remained at abundant levels, which naturally keeps some downward pressure on repo rates.
As the size of the Fed balance sheet continues to decline, however, and as reserves transition from abundant to ample levels, upward pressure on money market rates is likely to increase. We are starting to see the early signs of this in the repo market, especially around key reporting dates, following a period when the supply of reserves was so far in excess of demand that repo rates hardly moved at all (Panel 4). This represents a normalization of liquidity conditions and is not a cause for concern; however, it does imply that, in the future, the SRF is likely to be more important for rate control than it has been in the recent past.
New York Fed President John Williams showed very clearly today that the presence of a facility like the SRF can help reduce the amount of reserves that a central bank needs to supply to operate efficiently in an ample reserves regime.14 The more frictionless the facility is, the more effective it will be, and the lower the reserve buffer needed to account for the uncertainty that is inherent in monetary policy implementation.
The SRF is an important facility, but it is not frictionless. Our SRF counterparties have mentioned some features of the SRF that discourage its usage even if market rates exceed its minimum bid rate. One such friction is the fact that SRF auctions are carried out and settled in the afternoon, while most repo market activity takes place in the morning. Afternoon auctions are important because they can provide cash later in the day should a need for liquidity arise when most repo market activity has already taken place. Still, that does not mean additional earlier SRF auctions cannot be helpful. In fact, as I mentioned nearly two weeks ago, in the not-too-distant future the Desk will start implementing daily morning SRF operations that will also be settled in the morning.15 This will be an important step in enhancing the efficacy of the facility, and, at the margin, it can contribute to a smaller SOMA portfolio than would otherwise have been the case.
Still, other frictions remain. A prominent one in the mind of our counterparties is the inability to take advantage of balance sheet netting when borrowing from the SRF. The uncertainty of award allocations, which is related to the size limits of the facility, is also mentioned frequently. These frictions add to the costs that counterparties face when using the facility and mean that counterparties generally require private market repo rates to trade materially above the SRF minimum bid rate before using the facility. We saw evidence of frictions in SRF usage over a few days in December 2024, when a substantial amount of tri-party repo transactions—for which the SRF should be a close substitute—took place in the market at rates above the SRF minimum bid rate.
So, there is room to improve the effectiveness of the SRF—this is why we are continuously evaluating its design and parameters. We will move forward with additional morning-settling auctions soon, as I said. But, where possible, the Desk and other Federal Reserve staff will continue to look for ways to address other remaining frictions.
By the same token, I encourage our counterparties to use the SRF when it makes economic sense—the facility is there to support the effective implementation of monetary policy and smooth market functioning. Those are the Federal Reserve’s goals, though I have no doubt that they are shared by our counterparties. In other words, it’s in everyone’s best interest if the SRF works as intended.
Other Approaches to Monetary Policy Implementation, and Conclusion
The Fed’s operational framework is working well and has proved highly effective in controlling short-term interest rates. But it’s not the only approach that works, as my colleagues on this panel can—and I’m sure will—attest. In particular, while most advanced economy central banks operate floor systems, the implementation details vary across jurisdictions. This is most notably the case for the Bank of England and the European Central Bank, which operate what have become known as “demand-driven floors.” The key conceptual difference between the Fed’s implementation framework and a demand-driven floor is that the Fed determines the quantity of reserves to supply in order to ensure that reserves remain ample—of course taking into account banks’ demand for reserves—whereas in a demand-driven floor commercial banks determine the marginal supply of reserves directly. This is typically done by the central bank offering full-allotment repo operations at a fixed rate closer to market rates, which commercial banks can then draw upon as much or as little as they desire.
Both approaches can deliver strong rate control and support smooth functioning in markets, but tend to differ in other respects, including the composition of the central bank’s balance sheet. Repo operations are a key element in the implementation of demand-driven floors, which naturally leads to repos comprising a substantial share of assets for central banks that operate such frameworks. By contrast, the Fed’s main tool for injecting reserves is outright security purchases, and repos typically make up a very small share of the Fed’s assets, or often no share at all. At the same time, I don’t want to suggest that this difference is larger than it actually is—central banks operating demand-driven floors typically also hold substantial outright bond portfolios, to meet at least part of the steady-state demand for their liabilities. Demand-driven repos constitute a key part of our framework here at the Fed, most notably in the form of the SRF, as I have just discussed.
Neither approach is inherently superior to the other, and choices around operating frameworks seem to relate to how policymakers weigh competing aims and to jurisdictional differences between financial systems, including the size and complexity of banks and nonbank financial institutions. But, at the end of the day, despite the technical differences, floor systems may be more similar than they first appear since they all allow central banks to accommodate demand for their liabilities and maintain strong rate control. The diversity of frameworks and approaches creates the opportunity to learn from one another’s experiences, including through our discussion today.
Thank you.
