-
Conference information
- Conference overview
- Agenda
May 20, 2025
Policy Session 4: Central Banks' Operational Framework—Putting Monetary Policy into Practice
The fourth policy session of the 2025 conference explored revisions central banks have made to their operational framework for implementing monetary policy. Moderator David Beckworth, senior research fellow at the Mercatus Center at George Mason University, lead a panel discussion that included Claudio Borio, former head of the Monetary and Economic Department at the Bank for International Settlements, Imène Rahmouni-Rousseau, director general of market operations at European Central Bank, and Patricia Zobel, head of macroeconomic research and market strategy at Guggenheim Partners.
Transcript
David Beckworth: Good afternoon, and welcome to Policy Session 4 on central bank operational frameworks. I'm David Beckworth of the Mercatus Center, and I have the privilege of moderating this panel. It's going to be a fascinating and fun-filled one, with some great guests.
Here in the United States, the debate around the Fed's operating system has centered on the pros and cons of a floor versus corridor system—or, alternatively stated, ample versus scarce reserve operating systems. These discussions have focused on the implications for interest rate control, market functioning, fiscal costs, and the size of the central bank balance sheet.
Internationally, though, a different path is emerging. A growing group of central banks is charting a new course toward what can be broadly defined as demand-driven operating systems. Three key central banks—the Bank of England, the European Central Bank, and the Reserve Bank of Australia—are leading this charge. They each use different names for their frameworks, but they have three common characteristics.
First, the marginal unit of central bank liquidity comes from a standing repo or lending facility. Liquidity is demand driven, and therefore determined by the banks. Number two, the spread between the central bank's lending rate and deposit rate is narrow, typically 10 to 5 basis points. And third, the targeted overnight rate is effectively anchored by the lending or repo facility rate.
This represents a fundamental rethinking of how central banks supply reserves. In a supply-driven system, like a traditional floor or ample reserve system, the central bank preemptively injects reserves into the banking system, often in large quantities, to ensure the banks are fully satiated with liquidity.
In a demand-driven system, by contrast, the central bank does not guess or target a fixed reserve quantity. Instead, it stands ready to supply reserves in response to bank demand, typically via repo or lending operations. Put differently, in the supply-driven operating system, liquidity is explicitly ample, and in a demand-driven system, liquidity is latently ample; it's there for the taking.
Although none of the central banks now pursuing this demand-driven approach would use this term, one can think of this approach—in the limit, taken to its full extent—as a demand-driven ceiling system. Now, no central bank has reached a true demand-driven ceiling system status, but the Reserve Bank of Australia has given us a glimpse of what this journey might look like.
On April 2nd, they announced they would no longer be reporting the interest rate on settlement balances, or what we call "reserves" here in the US. Going forward, the open market operation repo rate, their lending facility rate, will effectively now anchor their target policy rate, the cash rate. This would be akin to the Fed no longer publishing interest rates on reserves, and turning to the discount window or standing repo facility rate as the anchor rate for overnight market rates.
So, it is quite a remarkable development that we are seeing in other central banks as they begin this journey from a supply-driven operating system to a demand-driven operating system. This development raises lots of interesting questions, such as: Why is this development happening now? Why, after most central banks turned to ample reserves or floor systems during and after the Great Financial Crisis, are they now making a turn to a different system altogether?
Also, is this development the beginning of a bigger trend, akin to the adoption of state inflation targets in the early 1990s, when a number of similar central banks began a new journey towards inflation targeting, and eventually everyone followed? Additionally, how is this playing out in the biggest of the central banks taking on this journey—the ECB?
Finally, and for this crowd, what does this mean for the Federal Reserve? What if at some point, Fed officials decided they wanted to explore a demand-driven approach to their operating system? How could they do it? Is it even possible, given the facilities that we have?
We have three great guests with us today to help us answer these and other questions, including yours from the audience. Our first panelist is Claudio Borio. Claudio is the former head of the monetary economic department at the Bank for International Settlements and has written extensively on central bank operating systems.
Second, we have Imène Rahmouni-Rousseau. Imène is the director of general market operations at the European Central Bank and is in charge of implementing the single monetary policy for the euro area. In other words, she is leading the charge with this new demand-driven operating system.
Third, we have Patricia Zobel. Patricia is a senior managing director and head of the macroeconomic research and market strategy team at Guggenheim Partners Investment Management. Prior to joining Guggenheim, Patricia served as the deputy manager and manager pro tem for the Federal Reserve System Open Market Account bond portfolio, and as a senior adviser on financial markets.
We'll now turn it over to our panelists. They will each have some time to give their remarks, and then we'll go to the questions. Claudio, you're up.
Claudio Borio: Thank you, David; can you hear me?
Beckworth: Yes.
Borio: Okay. First of all, I'd like to thank the organizers for their very kind invitation, and I would really like to apologize for the fact that, unfortunately, because of personal reasons, I can't be with you today—but at least I can participate by video. So, this session, as David mentioned, is on operational frameworks, central bank operational frameworks—which is really about how central banks go about influencing closely a market interest rate (you can think of it as their operating target, which is typically an overnight rate) to make effective their policy stance.
Now, the topic had limited interest before the Great Financial Crisis, but has now shot to prominence. For one, frameworks were upended by the crisis, and by the expansion of excess reserves that financed a huge increase in central bank balance sheets. And more recently, as we heard, a number of central banks have been reviewing frameworks—including the ECB, the Bank of England, and the Fed.
Now, the issue, as we will hear even more during this presentation, is really highly technical. It deals with the nuts and bolts of day-to-day operations; it's not well understood outside narrow circles—and partly as a result, I think, its policy significance is often underappreciated. Indeed, I think that how central banks go about this task has implications for market functioning, for the transmission mechanism, for financial stability—and also for the economics and political economics of central bank balance sheets.
So, what I will do in my initial remarks is offer a specific lens to understand the issues involved, and provide a personal perspective on the merits of alternative systems. Now, you will hear a lot these days about corridor versus floor systems, supply-driven versus demand-driven floors, abundant versus ample reserve systems; clearly, central banks are still trying to develop a common language.
Analytically, however, I would argue that the key distinction that helps us understand the rest is indeed between scarce reserve systems and abundant reserve systems; and I will not make an analytical distinction between abundant and ample. In all systems, the supply of reserves has to be equated to the demand for reserves, allowing the market to clear at an overnight rate that the central bank wants to see. But in a scarce reserve system, the central bank limits the supply of reserves to settlement needs of banks so that the overnight rate, the operating target, is above the deposit facility rate (or zero, if there isn't any), so that—and this is the key—there is a clear opportunity cost to holding reserves.
In an abundant reserve system, by contrast, the central bank increases the supply of reserves so that the overnight rate falls to the deposit facility rate (or zero; that's why it's called a floor system). And in fact, typically it is actually below the deposit facility rate, if some market participants don't have access to the deposit facility of the central bank, hence the "leaky floor" (and I will come back to it later).
As a result—and this is the key—the opportunity cost of holding reserves vanishes. So, the analytical implication is that in a scarce reserve system, the demand for reserves is exclusively a demand for settlement purposes, including any additional demand for reserve requirements to meet supervisory liquidity requirements, and therefore related precautionary needs.
By contrast, in an abundant reserve system, the demand is also—and crucially—at the margin; it is only a demand for store of value, or investment purposes. In other words, the features of the demand for money for the reserves that matter are system or regime dependent; and, in this sense at least, demand is endogenous to supply.
Now, the implication for the functioning of the system is that in a scarce reserve system, demand is, in aggregate, hardly responsive to the interest rate. Banks cannot afford to miss payments or to violate the various requirements. And the heavy lifting to anchor the overnight rate—and this is very often not appreciated these days—is done by the central bank signaling to market participants where the interest rate should be. And the power of the signal comes from the central bank being a monopoly supplier of the settlement balances, so that the central bank could always actually peg the interest rate by forcing banks to borrow—by inducing a system-wide deficit—at the rate it wants to see.
Now, importantly in this framework, the central bank liquidity management operations played purely a supporting role, designed to ensure that the supply of reserves meets the demand for settlement balances, not to influence yields as such. In an abundant reserve system, by contrast, the demand for reserves is highly responsive to risk return considerations, because the risk return profile of reserves must be sufficiently attractive for banks to be willing to hold them—that is, the relative yields become an additional variable in this demand function.
So, put differently: in terms of the transmission mechanism, in a scarce reserve system, the overnight rate does not respond to exogenous changes in yields or interest rates in the system, but it does in an abundant reserve system. Now, following the Great Financial Crisis, these days if you look, say, at the G3 central banks, all of them operate versions of an abundant reserve (or floor) system. This is also true of the UK, and it is also true of Canada. This is because, following the Great Financial Crisis, central banks satiated banks' need for reserves.
The difference is what they plan to do now. Now, the BOJ has not announced any specific plans; the Fed would like to remain on the floor, but reduce as far as possible the supply of reserves (what they mean by moving from abundant to ample), and the ECB has argued, as we just heard, that it will move to a demand-driven floor.
Now, in the framework that I outlined, the ECB is an interesting case; and this is similar in some respects also to Canada, Australia, and the UK. But particularly in the case of the ECB, the description as "floor" and the emphasis on the deposit facility rate as the policy rate, the key policy signal, sounds very much like an abundant reserve system. But the mention of an opportunity cost to hold reserves sounds like a scarce reserve system.
Now, personally, I suspect that over time, the system will gravitate towards one of these two polar cases. Because in my view, the regimes that operate in the green zone are likely to be unstable, as the demand for reserves is unstable in this borderline region. And the observed interest elasticity can switch suddenly in response to exogenous changes in interest rates in the system, and it could be also in response to very rather idiosyncratic market frictions, including issues that have to do with distribution of reserves in the system.
Moreover, since the central bank provides liquidity on demand at a fixed price, at regular intervals, in unlimited amount, in effect the central bank is delegating to the banks the task of forecasting autonomous factors, as well as other banks' demand for reserves—because what really matters in equilibrium is the aggregate demand for reserves, relative to the supply. And complicating matters further is the possibility of borrowing from the central bank for general funding purposes, completely unrelated to any demand for reserves or any equilibration in the reserves market (which effectively acts almost like another autonomous factor).
Now, this is what happened to the central banks that engaged heavily in QE. Those that did not do that actually stayed within a scarce reserve system, and this is the majority of central banks around the world, including many emerging market economies like Mexico, Brazil, Korea, India, Turkey, Thailand, and China, just to mention a few.
Now, the key difference between the two systems is that in an abundant reserve system—and here, I'm moving towards the assessment of the two systems—there is no need for banks to transact with each other to redistribute reserves and to clear the market. And because they hold reserves also as a store of value, the distribution of reserves can become a factor influencing market rates.
By contrast, in a scarce reserve system—in normal times, at least—the central bank cares only about the aggregate volume of reserves, and the interbank market takes care of the redistribution; that is, the fragmentation in the interbank market that one very often sees in abundant reserve systems is, in fact, to some extent, endogenous to the system itself. But of course, even in a scarce reserve system, when the market is under stress, the central bank will have to intervene and take over and provide liquidity and direct it where it's needed (and this is very much the lender of last resort function).
So, the key question that everyone is asking is, "Does inhibiting the redistribution function of the interbank market matter?" And here, there are two views. One that basically says no, it doesn't matter; it's just churn. The other one, to which I tend to lean, actually says yes, it does matter; and one reason—not the only reason, but one reason—is that inhibiting the redistribution function could, over time, actually contribute market stress episodes.
Why do I say that? Well, simply because if you don't use a muscle, it atrophies. Banks have less need to monitor each other in good times, and institutional knowledge is lost. This has two consequences. First, when stress arises for whatever reason, the central bank will have to do more of the heavy lifting, but starting from a higher amount of reserves. And second, possibly even the probability of stress could be higher for at least three reasons.
First, banks are less able to deal with situations in which redistribution is needed (think, for example, September 2019 in the US); second, reserves may look deceptively abundant because the demand for reserves, given its store of value function, depends on exogenous changes in market-determined opportunity costs (for example, repo spreads could move for any reason that is completely unrelated to the demand for reserves); and finally, banks may take on more liquidity risk endogenously (for example, as Raghu Rajan and Vera Aarya have argued; that's one possible mechanism).
Indeed, one could see a ratcheting up of demand for reserves in these systems. If you take the United States, for example (and I think you're familiar with this), the estimates of the demand have ballooned, going from something like $35 billion in April 2008, to something—I think the latest estimates are in the region of $3 trillion. And in Norway, a central bank that actually adopted this system quite early; it modified it, taking a step back, introducing a tiered system—precisely because it had seen this demand for reserves ratcheting up, and it was not happy with that.
Now, if you are concerned about this, you would also be concerned about another property of abundant reserve systems, and that is that they tend to be self-entrenching, and this is for two closely related factors. The first is that they give the impression that a scarce reserve system would be hard to operate (let me stress "impression"). And the second is that they generate conditions that actually do hinder the transition back to that system.
Let me just say a few words about each, and I will finish with this. First, the impression. Well, an abundant reserve system is the perception of an exogenously higher volatility or unpredictability of the demand for reserves—that, in fact, is, to some extent, endogenous to the system. Think, for example, of the fragmentation of the interbank market that I mentioned earlier, or of the unpredictability due to the new regulatory and supervisory liquidity requirements.
Now, there is no question that such requirements increase the overall demand for reserves; but they do not necessarily make it more unpredictable, just as reserve requirements don't, and the reason is simple. In a scarce reserve system, banks would economize as far as possible on those requirements, because of the prohibitive opportunity cost (and indeed, the evidence suggests this). And moreover, let me stress again, most central banks still operate scarce reserve systems, especially among emerging market economies, even once they have adopted these various liquidity requirements.
Now, what about the second factor, the idea that it complicates the transition back to a scarce reserve system? Well, for one, as time passes, institutional memory is lost; not just in the banks, actually, but also in the central banks themselves. Sometimes when I talk to people there, I have the sense that there is a certain fear of floating in a different context than in the one about exchange rates; but it's very similar.
In addition, an abundant reserve system may—let me stress "may"—increase the volatility of autonomous factors that need to be forecast in a scarce reserve system. Let me mention just one, the "leaky floor" that I was talking about earlier—because one solution to deal with the leaky floor is to introduce a facility open to nonbanks to place funds with the central bank on demand, and this is what happened in the US and also in New Zealand.
Now, this in fact introduces a new autonomous factor—and by the way, one that introduces the possibility of a run on the banking system as a whole, which is not possible under the scarce reserve system. But let me stop there.
Beckworth: Thank you, Claudio. We'll now go to Imène.
Imène Rahmouni-Rousseau: So, thank you; thank you, David—and thank you very much for the invitation to speak in this very topical conference. The normalization of central bank balance sheet continues, and as it's progressing, it poses unique challenges—but also opportunities—for the implementation of monetary policy.
Since the Great Financial Crisis, the ECB adjusted the way in which it implements monetary policy. Starting in 2014, it decided to conduct large-scale asset purchases, resulting in a significant increase in the size of its balance sheet. This increase accelerated in 2020, with the introduction of another wave of asset purchases linked to the pandemic, as well as longer-term repo operations.
The chart nicely illustrates the rapid rise in excess liquidity—the red line—peaking in the middle of '22. Once the worst of the pandemic was over and the stance of monetary policy evolved away from the effective lower bound and shifted to a rapid succession of rate hikes, crisis-related measures were phased out. Longer-term repo operations were repaid, net asset purchases stopped, and then QT was implemented.
As a consequence, our balance sheet started to shrink; and currently, our QT is proceeding smoothly, helped by robust absorption in bond markets. In March '24, as David mentioned, the governing council of the ECB announced changes to its operational framework—or, as I hear here: operating system—to ensure that it remains fit for purpose. I will use the next few minutes to explain the new system, the reasoning behind its calibration and design, and provide a first assessment of its functioning.
The new framework is based on six high-level principles that you have on the screen, with the main objective of maintaining interest rate control. Its design is tailored to the specificities of the euro area, the prominence of bank-based intermediation, the diversity of bank sizes, business models, and the remaining fragmentation risks across jurisdictions. Under the new framework, the ECB will indeed continue to steer the stance of monetary policy through the deposit rate, in what could be described as a "soft floor."
So, why are we calling it a soft floor? Because the governing council decided that it will tolerate some deviation of short-term interest rates from the deposit rate, to the extent that this doesn't blur the signal about the intended monetary policy stance. The soft floor has two key features; first, liquidity will be provided through a mix of instruments, which makes the system more robust.
Our short-term repo operations will play a central role in meeting bank liquidity needs, and provide the marginal unit of liquidity in a demand-driven way. These operations run as fixed rate, full allotment against broad collateral, ensuring that liquidity reaches those banks most in need, in an elastic manner. To complement repo operations, structural operations, longer-term repo operations, and a portfolio of securities will also be introduced at a later stage.
The second characteristic of the soft floor, there is a wedge between the lending rate and the deposit rate in the form of a 15 basis points spread, and the calibration of this spread is a balancing act. The spread is large enough to leave room for money market activity and set incentives to seek market-based solutions. But the spread is also sufficiently narrow, in our view, to limit the scope for volatility, reduce the risk of stigma in the operations, and incentivize the usage of these operations.
You may ask why the governing council opted for a system where the marginal unit of reserves is demand driven. In my view, following years of abundant reserves—and also, in the context of changing bank liquidity regulation, more complex banks' risk management, and evolving payment system landscapes—it's very hard for any central bank to reliably estimate reserve needs.
The beauty of the system we introduced is that it relies on banks themselves determining the exact amount of reserves they wish to hold, and then satisfying this demand through the repo operations. As you can see from the left-hand side here, we expect the Eurosystem balance sheet to continue to decrease for some time, driven by the roll-off of our QE legacy portfolios—and to decrease up to a point where the excess liquidity will have shrunk so much that the banking sector, in aggregate, takes recourse to the repo operations in amounts that exceed the base of our QT. And at that point, the balance sheet will start to grow again.
And it's after we reach this inflection point that you see, at the bottom of the curve, that we will introduce structural operations—the red area here—to cater for the structural liquidity needs coming from autonomous factors and minimum reserve requirements. On the right-hand side, you see that demand for our repo operations is currently low, and that's consistent with the fact that reserves remain abundant; but survey-based information suggests that demand for this operation will pick up in the not-too-distant future, although at a slow pace.
An important condition of success, and that is key to our framework, is that recourse to repo operations needs to be perceived as business as usual by banks, and we have therefore invested a great deal in communicating proactively with market actors. A powerful communication action was the recent publication of a joint blog post between the chair of the supervisory board, Claudia Buch, and the ECB executive board member in charge of markets, Isabel Schnabel; and in this blog post, they jointly reiterate (and I quote), "The ECB, both as a monetary policy authority and as a supervisor, expects that banks should consider central bank repo operations as an integral part of their day-to-day liquidity management."
And in fact, this is nothing new for euro area banks. They used to resort routinely to our repo operations, pre-Great Financial Crisis; and they also took a large recourse to our targeted longer-term operations during the pandemic. The blog also highlights the importance for banks to be operationally ready to source reserves in a scalable manner, both from the market and from the central bank.
As I mentioned earlier, for the moment the usage of our repo operations is limited; and this is perfectly understandable. Excess reserves remain ample, the market is functioning smoothly, and the market pricing is quite benign—so, banks can satisfy their liquidity needs in the market. And I'd like to illustrate this point about pricing here; so, as you can see from the slide, borrowing reserves in the market is currently more attractive today for banks than using our repo operations.
And in the overnight segment, as a matter of fact, borrowing reserves is even done at rates below the deposit rate—so, these are the yellow dots—both in the unsecured market, through interbank borrowing and issuance of CPs, and in the repo market. And that makes the equilibrium of the system, where our 15 basis point spread achieves a sweet spot, where incentives for market activity and control over interest rates are both met.
So, controllability; central banks like to see that money market rates, at which banks lend and borrow among themselves, align well to the rate used to steer monetary policy (and that's the deposit rate, for us). And so far, this is the case, as you can see on the chart.
So, the €STR, which is the benchmark rate for unsecured borrowing in the euro area, is broadly stable and had a near perfect pass-through of all our interest rate decisions. The repo rate—that's the yellow line—has gradually moved up, closer to the deposit rate, since the end of '22. And this reconvergence of repo rates to the steering rate reflects a clear shift from the collateral scarcity period in '22 to collateral abundance, owing to a lower central bank footprint and rising bond issuance.
Banks have responded to the decrease in reserves by increasingly managing liquidity through market activities, as I highlighted; and there, I'd like to make reference to what Claudio said about the importance of redistributing reserves in the market, to show that the repo market is currently the main channel for redistributing reserves in the euro area. And so, banks that are close to their reserve targets borrow in the private repo market—so, they gain collateral—to keep reserves about their desired levels.
However, at the moment, the portion of repo market activity motivated by reserve needs—that's the yellow part here—is still limited; but as we go forward, and given the large and robust size of the repo market, we think that there is significant additional capacity for the repo market to grow in importance, in the interbank fashion, to redistribute reserves. And in case of stress, the demand-driven nature of our repo operations will ensure that any surge in liquidity needs can be satisfied in full, without the need to introduce ad hoc types of operations.
So now, to wrap it up: one year after its announcement, the framework is working as intended and supports the smooth normalization of the size of a balance sheet as QT continues. Banks have been adapting very well, money market rates are well controlled; and so far, at current levels of excess liquidity that are still ample, banks have almost exclusively met their liquidity needs through market sources rather than the central bank.
However—and that's an important caveat—I should note that most banks' desired reserve levels have not been tested yet. And as reserves decline further, it will be crucial to ensure that money markets continue to function smoothly, and that banks are prepared to use the Eurosystem operations (as I said, as business as usual).
Finally, the governing council will review the key parameters of the framework at the end of 2026, and stands ready to adjust them earlier, if necessary, to ensure that the implementation of monetary policy remains in line with the established principles.
Beckworth: Thank you, Imène. Patricia, you're up.
Patricia Zobel: So, maybe while they're pulling that up, I'll say: I wanted to say "thank you" to President Bostic for holding this conference. It's an important time for this conversation, and it's a pleasure to be here at the Atlanta Fed.
What I wanted to do, to add to Claudio and Imène's good remarks today, is highlight how I see the US system fitting into the international context, and some of the innovation going on. I want to leave you with three points. One is that while central banks, broadly, are implementing floor systems, they're innovating to try and keep some of the benefits that floors have while limiting some of the costs; two, I see the US is requiring more standing reserve supply than in other jurisdictions, because of high payment intensity and deep capital markets; and three, but most importantly, the US still needs flexible liquidity for resilience. And this is similar to what Imène talked about, having some demand-driven liquidity in the system.
This slide just shows where central banks are at. Most have been reducing their balance sheets after pandemic expansions, and a lot of progress has been made in that regard. I would note that balance sheets are down quite notably now, as a share of GDP. On the right-hand side, it illustrates that as liquidity declines in the system, central banks are moving away from asset-driven abundance of reserves, and are focusing on how to operate with fewer reserves.
What's interesting, in looking across central banks, is that although they face very different operating environments, in my own assessment there are similarities in what they're trying to achieve—and many central banks are coming to the conclusion that floor-type systems are generally meeting their principles.
Still, the emphasis can be different across central banks; and looking on this list, effective, robust rate control and adequately efficient balance sheets are broadly shared principles. But I would say that flexibility has been more emphasized in the ECB's new framework, and it's gaining importance in the US dialogue right now. Interbank trading is something that I see as less favored, and less important to the US dialogue, in terms of its framework.
Still, there are drawbacks to floor systems, too. There's no free lunch in life. I like to separate out drawbacks related to steady state floor systems from ones that are really related to the effects of asset purchases. So, for example, on the right-hand side, studies find that large-scale liquidity injections from asset purchases can ratchet reserve demand; central bank income can also be volatile, after central banks take duration risk to support the economy during a downturn; and some believe floors leave central banks open to political influence by making asset purchases easier.
Some of these may be valid issues, but they seem to me to be less salient to the choice of operating framework, as they aren't features of steady state floor systems. And if truly needed, I sense that central banks would use asset purchases irrespective of the operating framework that they had in place; they did so during the GFC, to prevent a financial panic from becoming a deep, deep downturn.
So, central banks have generally determined that more reserves are needed in the post-GFC world, to meet higher precautionary settlement demand and reduce payment system risk. I just wanted to touch briefly on the pre-GFC system, when the US also operated with scarce reserves. Reserves in the system at that time were around $20 billion, including clearing balances; and banks ran large, intraday overdrafts to settle payments—they weren't settling them with reserve balances—and they also had high interbank trading in order to meet their payments and get flat at the end of the day.
During the GFC, I think what we saw is that these added to the run dynamic in markets, and increased stress in the system. The counter to that, though, is just as Claudio said: As you increase the reserves in the system and you move to a place where there's no longer a cost of holding reserves, you do see less liquidity distribution among banks. And the right-hand side chart shows that US bank lending—in interbank markets, at least—is quite small today.
And this can create frictions to the liquidity redistribution that Claudio was talking about. So, central banks are innovating; they're creating systems intended to supply reserves in an environment where the demand is both uncertain and elevated, and hopefully they're trying to limit some of the drawbacks of traditional floor systems.
For the next couple of slides, I just wanted to level-set briefly on supply-driven floors. What you see here is that in a steady state floor system, central banks supply enough reserves to keep market rates near the interest on reserve balances. So, on this chart it's no less than the area in green. This is consistent with the Friedman rule, which suggests that a safe, liquid financial system is a public good, and charging a premium for liquidity isn't socially optimal.
If everything is stable in the system, the issue of how to supply central bank supply reserves wouldn't be that interesting, really, and we might not be having the panel today; but in reality, reserve supply (in blue) is shifting around daily as non-reserve liabilities—like the TGA—fluctuate, and liquidity demand (in the dark line) also changes over time. Supply-driven floors rely on permanent securities holdings, so they can maintain standing reserves at the outer edge of the dark blue line, to keep fluctuations in non-reserve liabilities from pushing rates higher. It requires, as both Imène and Claudio mentioned, a higher level of asset holdings.
And then as noted, demand-driven floors instead intend to supply reserves flexibly through full allotment lending operations, and this limits the need to estimate the demand for reserves or allocate extra supply for fluctuations. So, with that in mind, for the next couple of slides I wanted to highlight why I think the US benefits from higher standing reserves, and that migrating to a demand-driven floor wouldn't be appropriate in the US context.
First, the payments intensity of the US system is extraordinary; as shown on the left, Fedwire settles transactions each year equal to almost 40x GDP—and when you include CHIPS, it's around 60x. And because of the US's deep and liquid securities markets, there are large peaks and troughs—in particular, for Treasury trading and settlement.
At the same time, the supply of reserves also swings around. The dark dots show that other liabilities can move sometimes over $300 billion in one week, shifting reserve supply. With large and less predictable macro liquidity flows, it would seem challenging to me for banks individually, understanding how much they should take from an operation each week in the context of where all other banks are making that decision. Will they make the right decision for the system as a whole, about the amount of liquidity that's needed?
The US also operates with narrow counterparties for open market operations. This is a historical vestige of the US having separate securities—firms and banks—and this has worked well when needed, but it's not necessarily well suited for a demand-driven floor. The ECB, on the other hand, operates with the banking system directly, against a broad range of collateral—so a broad range of banks can access liquidity directly, moderating the need for markets or a system like FHLB's to redistribute liquidity.
The US system—and it's been recommended—could be set up differently; looking at the gold bar on the right, the Fed could actively operate through banks. TAF auctions were held during the Global Financial Crisis; and while this could be quite beneficial, though, to me it seems unlikely to be sufficient for a demand-driven floor, given the large non-bank sector in the United States. I think it would be helpful at distributing liquidity, but not necessarily sufficient.
Finally—and most importantly—US money markets are highly complex. The US Treasury market depends on the smoothly functioning funding markets, but US repo markets are segmented and less cleared, with trading through a narrow group of dealers to a large group of non-bank financial institutions. The chart on the right shows repo rates in the US relative to interest on reserve balances; the median is in the dark line, and the 75th percentile in tan. As liquidity declined in recent years and issuance rose, the spread that dealers require to intermediate between those segments increased. Some of this represents reserve balances required to settle repo, but it also is related to balance sheet costs.
So, in looking at this, I think what I take away is that the Fed's operating framework is facing a system that has very volatile and very diffuse repo rates; and what you can see is that the SRF rate, in the blue there, isn't clipping the tops of the trees of these spreads. And what that means is that the SRF, because it's uncleared, isn't reaching those segments of the market, and I take two things away from that; one is that the SRF should be cleared in the United States, and the other is that repo markets themselves could be set up more efficiently.
Success on implementation frameworks is determined by money market rates. Whether central banks choose demand- or supply-driven floors, it will be that they will assess whether or not it's working. We haven't seen demand-driven floors work in practice, as Imène said, but central banks have been impressive in their comprehensive efforts—and I think it's encouraging that euro area banks are using facilities already, and that repo is actively traded across banks.
In the US, the ample reserve system has achieved excellent rate control for the vast majority of time; but as you can see on the right, in 2019, it did not. This reflected reserves being allowed to fall too low to absorb a normally-sized payment shock; frankly, there were also repo market issues, but I think that was yesterday's panel. Another lesson from that day, though, is that there was no flexible liquidity readily available to short circuit what should have been a solvable problem.
Standing liquidity supply isn't sufficient to protect against all events; as you can see on the right-hand side, six months later the pandemic represented an enormous global liquidity demand shock. Repo rates—repo operations—surged $300 billion over the course of two weeks, and that was enough to help offset that shock.
In fact, the chance of occasional shortfall is inherent. The Fed was operating with too high of a chance in 2019—but at the same time, greater standing liquidity supply would not have protected against March 2020. Looked at in this way, backstops are part of the monetary policy framework, and the flexibility principle that the ECB is emphasizing in its demand-driven floor is part of the US supply-driven framework as well.
The FOMC can choose whether it wants to maintain more reserves to protect against a broader range of events on the outer edge of the gray line, or one that includes a chance of occasional backstop usage. Either way, it's important for market participants to have solid confidence in the resilience of the US operating system.
In summary, I see the US as benefiting from more standing reserve supply than other countries; but there is unfinished business, and talented Fed in New York, at the Board, and around the Fed system are working on it. I see several components to that.
First, liquidity tools should be consistently positioned as part of the regular monetary policy framework. This is business as usual activity, and facilities are open for business. Communications and regulations should be actively aligned on this principle, to reduce stigma. It may even be helpful to draw brighter lines between discount window liquidity and the LOLR function.
Second, enhancing operations increases effectiveness. I applaud morning operations for the SRF, but the operation also needs to be cleared. This reduces balance sheet costs for dealers, and increases intermediation out to the non-bank sector—which, we saw in those charts, isn't happening right now. Work is ongoing for the discount window, including a lot of work around automation and collateral streamlining.
But finally, I would just say that the Fed should be unabashed about promoting efficient and resilient money market practices, too—like central clearing and appropriate risk management. Thank you.
Beckworth: Thank you, Patricia. We encourage you to submit your questions. We have some already. I have a few of my own, but I'd love to get to your questions as time permits.
I want to start with Claudio. I want to start with a general question, because this conversation gets technical pretty quickly. But just to paint the big picture, Claudio, since you're at the BIS, you follow this; you listed a number of countries—in fact, I heard you say that most countries are still on some form of a corridor system, if you look at all the emerging markets, and it was the advanced economies that went to more of the ample reserve floor system.
So, my question is just, again, the bigger picture here: Why are some of those advanced economy central banks making a move now? We spent a decade with the floor system, ample reserve—over a decade. What do you think is the impetus, the drive? Imène talked about some of the reasons at the ECB, but can you paint a broad picture for why there's interest in demand-driven systems at this time?
Borio: Yes; can you hear me?
Beckworth: Yes.
Borio: So, let me tell you: the first point that I would like to make is that while often—there was an argument some time ago that basically said, well, one can wait to discuss issues of operational frameworks because the balance sheets of the central banks are so large, it will take ages to absorb all of those reserves, and so on and so forth. But I think it's important to understand that operating frameworks have to do with the liability side of the balance sheet, not necessarily with the asset side; because for any given asset size, it's the composition of the liabilities—for example, the instruments with which you absorb reserves—that really matter for the system.
And this explains why a number of emerging market economies that have very, very large balance sheets still operate—largely because of foreign exchange reserves—still operate a scarce reserve system, or what you call the corridor system. In fact, the US has a corridor, except the deposit facility rate is the upper part of the corridor these days.
So, that's the first point to make. The second point is that, people have brought these two aspects together in a way that they need not be; but given that they have associated the operational framework with also the size of the balance sheet, well, effectively what happens is that many central banks that are moving now towards reducing the size of the balance sheet—for very good reasons, that may have nothing to do with the amount of reserves in the system—basically think that it is useful to start rethinking about the operational framework.
And so, the idea that we see nowadays is largely about reducing the size of the balance sheet, and in the process, basically seeing how you can adjust your frameworks in such a way as to still be effective with a smaller balance sheet. And that's basically why you have the shift that you have among those countries that basically have done a lot of QE in the past. Because remember, the others are not thinking of moving to such a system, even if they have very large balance sheets.
And the other aspect, of course, has to do more with financial stability considerations, and the idea that maybe it's good for banks to have a large amount of reserves, as opposed to relying on the central bank, in order to provide more self-insurance (and I discussed how far that takes you, and how far that doesn't take you, in the discussion of the two systems). But basically, I would say the reason why many central banks are thinking about it nowadays is because they are eagerly reducing the size of the balance sheets—and, as a byproduct, reserves are falling.
And there is no decoupling, in their mind, in terms of the size of the balance sheet, and the specific composition of their liabilities—so, the two issues are effectively commingled. Now, indeed, what you see here is the risk that at some point you have the size of your liabilities actually dictating the size of what your assets should be; and I think that that's the wrong way of thinking about it. One should basically decide on the optimal size and composition of the balance sheet, and then in that context also think of what operating framework would make more sense; otherwise, there is the risk of the tail wagging the dog, which is one reason why we have seen very little changes in operating frameworks so far.
The key point is that, to me, a very good principle for the central bank balance sheet is that the central bank balance sheet should not be large because of the cost that it brings, but it should be very elastic—lean and elastic, and ready to increase when the time requires.
Beckworth: All right; lean and elastic. Imène, what would you say? You can speak to the ECB, and you've already touched on it, but maybe repeat again: What are the important reasons why you're going toward a soft floor, demand-driven system?
Rahmouni-Rousseau: I think, interestingly, many of the reasons that Claudio gave, and especially this thinking of providing reserves elastically, is really at the heart of that. So, coming back to what we learned, I think, from the financial crisis is that leaving the system short of reserves carries costs in terms of financial stability; and so, I think that the sweet spot you want to find in your operating system is that you supply sufficient reserves—not too much, because then, indeed, you have costs associated with too large of a central bank balance sheet—but enough reserves to avoid that you reach that point where banks are short of reserves (and also, money market volatility becomes higher).
And in our case, this demand-driven system—where, basically, banks express their demand for reserves, and we satisfy it in full but we don't satiate it (so, we don't give them more than they need)—meets a number of goals. So, Claudio was mentioning that interbank activity and redistribution of reserves in markets is something you want to have. I think that this is not necessarily associated always with scarce reserves. I do think that we can have a hybrid system like the one we run, which is probably somewhere between ample and scarce, where you can preserve this possibility for market activity.
And of course, this means to carefully calibrate this opportunity cost of holding reserves that Claudio was mentioning—and one of the innovations of the framework that I mentioned already is this; it's basically this wedge that we introduced, relative to the Bank of England, that has the pure version of a floor, which is basically you provide reserves at the same rate as the deposit rate (so, no opportunity cost at all). In our case there are these 15 basis points, and we think that this allows for this market activity, and we also think that this is a signal that's saying we don't want necessarily rates to be completely flat at the floor, but we would allow some deviation.
Beckworth: Okay; Patricia, do you want to add anything to that?
Zobel: I would only say that, just building on what Imène said, that I think pre-financial crisis what we found is that imposing costs on banks' demand for reserve created a lot of payment system risk, and I think that moving to ample systems addresses that. I think in the US context, there's another factor that means that we have to supply somewhat more reserves with asset purchases, and that's really the large swings in reserve balances coming from the TGA—which means that there does have to be a larger amount of standing reserves in the system. I'm not sure that other central banks have that same degree of fluidity that the US does in the reserve supply.
Beckworth: So, you and Imène had a point that overlapped—that in the ECB, the ECB works through the banks, and in the US it's not so much. However, you did bring up TAF, which is one way to maybe get at something similar, a similar system; but you argue that it probably wouldn't be enough—that even if you had TAF, maybe even if you got more regular use of the discount window, it probably wouldn't be enough to get us all the way to a demand-driven system.
There were conversations yesterday about adding central clearing, as you also suggested, to the Standing Repo Facility; can you foresee any possible developments that would make it feasible? What would need to be done, if Fed officials at some point in the future said, "well, let's take a look"—what are the big hurdles that you see before we could even have this conversation?
Zobel: Well, I think one thing that is, as you mentioned, different in the United States context from the euro area and most other countries, is that it does have a very large non-bank financial institution sector participating in deep, liquid capital markets. And this is driving a lot of activity in money markets, so I think we could move to a system that has better distribution of reserves by implementing things like the TAF operation; and oftentimes it is small banks that, when liquidity declines, do end up being a little bit shorter of balances, and so that could help reduce the amount of standing supply that we might need to provide.
But it seems hard for me, given the capital markets activity, given the fluctuations in reserve supply, to move to a fully demand-driven system. I do think that the US context benefits from standing reserve supply.
Beckworth: Okay. You mentioned, Patricia, in your speech the Friedman rule, so maybe you could speak to that. What does that offer? Why should we consider it? And I think the point you made is that it occurs in an ample reserve system—and anybody else who wants to weigh in, Imène, Claudio, but let's start with you, Patricia, and then the rest of you can speak to: How important is considering the Friedman rule on an operating system versus all the other checklist items you might want to think about before moving to a certain operating system?
Zobel: So, I think that the essential concept is something that Imène and I both mentioned, which is that in scarce systems, pre-crisis, there was a large cost imposed on holding reserve balances. And in the US, that meant that banks economized to a fairly extraordinary degree on reserve holdings. In fact, it wasn't only the reserve requirement and the cost imposed; they used things like sweep accounts to try and avoid actually having any reserve requirements.
So, they were operating with a very thin level of reserves, and what we saw during the Global Financial Crisis was that the interbank exposures, the intraday credit—those things added to payment system risk. And I think Claudio's point is well taken, that the central bank liquidity could be quite flexible; but even so, I would think that you would want banks to have more balances than that.
So, in terms of what the Friedman rule is saying, is that a safe, liquid financial system is a public good, and you shouldn't charge a premium for liquidity because the central bank can provide it at a low cost. I think there are systems that are corridor-like systems; I know those have been proposed by folks recently, that use a voluntary reserve target system that can supply a lot more reserves, reduce the cost on it—but still operate in a corridor system, and have interbank trading.
Again, I go back to the US context. I think in terms of the TGA and the large swings in non-reserve liabilities, and a little bit more uncertain demand for reserves; I think banks would have high balance sheet costs of demanding a certain amount of reserves, so you'd have to impose a large cost to get that to work.
Beckworth: All right; Claudio or Imène, do you want to speak to it?
Borio: I'd just like to say that the Friedman rule is okay, as far as it goes; but honestly, it doesn't really go very far—because if you want to understand which system is superior, then you effectively have to answer all of the questions that we have been raising here, in terms of the pros and cons of all of the systems. To say that providing liquidity is free; well, no, it isn't free, because we know, in order to provide it, you have to have a certain balance sheet, and so on. There are issues about the operation of the markets, and so on.
So, I would say: the Friedman rule, in the specific context in which he used it, makes sense, but it doesn't take you very far. You really have to answer all of the various questions that were raised during this panel in order to understand whether you want to have that system or not. Plus, there are a few more that I didn't discuss, but . . .
Beckworth: Yes; so, there's a number of criteria you want to go through, not just the Friedman rule, when you're determining your balance sheet. All right; in the time we have left, let's go to the audience questions. I have a question for Imène here, from Bill Nelson, and he says: "In the new ECB framework, if money market rates are near the deposit rate and the lending rate is 15 basis points above the deposit rate, why would any bank borrow from the ECB when it could fund itself in the market for less?"
Rahmouni-Rousseau: Yes, that's a very fair question; and I think you have to look at this in, basically, a variety of ways. So, the operations that we run have a set of parameters, and the set of parameters is determined not only by the pricing, but it is also determined by the type of collateral that we take in these operations—and also by the fact that these operations are full allotment, where banks can have certainty to get reserves when they need it. And I think that if you take that into account—so, not only the pricing, but all the rest—you see that there can be motivations for banks to access our operations that go beyond the strict need to borrow reserves, and I think would then make these operations attractive over time as reserves go down.
So, just to give an example: The fact that we accept broad collateral means that non-HQLA—what we call "non-HQLA collateral," which is this collateral that doesn't have a value in liquidity regulation in the LCR—can actually be pledged in our operation, provided that it has sufficient credit quality and all the things you want, and this will convert it into reserves. So, banks have some incentives to come to these operations because of the breadth of collateral.
And I also believe that, even though there will be a cost between the market rates and our operations, this cost is not so large. So, some banks may also decide to basically take a little buffer in our operations in order to make sure that in all circumstances they have the certainty that they can basically meet their needs. And remember, reserves are the ultimate safe and liquid asset; and we learned that the hard way, with SVB and these kind of runs that can happen at any time.
So, I think certainty that you can get reserves at the central bank is also a very important point.
Beckworth: Okay; well, I think it's safe to say that this is a journey for all the central banks taking on the demand-driven approach, so there's a lot of learning still to do. It'll be interesting to follow this.
Another question from the audience; Michael McKee said: "The Fed has been encouraging use of liquidity sources, particularly the discount window, for decades, and never seems to be successful. How does that change?"
Well, Imène, maybe I'll start with you before I go to Patricia. You mentioned you want business as usual at the main refinancing operations—which is similar to the Standing Repo Facility, sort of similar to the discount window; it's the ceiling. How do you make it so successful? What can we learn from you?
Rahmouni-Rousseau: Now, that's a very good question. So, I'll start with the fact that our banks have the history, and the habit, of accessing our repo operations. They did it pre-2008; we had these longer-term repo during the pandemic, and large parts of the banking system came to these operations. The second thing is, I really think that terminology matters; and in our case, we say that the main refinancing operation is not a backstop. It's not a backstop, it's an operation—and it's not just a facility.
So, I think this matters, and this is part of communication; and what is also part of communication is what supervisors would say about the operation. So, the perception that banks, if they go to the operation, they have no scrutiny, it's "no questions asked"—and this is not only supervisors, but also rating agencies, for example. So, we're really trying to communicate this "open for business"—this is what you said, and I like the term—of the operation.
And finally, maybe just to mention something because I picked it up, I think things like disclosure around these operations matter. So, in our case, we disclose the aggregate amount but we never disclose the name of the bank, we don't disclose the geographical distribution across countries—and I think that helps banks to have comfort to come.
Beckworth: Patricia?
Zobel: Well, the ECB is a great example of a central bank that has been able to operate its ceiling operations without stigma; and I think there's a lot of great work going on in the United States right now, and so I'm really actually quite encouraged by the amount of intellectual energy that's being brought to this issue.
First, I think—like I mentioned in the presentation—it's really clear that we see the facilities in the United States as part of the monetary policy framework. It's not lender of last resort; it's not an emergency if it's used. If you look at the distribution of potential for shortfalls in the system, it's an unlikely event, or maybe it's an occasional event; but we can't look at it as LOLR, and I still see in some of the supervisory, some of the communications, where people say that the discount window is used in an event of stress, or it's framed still as something that looks like lender of last resort.
So, it has to be open for business; it has to be communicated like that. I think the supervisory and regulatory communications have to be clearly positioned, and I think you could be bold in the way that you do this. Maybe this is an opportunity to separate out the true "lending to insolvent institutions" from the "monetary policy operation," and create a brighter line to help reduce stigma.
Beckworth: Okay; another comment, from Ian Harnett: "Patricia mentioned the growth of US NBFI's role in the US system (a nonbank financial institution's role in the US system); should we focus on their liquidity needs rather than the needs of the banks, which now control less than 25 percent of US financial assets?" So, maybe we got it all wrong.
Zobel: Well, central bank supply reserves—and they're held by the banking system—in other countries there's a broader group of institutions that can hold central bank balances. I think the Bank of England is an example where they broadened that out. I think in the US, the system has been effective, but I think over time we're going to be thinking a little bit more about that as you see things like stablecoins and others getting access to the central bank's balance sheet. I think it will be a good opportunity to say: Where does liquidity risk exist, and where does the central bank's role stop and the markets begin?
Beckworth: Alright; I have one more question for Claudio. We're running short on time; I don't think I'll get to all the audience questions. But Claudio, another option for an operating system we haven't talked about, that some central banks do, is tiered reserve systems. So, it's option four, maybe. What do you think about that? Is that another step forward, or would that be a step backward for you?
Borio: Well, it depends which perspective you have. From my perspective—and it depends on the initial conditions where you come from. If you're moving from an abundant reserve system and you move to a tiered reserve system, it helps; it's good, because it reduces the need for the amount of reserves out there, and it also—provided it's done properly—it encourages more interbank activity.
Now, if you are there, the question is, "well, why don't you go the full hog?"—and you go back to the old system. But again, that depends very much on your perspective, and your pros and cons; but definitely, it's a step forward compared, I would say, with respect to a standard floor system.
Beckworth: Alright. Patricia, the last question to you. So, we're in the midst but soon to wrap up the Fed's framework review focusing on strategy and communications tools. Should we have a similar operational framework review like the ECB did for it? Would the Federal Reserve benefit from a process like that?
Zobel: Well, I'm a big fan of periodic reviews to rethink your operating framework, your policies, procedures; and so, I think that taking a look at the operating framework every several years to see if it is working as intended, to see if there's improvements that can be made, is actually an excellent idea.
I think that, somewhat different than the principles that central banks use, operating frameworks are put in place for long periods of time. There's a big cost to the financial system, there's a big cost to the central bank, of shifting it wholesale; so, I think those types of reviews should be of a little bit longer tenor, but I do think there's always opportunity to take a look at what you're doing, to take feedback from the public, and to assess if there are any improvements to be made.
Beckworth: Right. Imène, you have the final word; any parting wisdom to give to us here in America?
Rahmouni-Rousseau: Honestly, I'm not in a position to give advice here. I would just say—and really, reaffirm . . . because Claudio said it, and I think I also see it—that every operating system is basically also customized to the type of financial system that you run, and also to the institutional framework that you have as well. So, this was a great panel to understand, what are the moving parts there? But as you said, it's really a journey there.
Beckworth: It's a journey. All right; let's give the panelists a hand.
All right, so here are your parting instructions. We want to thank you for joining this session on day two; we're going to take a break at this point, both online and in person. For those in person, please return at 6 p.m. for a reception here, followed at 7 by our evening keynote session featuring Mary C. Daly, president and chief executive officer of Federal Reserve Bank of San Francisco, Beth Hammack, president and chief executive officer of Federal Reserve Bank of Cleveland, moderated by Raphael Bostic, president and chief executive officer of Federal Reserve Bank of Atlanta.