2026 Financial Markets Conference – Morning Keynote Transcript - May 19, 2026

May 19, 2026

Morning Keynote – Reserve Management and the Fed's SOMA: Recent Experience and Insights from Surveys

Transcript

Roberto Perli: Good morning. Thank you, Brian for that nice introduction. Brian was obviously my predecessor, two or three times removed—something like that. Also, my—not my favorite—you were my second section chief when I joined the board, just a few years ago.

All right. So... well, thank you. It's a pleasure to be here. I certainly thank you, interim President Venable, for the invitation. Thank you to the organizers. It's a really great conference, really flawlessly organized, so thank you for that.

So today, I just want to speak about three things: money market conditions, and how we decide how many reserve management purchases to do every month. We'll make some comments on bank demand for reserves, based on results from the recent Senior Financial Officer Survey (SFOS), which has some interesting tidbits in that. And third, I'll conclude with some comments about standing repo operations, which are very important from our perspective on implementation of monetary policy—again, using some tidbits from the SFOS as a starting point.

And some of these considerations will connect with the ongoing debate about the size of the SOMA (System Open Market Account) portfolio, the size of the Federal Reserve balance sheet, that we all heard and read about in recent weeks and recent months. Before I start, though, two things; the usual disclaimers: the views that I express here are my own, and not necessarily those of the Federal Reserve Bank of New York or the Federal Reserve System. Second, I want to thank the people that really helped me with these remarks, so—Natalie Leonard, Lindsey Molloy, and Eric LeSueur helped me write the speech, and Navya Sharma, Richard Wilson, and Zach Youngblood helped with the slides that you will see here in a second.

So, let's start with RMPs (risk management programs). So, as you know, in December, we started doing RMPs at the—of course, the FOMC told us to do that, because we foresaw the fact that in April (tax season) there will be a big drop in reserves (you can see on the right, in those slides over there), and we wanted to "fill that hole" well in advance. We didn't want to do it at the last minute, because it would be operationally challenging; so, we started doing it in December.

And, as you can see from that chart, around tax season the TGA (Treasury General Account) grew a lot—north of $1 trillion—and reserves dropped correspondingly, by more than $300 billion over the space of just less than two weeks. So, big drop, and fast. And yet, money market conditions were just fine. Reserves remained above, as you can see there.

I want you to look at the dark line, which is what actually happened; the blue line is what would have happened if we hadn't implemented the reserve management purchases. So, reserves remained above the troughs that we saw in late last year, when the FOMC said we reached ample (so that's good). And also, from a point of view of money market conditions, if you look at the chart on the left here, panel two, repo rates increased on the tax date—that's the last trip line, the last bar—but then they quickly reversed the next day.

And also, as repo rates increased marginally above the SRP (suggested retail price) rate on April 15, some of our counterparties participated in SRP operations—which is exactly what I expected to happen, given that SRP became economically sensible to use. I'll return to the SRP later, but basically all this suggests that the strategy that the FOMC put in place in December worked as intended—worked very well, I would say.

Now since the tax date, we saw a decline in the TGA, a corresponding increase in reserves, and a softening in money market conditions (you see there from panel two, again). The effective federal funds rate dropped one basis point (so far, at least), and the repo reference rate printed notably below (see it there), below IORB (interest on reserve balances).

And there are a lot of factors behind that; certainly, negative net bill issuance is one. The TGA declined again—a lot—to a position where it's not going to stay there; it's going to come back up. But in the meantime, that contributed to the softness.

As a result of all this, and consistent with prior communications, we reduced the monthly pace of RMPs substantially—and somewhat gradually, like we said: first from $40 billion to $25 billion for the mid-April to mid-May purchase period, and then down to $10 billion for the mid-May to mid-June period. As you know, we make this determination on a monthly basis, mid-month to mid-month.

So, looking ahead, we'll continue to manage the SOMA portfolio to keep an ample supply of reserves, but I want to emphasize that RMPs are not on a predetermined course; we'll make RMP decisions on a month-to-month basis to preserve the flexibility that is needed to adapt to changing market conditions. So, what's this mysterious process that we use to determine how much to purchase each month?

Well, very briefly, it's nothing too mysterious. I want to be as transparent as possible. We look at three things, all equally important: One is our forecast for reserve supply that I showed you earlier, and it's influenced by various things, but reserve supply in the future is important, particularly affected by the TGA and other liabilities.

Number two is our assessment of reserve demand—and this is informed by our outreach to banks and other counterparties, as well as surveys like the SFOS. And third is money market conditions. Money market conditions serve as a key input in our decision process but also serve as a key feedback mechanism.

So, as cited in a previous speech, we monitor a lot of indicators of various kinds of money market conditions, and so going forward, we'll continue to do that. We'll take all this information from these three broad types of categories; we'll look carefully at it, and we're ready to adjust the pace of RMPs up or down as necessary to maintain reserves within the ample range.

And by the way, I should have said at the beginning—you probably realize I'm not reading the whole speech; thank me for that, because it's very long. I'll just give you the main points.

To RMPs; so, obviously, RMPs are one of the factors that determine the size of the Federal Reserve portfolio, and currently the SOMA portfolio stands at around $6.4 trillion (this chart shows you the pace of RMPs, how it evolved over time). Now, the SOMA portfolio stands at about $6.4 trillion, which is roughly 20 percent of GDP (Gross Domestic Product), and as a result of the balance sheet reduction that the committee put in place between 2022 and 2025, late last year, that number is well below the post-pandemic peak of $8.5 trillion (or, 33+ percent of GDP).

So, assets declined. Now, if you look at the panel on the left there, assets are now growing in dollar terms. Why? Because of RMPs. But in principle, and everything else equal, as a share of GDP, the balance sheet—if we continue doing what we're doing—should stay roughly the same. So basically, you can take that chart on the right and project it flat going forward, everything else equal.

However, in real life, as we know, everything else is rarely equal. And so, let me spend a couple of minutes discussing some possible factors that could change the size of the SOMA portfolio—not in dollar terms only, but also in terms of as a percentage of GDP—and in doing so, this is where the SFOS comes in handy. So, I think some results there were interesting.

So, in simple terms, the size of the SOMA portfolio is a function of the demand for Federal Reserve liabilities, the FOMC choice of a monetary policy implementation framework, and the FOMC tolerance for money market volatility. The primary liabilities that drive the Federal Reserve balance sheet size are currency, as you see them actually here, on this table; the primary liabilities are currency, Federal Reserve notes, Treasury General Account (or TGA), and reserve balances. And together, these three components account for over 90 percent of all reserves.

So, let's go through this—not one by one, I want to leave currency aside because let's assume it just keeps growing at a constant rate. Let's talk about TGA and reserves. So, the size of the TGA is managed by Treasury, according to its cash management policy. And recently, interestingly, the Treasury asked the Treasury Borrowing Advisory Committee (the TBAC) for feedback on whether the Treasury should invest a portion of the TGA into overnight repo markets.

So, at high level, if the Treasury goes ahead and decides to do that, well, the TGA would decline by a corresponding amount, and consequently, the Federal Reserve assets needed to back the TGA would also decline by a similar amount—or somewhat less, depending on implementation details. But that would be one way to reduce the size of the SOMA portfolio.

As for reserves, obviously we implement monetary policy through an ample reserves framework. Maybe I should have said this earlier. What do I mean by ample reserve framework? Well, "ample" refers to the range of reserves that makes the federal funds rate only modestly sensitive to short-term variation in reserve supply.

So, this is the demand curve for reserves; we want to operate essentially where the slope of that curve is kind of a gentle slope (to the right of it we have abundant reserves, to the left we have scarce reserves). And this framework has been in practice since the Global Financial Crisis, and was affirmed officially by the FOMC in January 2019.

How have things gone since we started using this framework? Well, very well from the point of view of rate control. Rate control has been very strong—in particular, the effective federal funds rate has been outside of the target range only on two days since 2010. One was the year end 2015, and the other one was the famous—or infamous—September 17, 2019.

But other than that, very, very strong rate control—and importantly, the ample reserve framework also supports the resiliency of funding liquidity within the repo market, which in turn supports a well-functioning Treasury market and broader financial stability. I think this was one of the points that were discussed yesterday in the first panel.

Now, current implementation of this framework is demonstrably very effective, as I just said; but there's also an active public debate about the quantity of reserve supply that it entails, with a number of papers and speeches by policymakers on the subject—again, something that was discussed in the first panel yesterday. And conceptually, if policymakers wanted to reduce the supply of reserves and therefore the size of the SOMA portfolio, the size of the balance sheet, they have two broad approaches at their disposal.

The first one is to take steps to reduce reserve demands (so, essentially engineer, in some way, a shift of the demand curve for reserves to the left; that's panel nine there). Or the alternative approach, the second approach, is to move up along the existing demand curve, and that's panel ten.

So, let me say just a couple of words about these two alternatives. Current ample reserve system implementation framework is well equipped to handle a reduction in the SOMA portfolio, through a leftward shift of the reserve demand curve, and at a high level.

If reserve demand diminishes and reserve supply remains the same (or increases because of RMPs), there will be an imbalance between reserve demand and reserve supply, like there was when reserves were abundant. And our indicators would pick up this imbalance, and at that point—depending on the extent of the change—we at the Desk would be in a good position to adjust downward the pace of RMPs, stop RMPs (potentially), or if the increase... so if the shift in demand is large enough, even be in a good position to recommend to the Committee maybe a resumption of balance sheet runoff.

But in this hypothetical situation, reserves would be lower, yes; the portfolio would be lower, the balance sheet would be lower, but rate control will stay strong and money market conditions would remain stable because, still, reserve supply would meet reserve demand. So, in other words, we'll have a shift down in the scale of our operations, but nothing changes. So, it's lower reserve demand met by lower reserve supply; everything else stays the same.

Now, let me open a quick aside here. As I said, our current ample reserve framework would handle this pretty seamlessly, but of course I don't want to give the impression that this is the only possible efficient implementation of an ample reserve system.

In fact, as you know, there are other central banks that operate alternative versions with still strong rate control. Bank of England, for example, operates a system where the marginal quantity of reserves demanded is supplied via temporary open market operations rather than via securities purchases; and this alternative system, by itself it probably wouldn't have a big effect (or any effect) on the size of the SOMA portfolio, but it would affect the composition of the SOMA portfolio.

And so, if policymakers desire a composition that is more tilted towards repos than towards securities—well, this system could work in the US as well. Luckily, I dropped a useless page; good. So, my point here is just, we're not saying that what we're doing here is the only "right" thing in the world. There are other central banks doing things in a different way but they achieve very similar good results, and those systems could work if those are in the US as well.

Now, what could bring about a change in reserve demand? Well, many things; but a plausible one, given the current debate, is changes to liquidity regulations. And according to the SFOS, changes in liquidity regulations, as you see here, are the most important catalyst for a change in reserve demand. So, we asked them, "What do you think could change reserve demand among various options?" Well, liquidity regulations came up at the top. This is a scale from one to five—one being little, five being a lot.

Banks also mentioned other factors, including shift in liquidity management, transition towards 24/7 payments, the adoption of payment innovations, and so on. But all these catalysts that you see there tend to push reserve demand in different directions, but survey respondents told us that, on net, the expectations for reserve demand are biased to the downside, even over relatively short horizons (such as one or two years). So, reserve demand may not drop very quickly, because changes in liquidity regulations have not happened yet; a lot of the details are still unknown.

There is also the bank reaction time to factor in—so maybe not super-fast, but our respondents expect that if these changes happen, reserve demand will come down on net. And this will be a leftward shift of the demand curve, and it can be handled seamlessly by an ample reserve framework.

Now, what about moving along the reserve demand curve? Well, FOMC could simply decide to just cut the reserve supply and let's move up the demand curve—and this would lead to an increase in money market rates above IORB, and this will be in all likelihood accompanied by higher volatility, which would introduce risks. We have to be realistic; it would introduce risks in terms of rate control, funding market stability, and consequently, Treasury market stability, which is another of the things that was discussed yesterday.

And here, too, I think the SFOS is interesting; it gives us some ideas of what we could expect. So, these charts show you the answers that respondents gave to a question, a hypothetical question: Suppose interest rates increase by X, Y, and Z basis points; what would you do with your reserves?

And we have some confidence here; we can have some confidence on this data, because what the banks did between September (which was the last time this question was asked) and March, was in the same ballpark of what they actually did. So, these are not just made-up numbers; there is some relevance to them.

So, what do you see from these curves? Well, what I see is that they are very steep, in particular the one on the right for domestic banks; and this implies that banks would have to see large increases in overnight rates to be willing to shed modest amounts of reserves—or, flip it around: a modest decline in reserves could induce a substantial increase in overnight rates. And this poses risk, because we have learned from experience that increases in money market rates to levels even modestly above IORB can become disruptive very quickly, with all the consequences we have seen for not just rate control, but also the stability of repo markets and Treasury market, and the financial system in general.

So, let me finish very quickly with some consideration about standing repo operations. As you know, these are critical for our monetary policy implementation. They're intended for the sole purpose of supporting monetary policy implementation, and they are expected to be used when economically sensible. However, past market outreach pointed to frictions that were discouraging the use of SRP by some counterparties in making the operation not as effective as it could possibly be.

So, over the past year, the Desk and the FOMC have taken important steps to improve the effectiveness of SRP, and as a result of that we have seen encouraging signs that willingness to use the operations has increased. And for example, this year they weren't used very frequently because money market rates didn't justify their use; but when they were used, they were used when money market rates were just very marginally—just a basis point or two—above the SRP rate, which is a very good thing. It means that when they're used, they work as intended.

Now, it's encouraging; recent experience is also backed up by the result of the SFOS again. And what I show you here—look at the left side of the chart, these are hurdle rates for banks. Hurdle rates are just basically, ask how high above the SRP rates they need money market rates to be in order for banks to be willing to actively consider participation in those SRP operations. So, the brown lines are the medians, and the median hurdle spread came down for banks; the blue bars are interquartile range, and those came down and shrunk.

So, good news; lower hurdle rates on the part of banks to use to participate in the operation, and the dealers (on the right)—we don't have a comparison because we didn't ask the question to dealers in the previous survey—but dealers are at even lower hurdle rates. So that's very good news.

And in addition, banks and dealers rated various factors affecting their views on SRP participation, and the most positive factors (and those that showed the most positive change since a year ago) were recent increases in SRP take up, Federal Reserve communication and proposition limits—so, all things that we have promoted, essentially. So, this is very good.

To be perfectly honest, there are still things that hold us back a little bit; in particular, look at the last two rows there. Disclosures—which, we can do nothing about—and lack of balance sheet netting, which we could do something about (for example, via central clearing).

The SRP operation, of course, if you follow the minutes of the November/October FOMC meeting, that's a delicate subject; it's full of tricky questions, but the Desk continues to explore this and other possibilities to improve the efficacy of our facility. And an efficient SRP operation is especially important, in the case of a move along the current reserve demand curve.

I'll stop here, and I look forward to the discussion with Brian.

Brian Sack: All right; thanks very much for the comments. There's a lot there. I want to jump in, and I want to cover a lot, so I'm going to just go right into it.

So, it seems to me like the ample reserves regime is working very well. It seems highly successful to me. We've had a lot of volatility in markets, a lot of volatility in economic developments; funding markets have been resilient, rate control has been good. I assume you agree with all that.

And the foundation for this regime is really the decision in December to do the RMPs, which involved a directive that gave you a lot of discretion on setting the pace of RMPs to be consistent with ample reserves. I really appreciate that you talked a bit about how you make that decision month to month, but I wanted to pin down a little bit.

So, in the past you've talked about a variety of conditions that look like they're consistent with an ample reserves regime—but paramount in those conditions is really where market interest rates end up relative to IORB. So, first of all: do you agree with that? And is there some guidance you can give in terms of the degree of tolerance for that deviation? What would lead you to adjust RMPs in response to those market rates?

Perli: Yes. So, first of all, I agree that certainly the current system has been extremely effective, from the point of view of rate control, as I said earlier. So, RMPs—as I said, we look at a variety of information to come up with our monthly pace. So yes, certainly; it stands to reason that the federal funds rate, repo rates—in the neighborhood of IORB is one of those conditions. In fact, we've seen it up until the last couple of weeks; they dipped lower for a number of reasons, but the number of resources doesn't necessarily have to do with the quantity of reserves.

So, there's Treasury issuance, for example; bill issuance in particular has been negative (-$300 billion, I believe, between about March and now), so that injects more liquidity in the system. That is a story about money market funds assets—inflows into the money market funds—so, a greater supply of repo lending.

Well, there are a lot of other factors. We take all this into account, but our job is to not just be focused on what's happening now, so we try to look ahead as well. And if we look ahead, like we did back in December when we were looking at April—if we look ahead now, what do we see?

I just mentioned Treasury bill issuance; it's been negative for the past couple of months. Well, it's going to become significantly positive starting in July (or thereabouts—end of June, early July), and then it stays positive again in August. So, we don't want to get caught in too active management of reserves here, but we try to look ahead and try to strike what we think is the right pace to try to accommodate changes that we see coming.

Sack: You showed those charts from the Senior Financial Officer Survey suggesting that the reserve demand schedule is actually relatively steep, and I wondered—looking back at the experience since late 2025 to now—if we also should reach that conclusion, or what lessons we should draw just on how markets behaved.

So basically, you managed to end QT without a significant funding market disruption, which I think is a tribute to the information you provided to the FOMC. But if we look back and think about lessons, in Q4 we did see a lot of upper pressure on repo; we did see a lot of volatility (that, of course, is what led to the RMPs). And the level of reserves at the time was like $2.85-2.9 trillion. Today, the level of reserves is $3.1 trillion, and we see a lot of softness in market rates.

So, even though we avoided a real disruption, is there a lesson here that this was yet another episode showing that that reserve demand schedule ends up being pretty kinked, and can steepen pretty quickly? And when reserves get below that comfortable level—

Perli: Yes; if you look at it that way, you could say that what we have seen, both in the fall and what we're seeing now in the opposite direction, but kind of supportive of the idea that the demand curve is fairly steep. Which actually is one of the pieces of information that we take into account, of course.

We have one thing that we didn't have, or my predecessors didn't have, in 2019, which is the standing repo operations—which, I think we gained confidence from what we have seen, and from the results of the survey as well. We gained confidence that they would be effective in dampening some of the pressures. And we saw it also in the fall; they were used, relatively frequently, and to some pretty decent extent—I think year end we hit north of $70 billion; that's a decent amount.

So, we have that; that helps. But yes, I think the information that we have at our disposal at the moment, including the survey that I just showed you, is that the schedules are probably a little bit steep. So, we need to be careful about that because, as I said, it is certainly about rate control, but it's also about the stability of the repo market, the stability of the Treasury market, and the whole system.

Sack: So, I was going to get to the standing repo facility—the standing repo operations—in a bit; you're not allowed to say "facility."

Perli: I'm ready to correct you.

Sack: I was going to ask you if we're on a path to central clearing. It sounds to me like you're not ready to answer that question, so I'll ask a broader question. Will we see ongoing operational changes, whether it's central clearing or other changes, that can make it even more effective? Or are we at a resting point, where it looks—I mean, I agree; it looks like it's working better. The activity indicates that, the surveys indicate that.

So, are we at a resting point, or will it continue to evolve?

Perli: So, let me say just two words about central clearing. There can be no doubt that, from a pure implementation perspective, yes; it makes sense. And we hear it all the time from our counterparties, so everybody tells us "do it, do it," because netting is important; it frees up balance sheet space.

It's not the only thing that matters. Think about it: to start centrally clearing the SRP, we will have to become a member of a clearing house. And that by itself is a big step. Would we change the competitive landscape there? As you know, the FICC is the major one, but CME and ICE have also expressed interest of entering, or have already been approved to enter.

So, would we change the competitive landscape by doing that? Would we create some moral hazard issues? What models? So, there are a lot of thorny issues that I think need to be addressed, and thought about carefully. That's where we are, if you look at the minutes of the October FOMC meeting; there is some discussion along those lines.

Are there other things that we could possibly do? Sure. We're always willing to listen to our counterparties. In fact, we talk, as you know, to dealers and everybody else very frequently; we're open to change the timing. So, we already introduced a second operation. We already introduced an early settlement option. Is 8:30 in the morning the right time? Shift it earlier, could shift it later—could add another operation later in the day, maybe for banks, especially.

So, we're certainly willing to listen to a lot of ideas, but at the moment it seems like things are working fairly well.

Sack: I'm going to keep going, but I want to remind everyone they can submit questions in the app. Okay; I wanted to talk about Fed balance sheet reduction, as I said in the intro. There are a lot of operational issues involved with that; this broad framing of the issue of how to reduce the balance sheet, whether you're shifting demand in but staying on the relatively flat part (so, kind of maintaining the benefits of the ample reserves regime), versus whether you're moving up the curve—I find that a very useful framing. I know it's one others have used. I was actually going to ask you if you agreed with that as a broad framing, so I guess the answer is yes—so maybe I'll ask a more specific question, then.

So, one thing that seems different on these two paths is coming back to where market rates are relative to IORB. I think you can't move up the demand schedule towards scarcity, unless you're inducing positive spreads between overnight market rates and IORB. That's the mechanism through which you get up the curve. So, is that fair? Is that a good thing to gauge to determine which of these two paths we're on?

Perli: Well, to move up the curve, you need to—so, the demand curve is downward sloping, so if you want to move up you need to move to the left on the horizontal axis, and so, supply fewer reserves to move rates in the... we're already on average, except the last couple of weeks, in the neighborhood of IORB. If you want to shrink the portfolio more, you could move further to the left; it's perfectly (theoretically) feasible.

It's just up to policymakers, obviously, to decide what they want to do. Our job is to inform, and point out that there is a little bit more risk in this option with respect to others. But we're here to implement monetary policy.

Sack: Yes; I totally agree. I think we sometimes talk about these regimes like they're discreet, like we're either going to be in scarce or we're going to be in ample or abundant, and effectively moving up the curve introduces the spread, presumably begins to introduce some volatility and some vulnerabilities, and it's just a matter of how far if they want to do something.

Perli: Yes. So, you introduce some vulnerabilities. To be fair—I want to be fair—the SRP gives us some confidence that the currency could handle, because of the SRP, some movement to the left; I don't want to completely exclude that, but that comes with some higher volatility. Then it's up to the FOMC to decide how much volatility they're willing to tolerate. So that's how I would frame it, from my seat.

Sack: Okay; I wanted to get into this discussion of TGA, and the possibility of Treasury using the repo market themselves. I think this is pretty fascinating. So, as you noted, the last Treasury Borrowing Advisory Committee meeting had a charge on this—not at all focused on it being a vehicle for shrinking the Fed's balance sheet, but it being a vehicle for Treasury to earn interest on some of its TGA balances. And there were lots of different options discussed, but many of them involved Treasury coming into the repo market when repo rates were firm, specifically when they were above IORB.

And that's pretty interesting to think about, because if Treasury implemented that approach, it actually has some similarities to monetary policy implementation. You'd have a situation where Treasury would be sitting there willing to enter the market when rates are above IORB, at least up to some volume, and then the Fed is sitting there ready to come into the market when rates go above IORB+10, a little bit farther back from Treasury.

So, you'd have two government entities potentially coming into the market. Is that strange? Is that operationally problematic for the Fed?

Perli: I don't know. If rates go up, you will see a number of other market participants participating more in the rate boom; right? So, leave Treasury aside, and money funds will do the same; banks could do the same. In fact, banks tell us if rates go up, we will participate in repo; we have a higher threshold, but we will participate.

So, I wouldn't say that banks are doing monetary policy, or money funds are doing monetary policy. So, I think when rates go up people participate more in the market; it's kind of a natural thing. I view this as an additional source of funding in repo.

So, also from our perspective, the last thing I want to do is tell Treasury what to do or not to do. So, we take that as a given, but from our perspective I don't think that would pose challenges to the implementation of monetary policy.

Sack: Okay. And just for clarification, I think there are two different approaches; one is, Treasury is interested in earning interest, so they come in when rates are above IORB. Another (which was not discussed in TBAC) would be, Treasury could just take a chunk of its TGA—like, $200 billion—and put it in a steady rolling RP book or something, and that would actually, in the end, as I think you said, lead to a smaller SOMA balance sheet. So that would be—

Perli: Yes. So, in that situation, if the Treasury participated regularly—so, if there was like... let's say the TGA is $900 billion, the Treasury put a couple hundred billion to the repo market: yes, that, to a first approximation... so if they did that consistently, every single day, to a first approximation, assets needed to back the TGA will come down by a couple hundred billion, and the balance sheet will be smaller by a couple hundred billion. It remains to be seen whether the Treasury wants to do that, so that's why I said in the speech it depends on the implementation details.

Sack: One topic that came up yesterday was, should the Fed just hold less duration risk? So, maybe focusing on the overall size of the balance sheet isn't the right metric; maybe focusing on the amount of duration risk in the balance sheet is the right metric for thinking about the size of the balance sheet.

I wanted to ask an operational question related to that. So right now, the Fed in SOMA has maturing coupon securities, it rolls them over into new securities across the curve, based on whatever Treasury is issuing. So, Treasury has its announced auction sizes to the public, and then the Fed add-ons just get put on top of it. One option to reduce the duration of the SOMA would be to shift those add-ons into shorter maturities, which I know is something that's been discussed.

And that wouldn't seem to have any consequences for markets; their announced public auction sizes could be unchanged, and the Fed just has add-ons that go into shorter securities—say, like two years, three years, and five years. Is that right? That could be done in a seamless way, with little—or maybe no—market consequence?

Perli: I guess there is an issue of a steady state versus an issue of transition; presumably the transition will have to be gradual, in order not to create too much uncertainty for the Treasury. But there is also a question of, what do policymakers want in terms of the composition of the securities portion of the portfolio? And if you look back at the minutes of the—I think it was October again; September, October—and then another meeting about a year ago, there was a lot of support for having a composition of securities that's roughly consistent with the composition of outstanding Treasuries.

If you look at that, the interesting thing is that the current portfolio is actually pretty much in line with the composition of outstanding, with two exceptions. One is bills, and we're taking care of that (with RMPs), and the other one is the 10–20 portion result of the old QE (quantitative easing) operations.

And so, as I said, we are rebalancing the bill portion. If we continue with RMP, at some point it will reach the 22 percent (roughly) share of bills, like the share of bills in the outstandings. At that point, the committee will have to decide: Do we want to continue, or do we want to stop here? That's a discussion for the committee.

I think it's harder to bring back the 10–20 bucket; obviously, you could do it by, let's say not reinvesting in that bucket. So, you know how rollovers work; they're distributed across auctions. You could, say, exclude the 10–20 option; if you do that, though, because the other buckets are in line—forgive me the irreverent comparison—but we would be playing literally a game of Whack-a-Mole, right? So, we have this bucket that's above the others, so we hammer it down, and then the three to five pops up. And so, you get out of line in some other buckets.

So, anything can be done. It's a question of what policymakers will want us to do, and I think a full discussion didn't take place yet. So, a preliminary discussion date a couple of times last year, but we will do whatever policymakers tell us to do.

Sack: Okay. Well, I think we're at time, so I will leave it there. I was asked to say that we'll have lunch out there right now, and then if everyone could be back in the room by 1 p.m., that'd be appreciated. Thank you, Roberto.

Perli: Thank you, Brian.