Bill Nelson on the Using the Discount Window for Liquidity Requirements and Its Implications for the Fed’s Balance Sheet

Removing the stigmatization associated with use of the discount window is one way for the Fed to improve its role as lender of last resort.

Bill Nelson is the chief economist and an executive vice president at the Bank Policy Institute. Bill previously was a deputy director of the Division of Monetary Affairs at the Federal Reserve Board, where his responsibilities included monetary policy analysis, discount window policy analysis, and financial institution supervision. Bill also worked closely with the BIS working groups in the design of liquidity regulations. As a returning guest to Macro Musings, he rejoins David to talk about the recent proposals to improve the Fed’s lender of last resort role via the discount window, as well as recent developments related to the Fed’s balance sheet.

Read the full episode transcript:

Note: While transcripts are lightly edited, they are not rigorously proofed for accuracy. If you notice an error, please reach out to [email protected].

David Beckworth: Bill, welcome back to the show.

Bill Nelson: Thanks, David. It's great to be here.

Beckworth: It's always fun to have you on, because you know the plumbing of the Federal Reserve, the monetary system in the US, and there's always something exciting happening related to those topics. And you've been doing some work, some writing, and some articles we're going to share. But I want to begin with some developments that have been coming to the surface, have been making a splash— at least in your space— and I follow it from the sides, but that is this new push for the Federal Reserve to have a more robust discount window or a more robust user-friendly lender of last resort role because of the turmoil we saw last year in the banking system. We have a G30 report that came out, the Comptroller of the Currency. In one of your notes, you also mentioned that the Federal Reserve and other regulators, bank regulators, have issued new supervisory guidelines, so maybe talk about that. What is this push for, and how would it be different than what we currently have at the discount window?

Improving the Discount Window

Nelson: [I’d be] happy to, and this is an issue that I've worked on for about a decade and a half, and I continue to think it's really important. About two weeks before SVB failed, the Bank Policy Institute issued a call for just a holistic review of liquidity requirements, top to bottom. What we pointed out was that, before the global financial crisis, when a supervisor assessed the liquidity of an institution, it looked for stable, well-diversified funding, including contingency funding, and part of that contingency funding was being prepared to borrow from the discount window.

Nelson: After the global financial crisis, that turned entirely into, pile-up big stocks of high-quality liquid assets, assets that, it turns out, banks aren't even really willing to use under stress. Then, two weeks later, when SVB failed, it was an institution that was awash in high-quality liquid assets but had non-diversified, very concentrated, unreliable funding, and wasn't prepared to borrow from the discount window. Two days later, Signature Bank failed, and it was also a bank that was not well-prepared to borrow from the discount window. So, what happened in the wake of those failures was two things. One, of course, was a greater recognition that institutions need to have collateral at the window and need to be ready to borrow to meet immediate cash flow needs, because the discount window is available immediately, and it's available up until the end of… well after the end of the day. It's the only, really, source of funding that offers that, except for deposits at the Fed.

Nelson: The other recognition was that, well, everybody reassessed upward their views on how rapidly deposits can leave an institution. After all, three-quarters of SVB had left in 36 hours, basically, right? Yet, if the regulatory solution to that is just to make banks hold more HQLA— basically, entirely loans to the government in the form of Treasury securities or deposits at the Fed— you're just going to make banks look more and more like money market mutual funds or narrow banks and take them away from providing credit to the economy. But if you allow them to do what the Federal Reserve was designed to do back in its founding and allow those banks to make loans to businesses and households, and pledge those loans to the discount window as collateral to establish lendable value as your response, then you're making the system safer and more liquid. But at the same time, you're allowing banks to continue to lend to businesses and households.

Beckworth: Yes, and It's interesting that this discussion is happening now, because, as you noted, you have been working on this for a long time, this idea of having banks park collateral at the discount window. So, it's this contingent liquidity that they could draw upon if the need arose, and it could be counted toward their liquidity requirements, right? It would be something that bank supervisors would look at and say, okay, you guys are prepared for some future crisis. And this is something you've been working on for a long time, back when you were at the Federal Reserve, even. This was something that piqued your interest and you were thinking about. Mervyn King had a similar proposal, a similar idea, pawnbroker-

Nelson: Pawnbroker for all seasons.

Beckworth: -for all seasons. So, maybe trace that history of ideas, the history of thought behind this. I mean, I guess it is tied, also, to the fact that the discount window has had this stigma attached to it as well.

The History Behind Discount Window Collateral

Nelson: Right, so that goes way back. I'll turn to the stigma secondly, but the stigma has been associated with the window since the 1920s. The Fed has also— I would hasten to add, because there's a lot of misunderstanding about that—the collateral that's pledged to the window has been pre-positioned collateral for decades, at least back into the '80s, and I don't know how far back. That's because the Federal Reserve wants to accept the least liquid collateral possible, so that it can be enhancing the liquidity of the financial system by making loans, not just taking liquid securities and replacing them with cash.

Beckworth: Right.

Nelson: So, it's always sought to take all bankable assets, including, in particular, loans, as their collateral. And 80% of the collateral that's at the window is in the form of loans. About 40% is consumer loans, 20% [is] business loans, and then there's other types of loans that are in there. But what the real nugget of the question was, [is] when you were determining how liquid a bank is, should you or should you not be including the bank's capacity to borrow from the discount window? And in the wake of the global financial crisis, the perspective was, no, you really shouldn't, because that's the purpose of liquidity regulations, is to keep institutions from borrowing from the discount window.

Nelson: That was sort of a strange view, because the ECB and other central banks, their main asset was lending to banks. That's a very traditional, normal thing, for central banks to lend to banks, and discount window lending is mostly unremarkable. The vast majority of it is unremarkable. That's why I tend to not use the term “lender of last resort,” just because it conjures up all of the aspects of it that sort of reinforce stigma, which I'll turn to. Anyway, so, as one of the people that was working on the design of the liquidity frameworks, this was an area of specific interest of mine, and I advocated, since probably 2012 or further, that this be knit into the international liquidity requirement framework in a specific form, where central banks sell banks, for a fee, guaranteed lines of credit, and then those guaranteed lines of credit can count towards their liquidity requirements.

Nelson: The current area of concern and nature of the suggestions are a little bit different, because they're more about severe, immediate cash outflow needs, but it's all quite related. But, as you say, once you start recognizing… suppose, as Acting Comptroller Hsu recommended, that you have a requirement that banks have collateral at the window or deposits at the Fed equal to five-day cash outflows, or the G30 report, which is very similar, that they have collateral at the Fed or deposits equal to uninsured deposits and other short-term wholesale funding. Once this basic idea, that immediate needs can be met with the discount window, is accepted – and it seems to be accepted – you can actually make other changes that allow banks to rely less on reserves at the Fed. And a big implication of that, and a topic that we've discussed before, is that that should allow the Fed, actually, to end up getting much smaller, which I know we'll talk about. 

Beckworth: So, what you're saying is that the proposals out there wouldn't dramatically change the collateral currently being held at the Fed, it's just how it would be treated in terms of supervision and reporting for liquidity purposes. Is that right?

Nelson: I think that is right. I think that there would have to be more collateral pledged. The last estimate that the Fed provided was $1.6 trillion, but that was quite a while back. I imagine that it's quite a bit more than that now, but I think that more would still need to be pledged. To make that happen, one of the things that would probably be desirable would be to fix up the somewhat complicated relationship between the Fed and the Federal Home Loan Banks, in terms of the pledging of collateral.

Nelson: So, the Federal Home Loan Banks benefit from a blanket lien, meaning that if you take out a loan from the Federal Home Loan Bank, all of your assets are encumbered. The Fed can't take them as collateral. What happens is that each Federal Home Loan Bank and each overlapping Federal Reserve Bank, for each individual Federal Home Loan Bank customer, work out a collateral sharing agreement that basically says, for example, that the Federal Home Loan Bank subrogates its interest to the Fed on all consumer loans. Then, that bank will pledge the consumer loans to the Fed. But that takes time, and that was the kind of thing that clearly wasn't going to happen over SVB weekend or Signature Bank weekend. But even just getting that prepositioned collateral available, that would be helpful.

Nelson: There's a widespread view, I think, out there, unfortunately, that there are significant improvements that can be done in how the Fed accepts collateral. I'm not sure that I completely buy that. Maybe the way that the Fed processes securities can be improved, but that's really only 20% of the collateral that they receive. I've been talking to a lot of regional banks and money center banks about these issues, and what I'm hearing is that banks are generally satisfied with the way the Federal Reserve processes its collateral.

Beckworth: Okay, so, the collateral is mostly there, maybe a little bit more collateral. The big change is going to be more from the way that the regulators, the supervisors, interpret that collateral, those assets, that the banks hold.

Nelson: That's right.

Beckworth: Then, also, for the banks themselves, the banks need to be open to tapping that liquidity and not feeling stigmatized, so it's a nice segue back to stigma.

Nelson: Right, well done.

Beckworth: And one thing that was interesting in reading and preparing for the show, is that the difference at the discount window between a primary and secondary credit… so, if you're a bank that has lower credit standing, you go to the secondary credit window. Is that right?

The History of Discount Window Stigma

Nelson: That's right, but that's very rare.

Beckworth: Okay, but it does create extra stigma and reluctance to even go to the Fed in the first place if you're in that category, but you're saying it doesn't happen very often.

Nelson: It's extremely rare. The stigma that's of concern is the one that's associated with-

Beckworth: With the primary [credit lending], okay.

Nelson: -the regular discount window lending, which is primary credit lending.

Beckworth: Let's talk about that then.

Nelson: Sure.

Beckworth: Walk us through the history of that. I know that we've done it before on the show, but for the sake of being thorough here, how did we get to the place where there's so much stigma with the discount window?

Nelson: This is something that goes back to the 1920s. One part of the issue is that from its beginning, the discount window— discounting loans, providing credit— has a dual purpose. One of its purposes is a monetary policy tool. It's intended to make the level of reserves flexible to the demand. At the creation of the Fed, there was a big… this was largely to meet seasonal swings that had previously been associated with rising risks of financial crisis, and they just wanted to have a more flexible system of reserve creation. But, at the same time, it's also a contingency source of funding. Whenever you tap a contingency source of funding, there's an implication that something went wrong.

Nelson: Former Governor Betsy Duke— who had been a CEO of a community bank before becoming a governor— she described borrowing from the discount window as like borrowing money from your parents. You'll do it if you have to, but nobody likes doing it. So, the Fed, in the '20s and then again in the '50s, actually took steps to increase stigma. As they described it, they were taking advantage of the inherent tradition against borrowing, that a safe bank just takes deposits and makes loans. They don't go out and borrow additional money. And, in doing so, the Fed sought to amplify those views so that banks would borrow from the Fed less.

Nelson: In the '20s, this was because it was observed that a lot of banks that borrowed then failed. In the '50s, it's because, interestingly, the Fed was transitioning out of an episode that looks a lot like the situation that the Fed is in now. So, during World War II, they bought vast quantities of securities. That meant that there were vast quantities of reserves, which meant that nobody needed to borrow from the discount window. As they ran down those securities and reserves shrank, people started to borrow more and more, and the Fed was uncomfortable with that situation. And the way that they addressed it in the mid-1950s was to just make the rules about borrowing very punitive and very much expressing an expectation that the banks had done something wrong and needed to stop doing it.

Nelson: That then changed in 1972. It was the first effort to fix this by giving banks a more clear borrowing privilege. Then, again, in 2003— and this was the revision that I was involved with, where we moved to an above-market rate, and we got rid of all of the rules about borrowing. The idea was that banks should be able to count on it and use it when they want to, and the rate alone should be what encourages banks to borrow, only as a contingency. And that seemed to be working. I was told by a banker with a memory back into that period recently that that was gradually working. We were going out and telling banks that we wanted them to use the window. We were educating examiners. But then, that all went out the window with the global financial crisis, in the wake of which banks that had used the discount window or borrowed from the term auction facility— which is another discount window facility— even though they were encouraged to do it repeatedly, were treated like they'd received bailouts, and they were pilloried by the public. That vastly increased banks' unwillingness to borrow.

Nelson: And banks tell me, even if they're told that they should borrow, if they do borrow, their examiner will come back in after the fact and say, “well, there must be something wrong with the way you're doing things, because you had to borrow from the discount window.” It goes all the way up to the leadership of these organizations. I've been told repeatedly that… well, one story that I was told by a friend who's a treasurer at a large institution, when he took his job, he was told that if he borrowed from the discount window, there'd be two phone calls, one from the president of the New York Fed to his bank's CEO to ask why the bank borrowed, and one from human resources to him, telling him to clear out his desk. So, under those circumstances, it's very difficult to get banks to be willing to use the window. And so it's a very important tool, but it's currently hobbled by these problems.

Beckworth: So, maybe we can view this development, this push to make the discount window normal, less stigmatized, more robust, as a return to the pre-2008 trend. So, in 2003, you said that up, really, to the crisis, there was a re-engagement with the discount window. Maybe we're getting back on track finally after a decade-plus of really keeping it at a distance. Question, so let's say we all get on board, we support this decision, there's a collective agreement, and there seems to be growing consensus for it. Who actually sets the decision? Who's the main regulator? What has to happen so that everyone makes a coordinated effort toward it? So, we all agree… is it the Federal Reserve? Is it the Comptroller of the Currency? How do we get this concerted effort started and happening?

Implementing Changes to the Discount Window

Nelson: These changes would be done through notice and comment rulemaking, which means that the banking agencies— the Federal Reserve, the FDIC, and the OCC— would collectively come out with a proposal to make changes to how— for example, if there were some new requirement, they would need to go out, propose that new requirement, and receive feedback. I'd argue that this is an appropriate situation for what's called an “advance notice of proposed rulemaking.” I may not have that term exactly right, but, basically, what it is, is that rather than proposing a rule and getting comments on it, the agencies go out and say, “We've got this idea, and our idea is that things should broadly look like this, but we're seeking input before we put something down as a specific proposal.”

Nelson: And given the potential far-reaching nature of this, and the complexity of it, and the potential implications for the banking system, and the need to reconsider various aspects of the liquidity framework regime, that seems more appropriate to me. But, in any case, it should be done through a rulemaking process, rather than just, sort of, examiners stepping in and encouraging banks to do so. I'm sure that examiners are encouraging banks to be more prepared from the discount window, and that's appropriate, but if it's a systematic expectation or rule, that needs to be done with notice and comment.

Beckworth: So, if we go down this path, we get other regulators on board, [and] it happens, what would it mean for other liquidity regulations? In particular, I'm thinking of the liquidity coverage ratio. That's kind of set up and framed around this world of large reserve balances.

Nelson: Yes.

Beckworth: And this could change that. So, would that become a less binding constraint, or maybe even relegated to the dustbin of history? Or, would it work together with a rule like this, or an emphasis like this, on the discount window?

Nelson: Since the LCR is an international standard, it's hard to change. I think what's important, and where a lot of the sentiment for all of this is coming from, is that it not be tightened and simply have that tightening. So, the liquidity coverage ratio, it's a requirement that banks hold liquid assets in an amount equal to this formula that's meant to replicate the cash outflows that the bank would receive based on its assets and liabilities over 30 days of systemic and individual stress.

Nelson: Since we observe more rapid outflows on uninsured deposits than we expected— now, specifically in this case, it was a very concentrated group of individuals that all communicated on various different channels that move very rapidly, but, nevertheless, it was still stunning. You could see that, maybe, some of the dials in the LCR would be turned up, but that would just mean that banks have to hold more HQLA. So, one thing that I think, in the immediate term, that will happen, is that the LCR will largely be left where it is. I'll come back to the fact that it probably, ultimately, should be revisited, but there are changes that could be made now that would be very helpful, and that would be appropriate in light of this changed expectation. So, let me give you two examples.

Nelson: In the international standard for the LCR, banks are required to have the HQLA, high-quality liquid assets, that they need to meet their liquidity needs at the end of 30 days. Whereas, in the United States implementation, they need to be able to meet their peak need over 30 days. So, they need to be able to make it to the 30 days. And so, that's gold plating, as they say, that the Fed could remain compliant with the international standard, but not have this, what's called maturity mismatch add-on. But, there’s a certain thing appropriate to that add-on. So, one way that these two things could be brought together is to say, well, if you have the discount window collateral sufficient to meet your intra-30-day needs, we don't need this add-on anymore. And I think that would be a very reasonable change to be made. That would add, further, to the incentive for banks to hold collateral.

Nelson: Another change— and this is completely internal— the LCR gets a lot of attention, because people are familiar with it, but it's actually not the most important liquidity standard that banks are exposed to. There's two other things. Banks have to do internal liquidity stress tests and report the results to their supervisors. And those are at an overnight 30-day, 90-day, and one-year horizon. And as part of those requirements, they need to not only have liquid assets, but they need to be able to show that they can monetize those liquid assets, that they can convert them into cash as needed.

Nelson: Well, at least up until recently— and I think that this might be changing— banks weren't able to point to the discount window or even the Standing Repo Facility, which is the new facility that, really, is just the discount window by another name that the Fed has created, but, created to make banks treat it as just sort of a business decision, kind of no stigma thing. Anyway, banks aren't allowed to point to those things as the way that they would monetize their assets. And this is actually a reason why SVB had vast quantities of Treasuries and agency MBS, very highly liquid securities, but they weren't able to turn them into cash, because they wouldn't have gotten any supervisory or regulatory credit for setting themselves up to use the discount window, for setting themselves up for using the Standing Repo Facility. So, we could be looking at a very different world right now if those incentives had existed.

Nelson: And these are perfectly reasonable things. You really can use them to monetize your assets. The Fed really intends for people to be willing to do this, and you really are more liquid, as a bank, if you're capable of doing this. So, just adjusting these requirements to make them more reasonable will create the right incentives. And if banks are being required to have discount window collateral to meet five-day needs, [then] it would seem absurd to say, “Oh, but you can't plan on using that capacity in your contingency planning for liquidity stress.”

Beckworth: Which is currently the state of affairs, right?

Nelson: [It’s] currently the state of affairs.

Beckworth: So, these changes— we've been talking about the discount window, but they could also potentially involve the Standing Repo Facility. So, you could, in your liquidity contingency plan, say, “Hey, these many dollars in the discount window and these many dollars over there at the Standing Repo Facility.”

Nelson: Conceivably. Now, the reason why the Standing Repo Facility is strictly inferior to the discount window [is], for one thing, [because] there's no prepositioned collateral. You bring your collateral there, and you engage in a repo. But, also, you can only do it once per day, and you can only do it in the middle of the day. I forget the time that they have, but it's just one auction. Whereas, the discount window, which takes the same collateral, applies the same haircuts, in addition to other collateral— but it does take Treasuries and agency securities— provides you dollars up until 6:59, or after the close of Fedwire, at any time, [with] the same amount of funding. So, the reason why the Standing Repo Facility— it was created, in part, because of concerns about repo markets and providing credit to primary dealers. But, the other role— the role as a stigma-free discount window for banks— that's entirely about packaging.

Beckworth: Okay, one last question and then we'll move on to the Fed's balance sheet. But, you just mentioned something that reminded me of some of the history or the things going on in March of last year. You mentioned that the discount window closes at 6:59. Did I hear you right?

Nelson: So, I'm not 100% sure about that.

Beckworth: Okay, but it closes at some point during the day. And, if I remember correctly, during the March turmoil, they couldn't get the discount window. Some of the banks had a hard time reaching… or maybe it was Fedwire.

Nelson: I think it's a Fedwire problem.

Beckworth: It was Fedwire. But, it struck many of us like, well, that's bizarre that they're in the midst of a banking stress at these mid-sized banks, and they can't reach or make payments. It just really struck me that you would have 24/7 access in this modern age, but there are still limits on the hours and the times of the day.

Nelson: I think, first and foremost, what would be extremely helpful, for a lot of reasons, would be if the Federal Reserve were to keep Fedwire open 24/7. Fedwire is the backbone of the US payment system, and It's basically pretty simple. If you're a bank, you have your deposit at the Fed. I'm a bank, I have my deposit at the Fed. If you want to make a payment to me, then you tell the Fed, “reduce my account, and raise Bill's account.” That's how Fedwire works. It works similarly with what are called book-entry securities— so Treasuries and agency MBS— where the Fed also operates that system, which allows you to move those securities from you to me and then the cash to go in the other direction.

Nelson: And I think what caused the problem for SVB was that they were struggling, if I remember correctly, to get $40 billion from their correspondent in collateral, and that they were also struggling to get an additional $40 billion from the Federal Home Loan Banks. But what was reported was that, well, the wire closed at 4:00, and so they weren't able to get those securities from their correspondent. I think a lot of people read that, and said 4:00, without really thinking, well, that's 7:00 PM EST. It was striking that they didn't keep it open later, and they could have kept it open later, but 7:00 PM is normal for-- everybody else closes then. It's not like the Fed kicks off at 4:00 and everybody goes-- 

Beckworth: Right, but even 7:00 PM seems a little strange for a national market that spans multiple time zones. Again, in this modern day and age, it seems like it would be fairly straightforward to have this process running. Maybe you have a few people in the back office, or maybe it's automated. Maybe I'm just dreaming. 

Nelson: Well, I really agree. I mean, even if you looked at the banks' real-time payment system, RTP, or the later Fed real-time payment system, FedNow, those would all work much more easily if Fedwire, this wholesale payment system, was open 24/7. A lot of the things that made those complicated to implement had to do with, well, how do you deal with these periods when banks can't actually transfer funds to each other? It's just that it would really be a very valuable change for the Fed to make.

Beckworth: Alright. Chair Powell, if you're listening, let's get it started. Okay, circling back, though, to these changes to the discount window and lender of last resort rule, which, you don't like that framing— fair enough— I have to ask, would it dramatically reduce the size of the Fed's balance sheet? Because both you and myself are advocates of returning the Fed's balance sheet to a much smaller size, and would this make a meaningful dent? Would it really help on that front?

How Would Discount Window Changes Impact the Fed’s Balance Sheet?

Nelson: It absolutely could. As you know, when the Fed first looked at— so, just to go all the way back, the way the Fed conducts monetary policy, the way they used to conduct monetary policy, is that they would adjust their assets so that the level of reserve balances— which is their liability, which is money they borrow from banks— but, so those deposits of banks were just sufficient for banks to meet their clearing needs and satisfy their reserve requirements, and they provided that number. Whenever they would announce a new rate, they'd move the discount rate, interest rates would just move to that new requirement. It was all quite simple.

Nelson: The Fed needed to understand, what were going to be the shocks to reserves, because of tax payments, or we even used to have to worry about, well, if there was bad weather in Chicago, that meant that the planes that moved the checks around— but it was all relatively well understood and well known, and the implementation was simple, and the Fed was small. So, the level of reserve balances back then was about $10 billion. Starting during the global financial crisis, the Fed got bigger and bigger and bigger, first through lending emergency facilities, but then through waves of QE, and because the assets were getting bigger, their reserve balances were getting bigger, and the Fed switched to conducting policy differently.

Nelson: What they started to do was that they oversupplied reserve balances. They provided these vast excess amounts of reserve balances, and that then pushed interest rates down to the interest rate that the Fed was paying on those deposits, what's now called the interest on reserve balance rate, the IORB rate. Well, the more the Fed provides, what happens is the more that the banks adjust those added reserve balances, the more bank supervisors adjust to expecting banks to satisfy their liquidity requirements with all of those reserve balances. And this has been documented by working papers, by a paper written by the New York Fed, including President John Williams, that shows that this demand moves out as the supply moves out, but also through a variety of other academic papers. This is, I think, broadly accepted now, that there's this ratchet in the provision of reserve balances. That means that as you get bigger, it's very difficult to get smaller.

Nelson: So, right now, there's about $3.5 trillion of reserve balances out there. In April 2008, about six months before Lehman, when the Fed entered this new regime, the Fed asked the staff to do a big study to look at how they could conduct policy. The staff considered this excessive reserve framework, and they said that that would be a bad idea. That would be too radical a change. But, if you were to do it, you would need to provide excess reserves of $35 billion. Right now, the amount of reserve balances is $3.5 trillion, a hundred times higher than that amount. Now, it could probably come down some from there, but it could very well not get much lower than $3.25 trillion before they start meeting some resistance.

Nelson: Now, part of the reason is because banks satisfy their liquidity needs with these holdings of reserve balances, and they're expected by supervisors to do so. But the rules themselves say that you can hold Treasuries, you can hold other things to meet those reserve requirements. So, in a world where everybody became more comfortable with just counting on the ability of the institution to raise money from the discount window, maybe hold much smaller deposits at the Fed, but also to hold Treasury securities that they would convert into cash at the discount window or at the Standing Repo Facility, then, you're in a world where the amount of reserve balances that banks have to hold can get much smaller. It could be a very important component of a series of changes that could end up making the Federal Reserve much smaller.

Beckworth: Yes, I'm going to cite some numbers from the papers that you have written— one of them we'll mention in a minute— but you note that April 2008, $35 billion, I believe, is what it was back then. March 2016, the Fed— this is the Fed's estimates— $100 billion. March 2018, $600 billion. December 2018, $800 billion. September 2019, $1.3 trillion. May 2022, $2.3 trillion. So, what is it now? You say, maybe, $3 trillion?

Nelson: It seems like it could be about $3 trillion.

Beckworth: Yes, so, the point is that it keeps ratcheting up, and the way you eloquently put it in a previous paper is that the Fed does QE or large-scale asset purchases, and now these banks come, they begin to sit on them, and it feels normal, but then you've got to have a little bit of a buffer above the normal, and so this ratchet effect just continues to build. Every time there's QE, you've got to get a little more of a buffer above what was injected via QE, and then it gets really hard to move back, but it is possible with these potential changes to policy, so I'm hopeful. After reading all of this and thinking this through, I'm like, well, maybe this is our one route back. I was hopeful that the Standing Repo Facility would be one route back, and that hasn't done much, so, possibly, we get there. I mean, just to recap where we are in terms of the Fed's balance sheet, it peaked around $8.9 trillion, and we're now down to $7.6 trillion. We've come down a little bit, but how much farther can we go? We don't know.

Beckworth: The Fed's Treasury holdings are about $5.8 trillion, now [down] to $4.7 trillion. Reserves are on $3.5 trillion, as you said, and they were as high as $4 trillion. Overnight reverse repo is down under $1 trillion, and it was about $2.5 trillion [before]. So, there is a reduction, but everyone's a little, I think, worried that we're going to trip over that point, which throws us back into a scarce reserve system, and what's crazy, scarce reserve at, say, $3 trillion, right? $2.8 trillion may be… oh no, we fell back into a corridor system, which is just so ironic given the enormity of it. Let's use that as a segue into some recent documents that were dumped by the Fed, or released by the Fed, maybe, better put, and that is the FOMC transcripts from 2018. Now, those were exciting for people like you and me, because that's when the discussion of this framework took place, at least the official discussion. They had been doing it since December 2008, effectively, right?

Nelson: Yes.

Beckworth: But this was, okay, let's make it official, let's talk about it at least. And just off the bat, before I turn it back over to you, it's striking to me that they only spent, I think, three meetings talking about this; November and December, 2018, and [then] January. And, I don't know, that seems short-changed, not enough discussion for something as significant as your operating system.

The Past, Present, and Future of the Fed’s Operating System

Nelson: Right. Well, to be fair, they'd also discussed it in 2016, and this was, as you noted, how they'd been operating for many years. And, honestly, I think that the views within the Fed was that this was already pretty much--

Beckworth: It was written on… okay.

Nelson: It was pretty much baked in the cake that the conclusion that they were going to reach was that we should just keep doing it this way. After five years, each January, the Fed releases the transcripts from all of their FOMC meetings [from] five years prior, as well as almost all of the memos that the staff wrote for the FOMC and provided to them. For those of us who are very interested in how these things work, it's always sort of a gold mine of material when it's released. And 2018 was especially interesting, because, as you noted, in November and December, they had these big discussions of, how are we going to conduct policy? There was no suspense. We knew that, in January, they officially adopted this floor system approach, this over excessive reserve provision approach.

Beckworth: Yes.

Nelson: You may note that I avoid saying scarce reserves versus ample and abundant reserves, and that's for three reasons. One is that it's sort of value-laden. Scarce sounds bad. Ample and abundant sounds good.

Beckworth: Yes.

Nelson: The second reason is that it leaves you with the wrong impression. The prior system didn't require carefully calibrating the level of reserves so that you land right in the right spot of that steep point of the demand curve. It was, basically, just interest rates moved where they were supposed to, and the Fed kept providing the reserves that they needed. The third reason is that I can't keep ample and abundant separate. One is bigger than the other, and it just mystifies me. So, I tend to say excessive reserves and necessary reserves.

Nelson: However, okay, 2018 transcripts. There were several things that were really striking about this, to me. When you read them, it's not that surprising that the committee reached the conclusion that they did. For one thing, the staff basically said that the difference in the quantity of reserves that you're going to need to provide under the two regimes isn't that great. If you want to do it the old way, because the demand has gone up for liquidity requirements and these other reasons, you still have to provide $700 billion in reserves. Whereas, if you wanted to do it in this floor system manner, well, [then] you'd need $800 billion to get to the floor, and then you'd need another $200 billion buffer so that things can move around without having to do fine tuning operations.

Nelson: It's really only a difference between $700 billion and $1 trillion, which isn't that big of a difference when you're thinking about it. Now, if you compare $700 billion to $3 trillion, you're starting to talk about real numbers, right? But, strikingly, Chairman Powell, when discussing these numbers, said, “Well, if this turns out to be $1.3 trillion or $1.5 trillion, rather than the $1 trillion that you're telling us, I think that I'll have buyer's remorse in making this decision to do it that way.” And when you read the minutes from the November meeting, it recognizes that many folks, many of the participants, argued that, “Well, if it ends up being a lot more than we think, we should revisit this decision.” So, perhaps consistent with your sense that this hasn't gotten enough attention— by their own reckoning— they really need to revisit this decision.

Nelson: The other thing was that the staff just really overplayed how simple it would all be. You read the memos, and it's like, “Wow, we'll just oversupply reserve balances, and then all we'll need to do is to buy securities to match the growth in currency, and that will be the end of monetary policy implementation.” Well, just a few months later, in September 2019, we learned how wrong that was. The Fed was surprised by this big supply-demand mismatch in repo markets, at a time when they were shrinking, and they needed to suddenly scramble and do huge repo operations, and now it needs the overnight RRP facility to reach such vast quantities, that half of the assets of the money fund industry— of the government money fund industry— were loans to the Fed. There's this massive footprint of the Fed in the financial system in order to implement this regime. I would suspect that there's probably a fair amount of buyer's remorse going on, and I do hope that, as we enter into this period of the Fed's framework review, that they include, as part of that framework review, the consideration of how they actually move interest rates to where they want them to be.

Beckworth: Well, I hope you're right, but my understanding is that it's strictly the framework, like FAIT and inflation targeting. But it is a good point, because if you look around the world, many other advanced economy central banks, they are doing actual reviews of the operating system. The ECB, right now, is looking at, what is the best way to do this? In fact, I just read an article this morning that was critical of the ECB, because it was considering doing something like a tiered system where some of the reserves don't get remuneration, there's 0% interest on it, and this article was critical. But, the point was that the ECB is actually thinking, [and] it’s doing a formal review of its operating framework.

Beckworth: Other places like the UK, they actually have standings arrangements with their finance ministry, that if they experience losses, they have to get explicit capitalization from the government. The Swiss National Bank is taking big hits. So, it seems like everywhere else, this is something that's more on the minds of policymakers, politicians. Here, we've kind of lost interest. Understandably, there's a big election coming up, there's other things, but it would seem to be appropriate, given everything that the Fed's been through with the balance sheet, to step back and say, okay, is this the optimal approach? Just like they're going to do with their inflation target. And so, I'm hopeful that they will, at some point, do that. I suspect they probably won't do it at the end of the— do you know differently? Are they going to—?

Nelson: No, but the only thing that gives me a little bit of hope is that— and maybe this is wishful thinking on my part— but I do sense a slight change in the tone of the communication. Previously, the plan that they articulated was that, well, we don't know the minimum necessary quantity of reserves, but we're going to shrink until we're about $300 billion above that unknown quantity and then stop. That's, of course, not a very good formula for figuring out how small you can get. Also, as we were just discussing, that's exactly the formula that builds in ever-growing demand for reserves.

Beckworth: Right.

Nelson: Well, there seems to be increasing mention of [how] we're going to shrink until we start seeing borrowing pick up, at the discount window or at the Standing Repo Facility. And that's exactly how the Bank of England is doing it. They're shrinking until they see borrowing pick up. As many people noted in the last minutes— the minutes of the December meeting, I guess it was— there was an indication that several participants, or some participants, started to mention that we should maybe be starting to worry about QT going down. Of course, it wasn't expressed as the view of the group or the core group.

Nelson: Right after those minutes were released, soon thereafter, President Williams of the New York Fed went out and kind of said, “Well, no. I think there's a lot of liquidity out there.” So, it looks like there are a range of views in the committee about how they should be proceeding, and it does seem possible to me that there is an interest in exploring the lower boundary of where they could be. Now, it's important that, if they do that, they go back in and manage large swings in reserve balances. That's what they didn't do in September of 2019 that really got them in trouble. Everybody knew that there was going to be tightness. The Bank Policy Institute wrote a month before that there's going to be tightness. But once you start getting down into that zone, you need to go back in and manage the big swings in reserve balances again. If you do that, you can really start beginning to see how things could shrink. I'm optimistic that there is some thinking that way that’s going on.

Beckworth: Bill Nelson, the eternal optimist for the Fed's operating system. It is ironic, though, when we think about the arguments made for the abundant reserve system. What was the term you used for it?

Nelson: Excessive.

Beckworth: Excessive reserve system. And one of the key arguments was that it's going to be simpler. You don't have to forecast reserve demand. But, lo and behold, it's a constant struggle. Have we hit that threshold where we fall back into a corridor system? That happened in 2019. Like you said, they had to think about big swings in reserves. So, the job hasn't gotten any easier, and as you've told me many times, if anything, it's gotten tougher. They've hired more staff to help forecast. So, life hasn't gotten simpler for them, so you can throw that argument out the window for it. Well, let's move on to another item that came to our attention that the Fed released recently, and that is the losses on the Fed's balance sheet. They gave the year-end report, I believe, and [there were] $114 billion in operating losses in 2023. What are your thoughts on that?

Breaking Down the Fed’s Balance Sheet Losses

Nelson: So, yes, as you said, normally, the quarterly or the annual statements are released two months after the end of the quarter. They, I think, responsibly released, in the middle of January, a statement providing information about the quantity that they've lost over the past year. So, the Fed, like any organization, it has investments and it funds those investments with borrowings. It used to be that those borrowings were almost entirely currency, which bears no interest. So, the Fed makes money, is inherently a money-making machine, and that money is turned over to Treasury, and that's revenue for taxpayers.

Nelson: So, we are all able to have more stuff or lower deficits or lower taxes, because of that money that the Fed hands over. But beginning in September of 2022, because of the sharp increase in interest rates, the Fed owned a lot of longer-term securities, but was funding itself with a lot of shorter-term securities. Because its balance sheet is now vastly bigger than the amount of currency, most of its borrowing is interest-bearing, it's mostly overnight. It's almost entirely overnight. And so, as the Fed sharply raised rates, its own interest expense rose up by more than its interest income, and it started to lose money.

Nelson: And it's continuing to lose money. It's losing money— last week, it lost about $2.25 billion, each week, about $2.25 to $2.5 billion. In 2023 as a whole, as you mentioned, they lost $114 billion. Now, these are real losses. This is actual revenue that wasn't handed over to the Treasury, and it wasn't handed over to taxpayers. And It's even a bit worse than that, about double worse than that, and that's because, if the Fed had done nothing except invest the funding that they get in the form of currency, if they just invested that in Treasury bills and had done nothing else other than collect the interest— the seigniorage that they get from that activity— they would have made $119 billion. So, they actually lost— to be able to wipe out that profit— they actually lost $233 billion last year. Just to put that in context, that's a seventh of discretionary spending of the federal government, so it's not an immaterial amount.

Nelson: Now, the Fed isn't in the business to make money. They're in the business of managing the macroeconomy. And so, to me, the critical question is, when they took the decision at the end of 2020— after the economy was beginning to improve— when they decided that they were going to continue to buy long-term Treasuries and agency mortgage-backed securities at an extraordinarily rapid clip, lengthening the duration of their portfolio, lengthening the federal government's consolidated duration, funding that with an ever-growing amount of overnight borrowing, were they aware of the risk that they were taking? Did they make a calculation that said, “We think that the benefits that will likely accrue from this are much greater than the costs, the risks that we're taking?” We may not know the answer to that until 2025, when the 2020 transcripts and material are out, but I think that's the critical question. Losses themselves will not prevent the Fed from doing its job.

Beckworth: Yes, and so I want to be clear, we're not criticizing monetary policy, per se, but the way it's implemented, the operation of it, and whether that has distributional effects and also legitimacy, like who does Congress delegate the authority to manage the public debt? Treasury, not the Fed. I wonder, Bill, if one thing that really could drive the conversation towards a serious reconsideration of this framework is simply the interest rate environment that we find ourselves in.

Beckworth: So, the decade prior to 2020 was a low-rate environment. Large balance sheets seemed to be a win-win for everyone. But if we're returning to an interest rate environment that's much higher, more normal pre-2008, it may not be as advantageous to have a large balance sheet. The Fed may find it more challenging. Maybe even with higher rates, a normal yield curve will reemerge, and the Fed could still operate something large with 5% short-end rates, maybe 6% high-end rates. But, there is this risk, right? If you're in a higher interest rate environment, you might have greater interest rate risk than if you were at 0% and able to make some profit, some arbitrage between the short-end and long-end of the Treasury yield curve.

Nelson: You can certainly operate a profitable central bank if you're in a high environment rather than a low environment, and in fact, the benefit that accrues from having a big chunk of zero interest-bearing liabilities goes up in that environment. There was a really interesting paper written by the staff of the Federal Reserve System in 2016. It was subsequently published, and this was part of, as I mentioned, the review of their balance sheet and operations in 2016, and the paper was quite ahead of its time. It actually investigated and talked about a lot of the issues that got a lot of attention after the SVB/Signature failures, the importance of the value of a deposit franchise, and the loss of value of that franchise, potentially, under some circumstances.

Nelson: But what the authors calculated was that if you have about a trillion dollars in reserves, then you have about a 10% chance of being in a situation in the future where you're making losses. But if you have like $3 trillion in reserves or $2.5 trillion in reserves— remember that the more reserves that you have, the smaller the share of non-interest-bearing liabilities, and so the less inherently profitable you are. And they calculated that if you were in that situation, you faced about a 60% chance of making losses sometime or another. The more the balance sheet exceeds the amount of currency, the greater the probability that you're going to make losses. Now, it's a little disturbing to read another memo in 2018, another staff memo, that basically said, “Oh, the risk of this is very low. In almost all circumstances, your interest income will be above your interest losses.” So, it's not clear. It is a little worrisome when you ask, were they cognizant of the risks that they were taking? They seem to have gone sort of in the wrong direction.

Nelson: As our friend George Selgin often mentions, the other problem is that, when you have this unlimited balance sheet, you begin to solve more problems with your balance sheet, and maybe your willingness to— I mean, it'll be interesting. Maybe the willingness of the Fed to take all of that extra interest rate risk was abetted, to some extent, by the view that there was not much cost to getting bigger and bigger. [And] certainly, you hear increasing discussions from politicians about how, well, you can use the Fed's balance sheet to solve this problem or that problem. There are definitely systemic problems associated with operating with a floor system that can all contribute to increasing the likelihood that you will make losses.

Beckworth: Right, so, at a minimum, if you have a large balance sheet, you're going to have far more reserves than currency, which, [with] reserves, you've got to pay interest on them in this framework. Reserves are the zero interest-bearing liability, so there's going to be less profits. We can at least say that much, and in some cases, actual losses, depending on where we are in the business cycle, what happens to the yield curve. So, yes, I guess it's something to think about, and, again, if we're in an environment going forward, like we've been in, I think it will give Fed officials pause and, maybe, “let's think this through.” It's easy to become complacent when everything is going in one direction and you're $90 billion in remittances to the Treasury every year. That's a wonderful place to be, but maybe we won't always be there.

Nelson: I agree.

Beckworth: Okay, well, with that, our time is up. Our guest today has been Bill Nelson. Bill, thank you for coming back on the program.

Nelson: You bet. Thanks, David. Always a pleasure to be here.

About Macro Musings

Hosted by Senior Research Fellow David Beckworth, the Macro Musings podcast pulls back the curtain on the important macroeconomic issues of the past, present, and future.