Steven Kelly on the Financial Stability Implications of the Discount Window

Steven Kelly is the Associate Director of Research at the Yale Program on Financial Stability and is also a returning guest to the podcast. Steven rejoins David on Macro Musings to talk about the financial stability implications of the discount window. David and Steven also discuss the issues with FHLBs, how to fix the challenge of reporting requirements, restarting the term auction facility and committed liquidity facilities, and much more.

Check out our new AI chatbot: the Macro Musebot!

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: Steven, welcome back to the program.

Steven Kelly: Great to be back, David.

Beckworth: It's great to have you on, and I am doing this show because I saw you moderate a panel that the Atlanta Fed put together for its annual financial conference, and it was a really fascinating conversation. But before we get into that, Steven, would you share with us a bit about your center there, what you're doing on financial stability?

The Yale Program on Financial Stability and Steven’s Role

Kelly: Yes, so we're based on the fun presupposition that financial crises are not going to be prevented in every state of the world, and if you think about it, you don't want to design a system where the probability of a financial crisis is zero, right? That's not the optimal social percentage. And so, really, what we have here is freedom to think about crisis fighting, exclusively. The rule of crisis fighting policy amongst several international organizations is like, “don't talk about Fight Club,” and some of that is very real, because the second that Jay Powell comes out and says, "Well, we're thinking a lot about how to fight financial crises," people get nervous, right?

Kelly: So, it's a little bit of, there are thousands of excellent economists working on what to do to prevent the next financial crisis. Then, if you say, "Okay, how many of you are working on, what do we do and how do we design things when that prevention fails?" The hands go down. So, we're really focused almost exclusively on building a playbook for modern financial crises and how you get a financial system restarted again.

Beckworth: What is your role in the Yale Program on Financial Stability? What are you doing?

Kelly: Well, it depends, because when I started, it was to work on 2008. Then, we're supposed to be cicadas, us financial stability folk. We're coming up too frequently, basically, with 2020, and now 2020 is a historical example because of 2023. So, as things come up, we're very much working on current events. Obviously, we're going to talk today about a lot of the reforms that are coming up post-SVB, post-2023 banking crisis, but we do a lot of things here. We have a platform online called the New Bagehot Project, and this, again, speaks to our goal of updating the playbook for fighting financial crises.

Kelly: The one thing that you'll hear every central banker cite is Bagehot's dictum, from Walter Bagehot. About 150 years ago, he said, "Lend freely at a penalty rate against good collateral." And so, we know that that's not enough to stop a financial crisis. And so, we're building that out, and we have an archive of historical interventions, and whether it was capital injections, liquidity, restructuring, changes in the rules, we have hundreds of historical cases out on our platform that you can click through and see what design components worked and what design components backfired. The idea being that there are huge costs to getting these things wrong or to delay in implementing these things.

Kelly: So, we're trying to really gather all of the ideas that have been out there over historical crises for historical interventions, and let's come up with a playbook that works, and let's make it so that policymakers don't even have to think about it. They can just go to our New Bagehot platform, go to the other things that we're writing, and say, “Okay, here's the design features that we need to think about based on this problem that we're having in our financial system.” So, we have the New Bagehot platform, which I encourage everyone to go check out. We have interviews with historical crisis fighters, what they were thinking at the time. We have tons of interviews that we've conducted with folks who've fought crises in the past. We have a Journal of Financial Crises that comes out quarterly, where we publish case studies and other articles on financial stability. We have conferences. We're staying busy over here, and like I said, a little busier than we were expected to be with current events over recent years, but nevertheless…

Beckworth: So, one of the things that you did recently in your work is you moderated a panel, as I mentioned previously. It was a part of the 28th Annual Atlanta Fed Financial Markets Conference, and this year, it was titled, "Central Banking in a Post-Pandemic Financial System." So, you were there. Now, was this in Florida? My understanding is that it was off the coast of Florida. Is that right?

Kelly: Yes.

Beckworth: So, sunny Florida in May, what a perfect place to be talking about financial stability issues. And your specific panel was titled, "Domestic Liquidity Provision During Potential Crises," and it focused largely on the discount window and issues related to that. So, you led out, [and] you had your own remarks. In fact, we'll provide a link to your remarks. They were informative, too. Great person to lead out in this conversation, but you had several guests. You had Bill Nelson, as listeners will know, past guest on the podcast, your colleague Susan McLaughlin, and then Luc Laeven, if I'm saying his name correctly. He's the Director-General of the Directorate of General Research at the ECB.

Beckworth: So, we had two people from the Yale Program on Financial Stability there, and then Bill Nelson, BPI, and then an ECB official as well. To kick this off, Steven, let me read some of your remarks. I want to use your remarks as a way to get this conversation going, what you guys chatted about. I want to read the first two paragraphs. I think it captures the spirit well of what went down. You say, "It's rare that, after a financial crisis, political consensus begins to form around the government more ably supporting banks, but that's precisely the moment we are in.”

Beckworth: "Where political furor would normally spell the end of some crisis fighting tools, it has instead focused on how to make them more effective, at least in the case of central bank lending and the Fed's discount window." You go on to note that there's broad support for this. It seems to be having a momentum of its own, it's moving forward. And as it turned out, at that conference, Vice Chair for Banking Regulation, Michael Barr, had some announcements. So, I want to talk about this issue here. First, let's talk about the momentum behind it. You've touched on it already, the SVB crisis, 2023, the banking turmoil then.

Beckworth: Also, we can talk about Michael Barr's speech, and then maybe some of the issues going forward, as well as, is there congressional support for it? So, I had Bill Nelson on previously, and we chatted about some of this interest. There was a G30 report, I believe. The acting Comptroller of the Currency had a speech on it. Bank supervisors were talking about it. But was Michael Barr's speech the official launch, or what was accomplished in his speech at this conference that led into your panel discussion?

Building a Resilient Regulatory Framework

Kelly: Yes. This was the first real clarity we got from the Fed on what they're thinking about, liquidity-wise, post-SVB. Obviously, all of the stuff with Basel III endgame sort of predated SVB, and then it's all gotten mixed up, and that has been front and center. We haven't had much clarity from, exactly, what the Fed is thinking on liquidity, until this speech. Granted, it was all leaked in the New York Times several months ahead of this. There was some more reporting in the Wall Street Journal. But what we got is a few things of how the Fed's thinking about this. One, is that, broadly, he said that we're looking at the scope of our liquidity regulations, in general.

Kelly: And so, I think that the way to read this is probably like, okay, SVB shouldn't have fallen outside of the LCR world. So, there's been a general rethinking about what constitutes a big bank, post-2023. So, that thinking is going on. But then, he offered some more specifics, one of which was to limit banks' inclusion of held-to-maturity assets towards their liquidity regulations, so, limiting the inclusion of held-to-maturity assets towards what's called HQLA, or high-quality liquid assets. And so, this makes sense on its face, right? To the extent that you're telling the accountants, "Oh, we're going to hold this to maturity," you can't be telling the banking regulators, "Oh, we're going to sell this if we need liquidity."

Kelly: That's just a harmonizing, really, of accounting with the regulation, and it makes sense with the experience that we saw last year. Banks are very hesitant to sell held-to-maturity assets that have unrealized losses on them, because the second you sell them, you have to recognize the losses and, potentially, the losses in that whole accounting categorization. Perhaps more relevant for our discussion today were two other things that he mentioned. One was to ratchet up the deposit outflow assumptions in liquidity regulations for certain customers. He specifically mentioned high-net-worth individuals and crypto and VC firms.

Kelly: We can talk more about this in a minute, but just the last thing he said [was that] they're looking at, particularly for-- again, this is all for banks, he said, of a certain size. So, it's kind of a guessing game at that point, but you can think of, maybe $10 billion, maybe $50 billion, maybe $100 billion. So, he said that the Fed is looking at requiring a level of discount window preparedness. So, this is just the ability to borrow from the discount window, which banks are woefully unprepared to do. And as part of that, the Fed is looking at requiring a total sum, of both pre-position collateral and reserves, that equals some percentage of a bank's uninsured deposits. The Wall Street Journal has reported that the Fed is sort of circling 40%. So, figure [that] reserves plus the haircut adjusted value of pre-position collateral must equal at least 40% of your uninsured deposits. So, you can think of a bank like SVB, which was approaching $200 billion in uninsured deposits, would need some level of collateral at the window, plus reserves, that would effectively equal 40% of those uninsured deposits, instead of the $5.3 billion that it was able to borrow before it failed.

Beckworth: So, this provides clarity in where the Fed is going with this. Do we have any timetables for when it will actually go into action?

Kelly: We don't. We don't. And as you mentioned, there is activity on the Hill as well, and these discussions are all happening at once, both in the House. There's been a bill from Representative Barr— no relation, as far as I know, to Michael Barr— that, basically, just asks the Fed to study the discount window and come up with a remediation plan for all of its shortcomings. Senator Mark Warner has announced that he's releasing a bipartisan bill. It's sort of gotten punted a little bit, but it may be out by the time this podcast airs. But that's looking at mandatory testing of the discount window, giving banks credit for their discount window preparedness in their liquidity ratios. And we can talk more about that. Expanding the discount window hours— We saw that West Coast banks were disadvantaged, basically, by the Fedwire hours, as well as general things like reducing stigma and increasing coordination between the Fed, regional banks, and the FHLBs.

Beckworth: Okay, so there is much happening. There's work going on in Congress. The Fed has proposed what it's going to do. And as we mentioned, a lot of momentum is already behind this coming out of last year. So, one thing that Vice Chair Michael Barr mentioned, in addition to the discount window material— I want to come back to that, but he also mentioned, as part of a package, that it would be good for banks, also, to think about the resolution resources, I believe, is the term he used, so, like living wills, which we're familiar with. But he also pushed long-term debt, and I just want to ask you about that briefly. Again, we'll go back to discount window issues in a minute. But is he talking about contingent convertibles, CoCo bonds? Is that what he's thinking about here?

Kelly: So, in the US— this is not the Credit Suisse thing. This is really about, when a bank fails, a layer that you can then bail in, in resolution. The biggest thing that you get from this, I think, is protection of the deposit insurance fund. You have this extra layer of debt that, when a bank fails, you can bail in these debt holders. That's what he's thinking about there.

Beckworth: Okay, so, not CoCo, it's totally different. But that was an interesting development, you're right, with the banks in Europe recently. Okay, back to the discount window issues. So, we have this momentum going and a lot of excitement, maybe, behind it. I know that Bill Nelson and I have chatted about this some more, and I'll come back to some of the questions that he has raised about it. But in your remarks, going back to that, you mentioned this excitement, but then you also throw in some caution, like, would it have made much difference in 2023 with specific banks versus, say, a systemic crisis? So, maybe you can speak to that.

Addressing Issues in the Discount Window

Kelly: Yes, it's sort of put me in an awkward position as someone who's been thinking about ways to reform the discount window and really not believing that it can accomplish a ton of what some folks want it to. I share the examples in the remarks, which I encourage folks to check out, not just because my remarks are so great, but it's also heavily cited. So, there's a lot of good reading stuff for folks interested in the issue. I mentioned in the remarks that Randy Quarles is on record saying, "Hey, if the discount window worked, if the discount window was functional and worked the way it was supposed to, SVB would still be here."

Kelly: And Michael Barr, his successor in the [Vice Chair] for supervision role, has said, "Okay, discount window access, or lack thereof, or its functionality, is not the reason that these banks failed," which I think is much more convincing, not least because, even from an accounting perspective, SVB was underwater. But I wanted to raise the issue of, okay, we're sort of agreeing on some of these technical changes around pre-positioning or hours or mandatory testing, but we have— and I'm saying we as the royal we, folks thinking about this— have, still, a wide spectrum of views on what the discount window can accomplish.

Kelly: And I think it's important that we give up the idea that the discount window alone can really rescue a bank that is facing a run in the market due to perceived non-viability. If you think about what it means for a bank to go to the discount window, that's facing a run, it's replacing its depositors with the Federal Reserve. So, not only are its interest costs going up substantially, from effectively zero to, in present case, over 5%, but your depositors are your future lending franchise. They're who your employees have customer relationships with. So, the idea that we can keep a bank in business that has been deemed non-viable just by reassuring customers of full repayment, I think, is lacking.

Kelly: That’s a necessary but not sufficient condition, and I share the examples of Credit Suisse where they put over $200 billion in the window for Credit Suisse, which was more than any estimate of how much liquidity could have run, how much it would have needed, and you saw the run continue. Then, I've talked about this on this podcast before, but the SVB Bridge Bank, which took on the deposits immediately— or, not immediately, but the Monday after SVB was closed— was advertising that it had unlimited FDIC protection for even new deposits. So, the SVB Bridge Bank was advertising, basically, like Fed accounts.

Kelly: This was the safest bank in America, and it continued to bleed deposits, because there are other reasons to run from a bank. And you're thinking about your future relationship with the bank, its ability to be there for you in the future. So, anyway, the point being is that once a bank has been deemed non-viable, it's very hard to recover from that, and the discount window certainly isn't enough. I like to share the analogy of being broken up with or attempted to be broken up with. Imagine Friday, ideally, after the bank closes, you get broken up with, and then Sunday night, before Asia opens, she says, "Alright, we'll give it another shot." Come Monday morning, you're still going to be like, "Whoa, I almost got broken up [with]," and you're going to have a wandering eye more than you would have before.

Kelly: And that's sort of an analogy for the depositors, for the employees. Nobody wants to be an employee at a bank that almost failed, or at a bank that is funding itself via the government. Anyway, I wanted to put that out there as one thing that we cannot forget, is that the discount window is great from a systemic perspective. It's not going to save that one bank. It might get that bank to the weekend, which would be fantastic. And if you think about what was most damaging about SVB's failure when it did fail that Friday morning, was that, originally, the government said, "Hey, we're not going to bail in, we're not going to rescue uninsured depositors."

Kelly: “They'll get some of their money next week, and then we'll find out what they get in resolution.” But by Sunday, they do the systemic risk exception, right? And so, if you have that time, if you can get SVB to the weekend, and maybe you get your ducks in a row to do other more substantive policy responses like the systemic risk exception, you contain some of the damage. So, it's great in a world where you're thinking about systemic demand for reserves going up, or when you're talking about getting a bank to the weekend, but you're not going to save a bank.

Kelly: Okay, so that was one thing. Then, the other thing is, fire sales aside, you think about the discount window like, okay, we're trying to get banks to not fire sale their assets and face fire sale values on whether it's loans or whatever else. But there's a step that happens before that, which is, when you take an asset off a bank balance sheet, you've lost the franchise value that comes with held-to-maturity funding at the deposit curve, which is lower than even the Treasury curve. So, irrespective of fire sales, once you take an asset off a bank balance sheet, there's immediate franchise value loss and asset value loss, and we can talk about the BTFP and the implications of that.

Kelly: And so, I sort of raised the elephant in the room, which is, if you're in a world where every asset has fallen substantially in value because of interest rates— 2023 wasn't really about credit risk. We're thinking about interest rates. That's much harder for the central bank to ignore. In 2008, literally FASB is coming out and saying, "Hey, if stuff's really illiquid, you don't have to market exactly to what markets are saying." They're never going to do that with interest rates.

Kelly: Interest rates are clear as day, and they're endogenous to monetary policy, not to the central bank's financial stability policy, which can contain certain credit risks and make its valuations true in the end. So, those were the challenges that I raised, and the BTFP, I think, shows that we need to think about valuing collateral the same way we think about setting the rates or collateral eligibility for a crisis time facility, which is— We think about what would be the case in normal times, right? You don't set a penalty rate relative to the crisis time rates, when market rates go way up.

Kelly: You, as the central bank, don't add the penalty rate to that. You say, "Okay, what was the interest rate the day before the crisis? And we'll add 50 basis points to that, and that's our rate." You do the same thing with collateral eligibility. Everyone's worried about housing bonds, but you say, "Okay, but yesterday they were AAA and calm, so we're going to let housing bonds be eligible collateral." But we haven't done that with valuation, and there's a valuation advantage to being able to keep assets on bank balance sheets.

Beckworth: These are all important issues, and one thing you noted in your piece that really struck me, and I've heard you say this elsewhere, is that the Fed could recapitalize the entire banking system simply by lowering rates, right? Get lower rates, suddenly those bond prices go up. So, a lot of these issues you mentioned are endogenous to policy. There's also credit risk, which is different. It's endogenous to the Fed facilities that you've outlined. To summarize, there are issues with the individual banks in terms of this increased use of the discount window, but overall, we're moving in the right direction for a systemic crisis. Is that a fair assessment?

Kelly: Yes.

Beckworth: Okay, so, let me give a little pushback on the whole notion of these liquidity regulations. So, as you know, I recently had Anat Admati on the show, and she's all in on the capital regulations, making sure banks fund with more capital, but she really has no time or place for liquidity regulations. So, let me just read a quote from that show. This was March 18, 2024. I want to get your response to it. I had just said, "Hey, I see you're shaking your head," when I was doing the video with her.

Beckworth: And she said this: "I'm shaking my head because I'm not a fan of liquidity requirements, number one, at all. We created central banks to solve pure liquidity problems. If a bank is insolvent, it doesn't have a run, because it has a central bank. I view liquidity requirements, the liquidity coverage ratio, and all of that as costly on good days and useless in the run. The cost-benefit is just not there." Then, she goes on to praise capital requirements. So, what would you say to someone like that who is skeptical of these type of exercises? The discount window discussion we're having today is about the liquidity requirements and giving banks better access to liquidity in crises. So, how would you respond to that?

Responding to Criticism of Liquidity Regulations

Kelly: Well, I think she's absolutely right about these regulations being useless in a run. If you look at banks’ high quality liquid assets as a percentage of total assets, they have effectively quintupled since 2008. What do we have to show for it but the fastest bank run in history?

Beckworth: Fair enough.

Kelly: So, it goes back to my point about SVB. It's never going to be good enough for SVB. It's helpful system-wide to some degree. Like, if you're a Schwab, you benefited from having liquid assets. Then, there's the question of, is that really what we want banks to be, to be doing all this self-insuring? And Anat offers an interesting corner solution, which is much higher capital. I have reservations about how much capital there is in the world. I think it's our scarcest resource, and you have to think about what exactly that means for the structure and size of the financial system.

Kelly: Perhaps the opposite corner solution is the Mervyn King solution, which he calls the “pawnbroker for all seasons,” which is that you get rid of all capital requirements, all deposit insurance, and you just require that all short-term liabilities from banks be backed by sufficient collateral at the discount window. And in that world, your capital requirements are basically the haircut that you face at the discount window. You're able to pay out all depositors, and you can wind down much more slowly. So, those corner solutions exist.

Kelly: I think there's drawbacks to both of them. We seem to be moving [more] in the direction of the King solution, and part of the value and the relative free lunch in pre-positioning for liquidity [regulations] is that there's a lot of collateral out there that really has no higher purpose. Banks are sitting on loan collateral. It's not as if they're repo-ing that on a daily basis, and pre-positioning collateral is going to reinvent the financial system. I think it would be the least disruptive, at least directionally, to get substantially more discount window pre-positioning.

Beckworth: I like how you approach this, Steve. You're very much an economist. There's trade-offs, right? There's costs, opportunity costs, and if you go to one corner solution, you're potentially giving up some gains, on the margin, going the other direction, right? So, maybe we've gone in one direction really strongly. It's time to look back, and maybe there's some big marginal gains going to the other corner solution, where we do park collateral at the discount window, make more use of it. And that is, in fact, what the world is doing right now, at least in the US. I guess the question is, are there similar developments, or have they already taken place in other major central banks like the ECB or Bank of England? Any knowledge there about what's happening?

Kelly: Not that I know of, really, substantially. Every jurisdiction looks at pre-positioning a little bit differently. I think the biggest reservation, which is a fair reservation, is, as central bankers, we don't want to imply commitment to giving some level of liquidity in a crisis. So, if you give me some collateral today, I'm not going to promise you that, when the crisis comes around, I'm going to give you X amount of dollars on it, and that makes sense. I think the problem is that— So, right now, we have something like $3 trillion of collateral sitting at the discount window.

Kelly: Banks are being told that that's worth $0 in their internal liquidity stress tests, in their LCR, in their net stable funding ratio, in their resolution plans. And so, you can understand the central bank view of like, okay, we have $3 trillion of collateral from you guys. Haircut adjusted, it would be, let's say $2.5 trillion, and in a crisis, the value might fall. It might fall to $2.5 trillion, in which case, haircut adjusted, it’s $2 trillion or whatever. But it's not clear that the most appropriate value to assign to that is $0. So, that's the trade-off world that we need to figure out, is there's a limit to how much the discount window can be encouraged by Jay Powell and Michael Barr going out and saying, "Hey, we love the discount window."

Kelly: There has to be some carrot involved. It's not going to be just a handout to the banks. Asking them to pre-position is costly. It will take tech costs and upfront charges and all of these things, and maybe you increase, maybe you tighten up. Like I said, Michael Barr is looking at tightening deposit outflow assumptions for certain kinds of depositors, so maybe you tighten liquidity [regulations] at the same time, but there's some carrot where you're giving credit for pre-positioning and not pretending that this discount window— which is not a crisis time tool, right? It's a through-the-crisis tool, same with the standing repo facility, not pretending that that's going to be worth zero.

Beckworth: Yes, so, the push here is to make all that collateral, that's currently sitting at the discount window, count towards liquidity requirements, at least some part of it, a haircut. And maybe in the future, banks will park even more collateral there if they see that it's a good investment of their resources. But, to get there, there's certain hurdles we've got to get through, one of them being the stigma. And I know Michael Barr, in his speech, he talked about this. He mentioned that they really have to get on top of the bank supervisors and examiners to change their mindset.

Beckworth: So, it's not going to be something that happens overnight. It's going to take some work. And there's a number of issues, and I want to bring them up beyond just the general stigma issue. One of them that you have written about is in an article titled, "Weekly Fed Report Still Drives Discount Window Stigma." So, you talk about how the reporting requirements that the Fed currently does still create challenges that makes banks uncomfortable to use the discount window. Tell us about that challenge.

Fixing the Challenge of Reporting Requirements

Kelly: Yes, so the Fed— As many listeners will know, the Fed, every week, publishes a balance sheet called the H.4.1, where it discloses balance sheet items as of that Wednesday and the week average. It closes every Thursday night. And as part of that disclosure, it's broken out by the 12 regional Fed banks. And so, the historical issue with this disclosure was, okay, if you're a reasonably sizable bank in one of the given Fed districts, based on where your headquarters is is sort of where you borrow from the Fed. If there's rumors going around about, say, Wells Fargo, and all of a sudden there's a huge increase in borrowing at the San Francisco Fed, that's going to further stigmatize Wells Fargo.

Kelly: It's going to encourage rumors, and it's going to encourage a run. And the Fed is very aware of this, and what it did in 2020, at the start of COVID, was it said, "Okay, now, when we break down the regional balance sheets, we're going to combine discount window lending, at the regional level, into our securities allocation. So, when we're doing QE, we have a massive SOMA portfolio that's allocated across each regional bank, and we're going to stuff the discount window borrowing into those numbers, and that's going to help protect and sort of disguise those numbers." And that's true to some degree.

Kelly: The problem is that QE is public and, basically, you can back out. And so, [in] the note that you mentioned that I wrote, I sort of played hedge fund analyst for the day and wrote this thing. It’s very real. Bill Demchak, the CEO of PNC, at a Brookings event a few weeks back, said, "Look, these disclosures are regional, and the second we borrow, it will show up in that data." And so, PNC is in the Cleveland Fed District, so, people are thinking about PNC, and all of a sudden, $10 billion gets borrowed. Who else in the Cleveland Fed District might that be? And so, that's sort of the concern.

Kelly: But, so, anyways, now you can still back out QE, basically,  based on assumptions and disclosures about the allocations across districts. But also, when the Fed is not doing QE, it's really easy to back it out. So, literally, the week of SVB, the Wall Street Journal published this chart of borrowing by district, because the Fed was done doing QE. So, it just looked at the increase in regional Fed balance sheet totals, and there was a ton in San Francisco, of course. There was a ton in New York, and so, folks were relatively calm with that, because it's SVB and First Republic and all of these names we already know out west, and it's Signature Bank in New York. But there's sort of an alternative world where PNC does want to tap the window, and all of a sudden, the Cleveland number goes way up, or I mentioned Schwab before.

Kelly: Folks may remember that Schwab was on the brink, or at least in the headlines last spring. Imagine that the Dallas district goes up $10 billion or whatever, and it's [like], "Okay, Schwab's at the window. Schwab's going down," and the run is perpetuated. So, this risk is still out there. The Fed did a little to mask it, but it's really a few Excel cells away from backing the number out. So, you can't casually look at the number, but any hedge fund can find it or back it out. Any reporter can do the same. And so, I sort of argue that they need a little more aggregation to disguise those numbers.

Beckworth: Yes, it was really interesting to see your exercise where you did back out the actual amounts, whether it's from looking at the aggregates when QE has ended and you can still see the change, or if you go back, historically, to when they did report the differences, and just simply add things up to the present. So, you want to aggregate more information. Would you also change the timing? Do you still want it to come out weekly, or just aggregate it up?

Kelly: So, the Federal Reserve Act requires the Fed to disclose its balance sheet weekly, so, absent a change from Congress. But the fact that they've already done this aggregation— I'm arguing for just a little bit more aggregation. And really, I think, if you think about the spirit of the Dodd-Frank Act, which now requires that the Fed release borrower identities on a two-year lag— technically it's eight calendar quarters— that's already proven to be stigmatizing. And so, the fact that we're going from that to, “if you're a sizable regional bank, we can find you every week,” that's incredibly stigmatizing. And really, even folks who are for the disclosure in Dodd-Frank, you must see that this undermines the spirit of that compromise of, "Okay, we'll at least wait two years."

Beckworth: Yes, so, it'll be interesting to see where this goes, and this is one practical suggestion on this journey to better use the discount window to meet these liquidity requirements, and hopefully make banks more secure as a whole of the system. And we'll come back to this later, but Bill Nelson's goal of also shrinking the Fed's balance sheet— he would hope that this would reduce the structural demand for reserves, if they can simply go to the discount window. Alright, so, that's one potential solution, or a fix towards getting us closer to this destination. Another one I want to bring up comes from Bill Nelson, and he had an article recently titled, *Something Old and Something New: Two Potential Beneficial Discount Window Facilities.* His basic idea is this. He wants to restart the term auction facility and, also, really work up these facilities called committed liquidity facilities.

Beckworth: So, the term auction facility, many will remember, was used after the great financial crisis. The Fed would auction off funds from the discount window, and it had a whole lot less stigma associated with it, because you weren't going to the discount window, or you were going to the auction. The committed liquidity facilities would be collateral that is committed to the Fed, and in turn, the Fed commits lines of credit to the banks. So, his point is, together, if we start doing the term auction facility, so banks get used to it, they get comfortable going to the discount window, and then you combine it with the committed liquidity facility, you're really on a path of getting past the stigma barrier and to making the use of the discount window normal operating procedures. Any thoughts on that?

Restarting the Term Auction Facility and Committed Liquidity Facilities

Kelly: Yes, I buy that totally. The TAF, the term auction facility, in 2008 or 2007, I guess, really was stigma-reducing, partially because of this auction mechanism, which we found here at the Yale Program on Financial Stability, is really helpful with stigma. The other thing is sort of a historical accident, which is that, for technical reasons, the funds from the TAF auctions didn't settle until a day or two later, and so the market was able to view this as like, "Oh, this is not a bank who is in desperate need of funds today. This is a bank that can wait two days for the funds, and it's just getting a good rate on some market-based funds." So, there are some interesting things there.

Kelly: The committed liquidity facility idea is a sound one. I think the challenge— and I've been trying not to mention the FHLBs until this point— but they are a challenge in this space, because the Fed's discount window is legally constrained to four months max. You can roll that over, but it cannot make more than a four-month loan. The FHLBs have term structures out 30 years, and so the FHLBs are, really, partners with banks when it comes to establishing liquidity lines and building out a liquidity framework, and it would be very difficult, again, absent some regulatory stick, for the Fed to usurp that role as long as the FHLBs exist as they do, and the Fed's limits are what they are.

Beckworth: Yes, so Bill does bring that up, the FHLB issue, and he hopes that the TAFs, the term auction facilities, would maybe take away some of that, but you're saying that the advantages offered by the FHLB, in terms of longer-term lines of credit, may outweigh getting familiar with the discount window and all of that. One question on the committed liquidity facilities, and maybe this is a question for Bill Nelson, but I'll throw it at you, too. How is that any different than what's being proposed currently for the discount window? So, my understanding is, again, we want to use the collateral at the discount window to count towards liquidity requirements. Is that any different than what he's proposing with the CLFs?

Kelly: So, it's slightly different in that pre-positioning doesn't give you— when you think about the way that you get a credit line from a bank, it's, "Oh, I have a $100 million credit line." Pre-positioning doesn't give you a $100 million credit line with the Fed. You can put $120 million of collateral and expect to get $100 million, but it's not, to use the word, committed, and that's part of the challenge of liquidity planning and the culture of the Fed and whatever else.

Beckworth: Yes, he mentions, I believe, Australia and South Africa. Their central banks have used these before. One other interesting thing from his article, just to throw it out there, [is that] the term auction facility was actually an idea that was originally proposed, and it was called the auction credit facility in the early 2000s in response to the fact that we were actually running down the national debt. There was a worry that we would run out of Treasury securities for the Fed to hold on the asset side of its balance sheet, so, "Oh no, what will the Fed do?" And they figured, "Well, we can maybe do more discount loans, but how are we going to do that with stigma? Well, let's do an auction credit facility,” was a term they called it, and then it was renamed to term auction facility in the crisis.

Beckworth: So, [it was an] interesting turn of events, from a shortage of public debt to now an overabundance, where we have a Treasury market that struggles with handling all of it. Okay, so, if Bill Nelson's ideas go forward, add those on top of your proposal for aggregation, maybe we make some progress. But let's circle back to the FHLB issue, because you have an article that you wrote with some of your colleagues at the Yale Program on Financial Stability and maybe some suggestions for dealing with that big thorny problem.

Addressing the Issue with FHLBs

Kelly: Yes, so, there's a lot wrong with them, to some degree. First, I should say that you can't just destroy them overnight. As I alluded, they are liquidity partners of banks and they're an essential part of the system now for reallocating liquidity and [they are] part of banks' contingency funding plans. What I wrote with colleagues, Susan McLaughlin and Andrew Metrick, [is] we go through how the FHLBs actually reimburse— Because the FHLB’s members are their member banks, they pay dividends to their member banks, and it's not based on size or housing activity or anything like that. It's based on how much you borrow.

Kelly: So, imagine the Fed paying out its profits to the banks that borrow the most from the discount window. It goes into the pricing advantage that FHLBs often have, and so our proposal is, "Look, if you want to keep those dividends to members, fine. That's part of how the system works, but let's pay them out based on the FHFA's housing goals as opposed to how much lending banks are actually doing, into whether it's low-income housing or housing, pick your allocation,” but not based on, "Oh, I took a bunch of Treasuries or MBS to the FHLBs and did some low-cost borrowing and now I'm getting a dividend."

Kelly: So, that was our proposal in that article, but there are various challenges. What we hear from banks, and the surveys reflect this, is just that the presence of the FHLBs adds to the stigma of the window. The FHLBs have much less disclosure. We talked about disclosure before, with the discount window. [With] the FHLBs, there's much less disclosure. Really, in this article, we talk about how really inaccessible even the lending terms are at most of the FHLBs.

Kelly: But there's all of these problems. We saw this with SVB too, which, one of the issues of the FHLBs is, if you want to borrow a substantial amount of money, the FHLBs have to go out and raise it. And so, you may be getting money at T+1 as opposed to T+0, which in SVB's case, made all of the difference. If you're getting to them late enough in the day, they can't print money the way the Fed could, so, their existence just kind of gets in the way. It can slow down the Fed's ability to get sufficient collateral. The FHLBs don't really have the prudential insight that regulators do. And so, what we see is, basically, when a crisis gets bad enough, the FHLBs raise haircuts, or they'll just cut off lending, and we don't see that behavior from the Fed. So, there are all of these issues [that are] roped together, but the FHLBs are basically capturing seigniorage that they get from being counted as government debt, to issue into money market funds, and then they pay it out to those who borrow the most from them.

Beckworth: Okay, so, if you had to rank all of the challenges we've gone over, so, too much disclosure, just not being familiar with it, that's Bill Nelson's proposal, "Get us used to it. Get us going, " or the FHLB, and maybe some other ones as well, where would you rank those? Would the FHLB be near the top or would these other ones be pretty important, too?

Kelly: I mean, they're all at the top. These things all work together. I think that the biggest thing is that you have— as of now, you basically have no regulatory or supervisory incentive to be prepared to use the discount window to have collateral there. You can't have no carrot, basically, for using the discount window. You're never going to solve the stigma problem. You can't really solve it with pricing. I mean, part of the issue is pricing, and if you think about [what] I talked about before, going to the discount window, you're replacing low-cost deposits with market-based rates, and the Fed can't really go any lower on that.

Kelly: Right now, the discount window is priced at the top end of the fed funds rate. It used to be 100 basis points premium. You can't really put it below fed funds, because then you have an arbitrage, or you can't put it below IOR or whatever. So, you can't really do much more on pricing, and as long as the pricing exists the way that it does, you have to have other incentives. You have to at least beat out the FHLBs. You have to have credit for regulatory purposes or some other incentive, because as long as there's no incentives, it's going to be stigmatized, because it's going to be seen as the lender of last resort. It's going to be seen as, "Why are you even talking about it? Why are you practicing it? Why are you prepared to use it? Why is it part of your funding plans if you don't actually intend to use it?" So, just the lack of incentive is a problem, and it creates all of these other issues. It creates the stigma, it creates a lack of competitiveness with the FHLBs and all of these things.

Beckworth: So, just imagine a world where the FHLB went away. So, Congress acted, and I know it's unlikely, but just pretend with me that they disappear quickly. Would there be a missing market for banks? We mentioned earlier that the Fed can only go four months out on the discount window, and so the rest of that way, that term structure is being met by FHLBs. So, would there be a missing market? Would the Fed need to have some changes to its Federal Reserve Act, as well, to meet that market?

Kelly: Potentially. I mean, potentially, you get the private market to do this. Remember, the FHLBs are just a collective of all of their members, which is banks, credit unions, insurance companies. You lose the seniority and the joint issuance of debt and all of those things. But like I said, they're liquidity partners to their members in a way that the Fed will probably never be, even if you change the rules. So, you get some replacement with the private markets, but you run the risk— Like I said, you can't just destroy the FHLBs and then hope that the Fed improves. You have to have other incentives in place. The FHLBs, the week after SVB, I think that they borrowed half a trillion dollars in the market. Is the Fed really willing to step into that breach and do that much lending? You're talking about a cultural change, too. Again, it depends on the supervisor, it depends on the regional bank, but you're talking about a cultural change about how the discount window is viewed across the Fed system.

Beckworth: Let me ask the question this way. Is the FHLB truly a market innovation, or is it created by subsidies, implicit or explicit, from government? Is it there because there's truly a market need for it, or has it arisen due to government structure and design?

Kelly: I would say it's both. It definitely benefits from, call it seigniorage, call it arbitrage. The CBO put out a report that put its expected implicit subsidy, this year, at over $7 billion. Last year, which was a super profitable year for the FHLB system, they made a little over $6 billion. So, to some extent, yes, the profitability of the system only exists because of the subsidy. The main one being that it can issue debt into government money market funds.

Kelly: But there is some innovation, and there is, in theory, a private sector alternative to what it does as, effectively, a bank clearinghouse and the joint and several liability of all of its members that it uses, of the joint and several liability of the system, and it's supported by its members. That's advantageous. If you could recreate that, [then] you're back in the 1900s world of clearinghouses, but there were advantages to those. But, yes, you just can't destroy it overnight.

Beckworth: Yes. Well, it sounds like Vice Chair Michael Barr has his work cut out for him along with all of the other people at the Board of Governors and other bank regulators who are wanting to go in this direction of more use of the discount window. Now, let's circle back to your panel, because that was the original motivation for this show. So, again, it was at the Atlanta Federal Reserve Financial Markets Conference. It was in May, and you had two other guests. I mentioned Bill Nelson, but did Luc Laeven or Susan McLaughlin— did they have any other thoughts that you want to share with the audience?

Additional Thoughts from the Atlanta Fed Conference Panel

Kelly: Well, Susan, she has a novel idea. She really wants to separate more primary credit and secondary credit. So, these are the two levels of the discount window, primary credit being through-the-cycle. It's supposed to be no-questions asked. We sort of don't see that in practice everywhere, in every case, but it's for solvent, well-capitalized banks. Then, secondary credit is a little more punitive, and it's designed to be for banks that have low capital or that are otherwise going through solvency issues.

Kelly: She really wants to separate these facilities and, really, just have a no-questions-asked facility for solvent banks that— She talks about automation and things like that, which, again, closes some of the gap with the FHLBs, but automation, not all of this, "Oh, I've got to call the discount window, and then wait for approval." So, separating the so-called “good guys” and the good banks from what's, effectively, a resolution financing facility. She's got some articles out showing exactly how often people use secondary credit, or come off secondary credit, which is not frequently, It's effectively a bridge to resolution or other things. And she's looked at other jurisdictions as well, and there's much more distinction between the two facilities. And so, her hope is that renaming and redesigning these facilities can reduce some of the stigma associated with running them both in parallel. She's very focused on that.

Kelly: Luc was much more critical of the uninsured deposits situation that we ended up in. And, really, he saw this as a much higher-order problem than anything to do with the discount window, and [he] really thinks that we should have never let banks fund themselves with north of 90% uninsured deposits or 80-some percent uninsured deposits like we saw with SVB and Signature. There are challenges to that, because if you want to run a business like banking Silicon Valley, [then], by its nature, you're going to have uninsured deposits. There are some business models where, by their nature, you're going to have uninsured deposits, if you're BNY Mellon, for instance, or if you're trying to bank a certain sector. But he was very much focused on that more than the discount window and the BTFP.

Beckworth: Well, we'll provide a link to that panel in the show notes, so check it out, listeners. It's a great discussion. The whole conference is very fascinating, and I was really surprised to see how long this conference has been going. I believe it's at 28 years, so it's becoming the financial stability equivalent of the Kansas City Fed Jackson Hole meeting. Is that fair? I'm sure that the Atlanta Fed would say, "Yes, definitely."

Kelly: Yes, right, right.

Beckworth: They would say, "Come on. That's right. We are it." Alright, so, in the time we have left, I want to go back to Bill Nelson, as a friend of the show. And I share many views with him. [I’m] very sympathetic [to] many of his positions, including the ones we've just talked about. But one in particular [is that] he wants to see the Fed's balance sheet get smaller for a number of reasons, shrink its footprint [in the] financial system. Also, though, to, in part, return, at least, to some extent, some overnight interbank unsecured lending.

Beckworth: One of the motivations for this piece that I cited previously, where he called for reintroducing TAF and using these collateralized lines of credit from the Fed, is that it would, in theory, all else equal, reduce the structural demand for reserves by banks. If they know that they can go tap into the Fed, they don't need to sit on as big of a stock of deposits at the Fed. In fact, he noted in his piece— very interesting, we'll provide a link to that as well— that QE has, actually, if anything, added to the stigma of the discount window, because if you have all of these reserves, why would you ever go to the discount window unless you are a problematic bank?

Beckworth: So, QE, again, it's an important tool when it's used appropriately, but it can have side effects, like the ones he outlined. In any event, he wants to see us go in that direction, and I bring that up, Steven, because there have been a number of other central banks that have taken a look at their operating system. And so, you may have seen that I had Isabel Schnabel on the podcast. She's a senior executive at the European Central Bank, and she led a 15-month review of the ECB's operating system.

Beckworth: And what they're doing is, they're going to go from more of a floor system to what they call demand-driven, but somewhere in the direction of a corridor. It's not going to be quite a corridor. In fact, they're not the only one doing this. There's a number of central banks. There was the Reserve Bank of Australia, the Riksbank from Sweden, the Bank of England— they're all inching away from a floor system to what they call a demand-driven system, where there'll be more activity, like the repo facilities that they have, so banks will go to the facilities when they need more reserves or [to] get rid of reserves.

Beckworth: That's more market-driven as opposed to just loading them up with reserves. And, sometimes, the definitions get tricky. What they call ample might be different than what the Fed considers ample reserves. But, I guess, my question to you is, have you heard or seen any other individuals talking about this potential implication of using the discount window, that it could also lead to fewer reserves being held by banks, and that it might push us in the direction that some of these other central banks are going?

Could Increased Use of the Discount Window Cause a Shift in the Fed’s Operating System?

Kelly: Yes. I think, most prominently, Andrew Bailey, head of the BoE, has sort of advanced this view, and if you look at the BoE's QT program thus far, they're starting to see pickup at their standing repo facility. So, not exactly a discount window, but in this case, effectively, the same thing. And they're much more comfortable with that, and they want to see ongoing activity through the standing repo facility that they have, where that's the allocation of reserves into the system.

Kelly: My sense is that the Fed is— while there's internal debate, the Fed is much more of the view that, when the standing repo facility starts getting substantial value, that's the time to think about ending QT, as opposed to, "Okay, this is sort of our smooth transition." They really talk about it like a backstop. So, I don't know that we'll get to, exactly, that world that we're seeing at the ECB and the BoE and others, where they're more comfortable with playing with fire as far as how many reserves are left, and living in that world where there is a little more volatility, but not, obviously, September 2019 repo crisis level. It's a challenge.

Kelly: The other thing to note about the UK is that they are actively implementing a non-bank standing discount window. It's going to start with government collateral, but they'll look to expand it after that. And so, part of the challenge with using something like a standing repo facility or the discount window is, can you get banks to on-lend liquidity? And we found— sometimes, it's liquidity [regulations], but we've found that a challenge can be balance sheet constraints to, really, on-lend liquidity, and that's something that QE, or large balance sheets, doesn't suffer from, to the same degree, right? The central bank is not relying on intermediaries to increase their leverage to get liquidity where it's needed, to the same degree. So, that could be a challenge as well.

Beckworth: Yes, it was interesting in talking with Isabel Schnabel from the ECB, and they also dealt with the issue of a leaky floor, that rates were dropping beneath the bottom level of their corridor, of sorts. And they didn’t set up something like the overnight reverse repo facility. That's what we set up. We set up something to catch that leaky floor, the money market funds. So, rates would go down in that market. They would then go park at the overnight reverse repo facility as a way to definitely lock in the floor of where the Fed wanted its rates to be based on what they thought appropriate monetary policy was.

Beckworth: And I asked her about that. "Why don't you guys have one?" She said, "Money market funds aren't as big of an issue, or deal, in Europe as they are in the US." So, I guess, to some extent, we already have something of a repo facility to deal with this problem, but it could be even more so if we were to, for example, expand the standing repo facility, who has access to it, or who has access to the discount window. And, I guess, the concern is then that the Fed ends up subsidizing these other players or encouraging more shadow market activity that, maybe, doesn't want to go down that path. So, it is really complicated, and it'll be interesting to see where this all ends up.

Kelly: Yes, absolutely. It's a constant challenge of, who do you want to interact with as a central bank? What constitutes an open market operation versus like, "Okay, is this, effectively, just emergency lending to everybody or some sort of engagement that we're not supposed to be in, from a legal perspective," versus something like QE, where it's more plainly, "Okay, I've just gone out on the market and bought a Treasury." All of these things are at play here.

Beckworth: Okay, with that, our time is up. Our guest today has been Steven Kelly. Steven, thank you for coming on the program.

Kelly: Thanks, David.