2024 Financial Markets Conference

Policy Session 1: Domestic Liquidity Provision during Potential Crises

Moderator Steven Kelly, associate director of research at the Yale Program on Financial Stability, led panelists Luc Laeven, European Central Bank director-general of directorate general research, Susan McLaughlin, executive fellow at the Yale Program on Financial Stability, and Bill Nelson, Bank Policy Institute executive vice president and chief economist, through a discussion about the Fed's role as the lender of last resort to the US financial system.

Transcript

Steven Kelly: Good morning. Welcome to the first policy session, "Domestic Liquidity Provision during Potential Crises." 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're in. 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 emergency lending and the Fed's discount window. There's broad directional agreement on technical issues like operational readiness, collateral prepositioning, infrastructure enhancements, and destigmatization—issues that we'll dive deep into during today's panel.

To some extent, consensus on the "what" of reform is what's most important, but for getting all the way to the most effective policies we also need to further our shared understanding of "what for." While we all agree that better central bank liquidity provision leads to better crisis response, there is little agreement on how or how much. For instance, Vice Chair for Supervision Michael Barr, today's keynote speaker—who is very dialed in, as you heard, on improving liquidity provision and regulation—said in December, "The banks that failed because of the stress event that began in March had access to and utilized the Federal Reserve's discount window, and their failures were not the result of a lack of access to the discount window."

Meanwhile, his predecessor both in the vice chair position and as a keynote speaker here, Randy Quarles, said if the Fed had been focused on its fundamental reason for existence, which is to provide liquidity to viable institutions when they're undergoing a run, Silicon Valley Bank would have survived. It seems highly unlikely that the two vice chairs' differing political persuasions can explain their difference regarding what the discount window can accomplish. It thus remains incumbent on us to make, to borrow a phrase, "substantial further progress" in our shared intellectual understanding of the boundaries of the discount window's potential influence.

The discount window's performance last March didn't make anyone particularly happy, with the possible exception of the Federal Home Loan Banks. However, could even a "perfect world" discount window really have saved SVB? Could it have prevented the need for the Fed to create the Bank Term Funding Program? If so, could we ever realistically build consensus around the necessary reforms while offsetting the associated drawbacks?

I'd offer as a starting point that when the market is running on a specific bank's business model, as in the case of the 2022 bank failures, the discount window should not be expected to save the bank. In this case, the window can do little more than buy you time, which is not in itself an unworthy goal.

While these bank failures tend to look like the result of a liquidity issue, that is simply the final manifestation of a viability issue, notwithstanding what their CEOs say in subsequent testimonies. We should abandon the revenant notion that targeted bank runs are simply the result of horoscopic liquidity demands: solvent banks that randomly become illiquid. Rather, such runs are the market's expression of fundamental viability concerns. Customers' confidence in full repayment of deposits, which an effective discount window helps instill, is a necessary but not sufficient incentive to stay with a bank that is nonviable as a going concern.

In addition to at least buying time, however, the discount window should be incredibly effective as a systemic tool, when the system is facing elevated and dispersed demands for reserves. The discount window should function as an elastic and precise allocator of reserves in crisis, whether simply until uncertainty dissipates or as a bridge to more structural policy responses.

As we've started to circle some low-hanging reforms to the discount window, it's important we do not lose sight of an issue that preexisted the 2023 banking crisis but was laid especially bare during it: the challenge of collateral value. Unlike the perhaps prototypical financial crisis, it was interest rate risk that provided the initial vulnerability to the banking system last year rather than credit risk. This poses a particular challenge to central bank liquidity provision, a risk that very clearly manifested in the BTFP's novel feature of valuing collateral at par.

Par valuation allowed banks to fund the book value of government securities at the BTFP, end-running around the securities' mark-to-market losses that followed from the Fed's unexpectedly steep and rapid interest rate hikes. Under normal procedure, central bank lending operations would only consider the fair value of the collateral and impose a haircut. It's often stated that a bank's unrealized mark-to-market losses on assets only become realized when the bank is forced to sell; however, this is only part of the truth.

Mark-to-market losses from interest rate increases represent the lifetime cost of funding the asset at market-based pricing; so, a bank that does hold on to an asset until maturity but has to pay market rates on the funding side (such as at the discount window), will likewise realize those losses. Banks specifically, however, actually can be particularly well situated to avoid these losses implied by mark-to-market valuations, and record things at held to maturity value.

Take a hypothetical bank in the 2023 environment. It's paying a deposit cost of 0 percent, for a security yielding 3 percent, and market rates are 5 percent. This bank has unrealized losses if the asset is marked to the Treasury curve—that is, market-based rates. If the asset were instead marked to the bank's deposit curve, however, it would have an unrealized gain. Thus, a bank that's able to hold on to its cheap deposit funding can safely account for these assets at their held to maturity or par value. This is basically the Bank of America story from the last couple of years.

By contrast, a bank that can hold its deposit funding should be marking to the Treasury curve. This became SVB's reality as it's deposit base eroded, first steadily for a year and then suddenly.

So, what does this have to do with central bank liquidity provision? As a bank replaces lost deposits with central bank funding, which prices at market rates or higher, the central bank is self-selecting for banks that are going to realize those losses. That is, when a bank replaces its zero-cost deposits with market-rate discount window or BTFP funding, the bank begins realizing those unrealized losses.

Credit and liquidity risk, by contrast, are more endogenous to the central bank's emergency lending actions, particularly in the acute phase of the financial crisis. The Fed's financial stability actions, by drawing a line under fire sales containing financial spreads and economic fallout and reducing tail risk, can help credit risk fall as an asset approaches maturity.

Interest rate risk, meanwhile, is endogenous to monetary policy. The central bank's standard adverse selection problem in lending operations is thus potentially even more of a challenge when stress originates in interest rate risk rather than credit risk. Whether a monetary shock, as in 2023, or something like the "dash for cash" that we saw in the 2020 treasury market, or the 2022 gilt market, tools like the Fed's discount window and Standing Repo Facility currently function no differently than a market player, when it comes to collateral marks. They are bound by the fire sale market value of collateral; they do not put a floor on the range of correct prices.

By contrast, lack of a market-based fair value in the case of, say, loan collateral when perceived credit risk has risen, frees the central bank to lend more flexibly, and appropriately so, given the endogeneity to the central bank's intervention of that risk. Indeed, benchmarking to "normal times" is a regular feature of central bank term sheets. Interventions regularly apply this kind of consideration to the interest rate, where any penalty is set relative to what rates would be normally, not what they are at times of stress.

Such consideration is also inherent to choosing the types of collateral eligible for lending operations, a process which is usually based on the soundness of the collateral in normal times. Again, this is the endogeneity of central bank liquidity provision. While the BTFP was framed as a Rubicon crossing, with respect to collateral valuation terms, other terms of central bank intervention regularly consider what is normal market conditions. As recent crises have shown, it pays to look at collateral valuation through a similar lens.

With those initial thoughts, I want to introduce our next speaker, Bill Nelson. Bill is an executive vice president and the chief economist at the Bank Policy Institute, and an adjunct professor at Georgetown. Previously, Bill was a deputy director of the Division of Monetary Affairs of the Fed, where his responsibilities included monetary policy and discount window analysis, as well as financial institution supervision. In 2004, he was the founding chief of the Monetary and Financial Stability section of Monetary Affairs, and in 2008 he helped design and manage several of the Federal Reserve's emergency liquidity facilities. Bill.

Bill Nelson: Thank you, Steven. And thank you, President Bostic and the Atlanta Fed, for the invitation to be here today. I'm really looking forward to this panel's discussions, and very happy to be involved.

The first thing I'm going to do is set the table for the whole group's discussions by talking a bit about some discount window basics and some liquidity regulation basics. Then I'm going to turn to the substance of my remarks, which is to relay some feedback and information that I've received from extensive meetings with bank treasurers over the course of this year discussing these issues, as well as make some suggestions for how liquidity assessments and liquidity regulations could be improved by better incorporating a recognition of discount window borrowing capacity.

I'm going to start with a little history. The Federal Reserve System was created in large part to address the weaknesses in the financial system that led to the very severe crisis, the Panic of 1907. In particular, one of the more significant problems was that the combination of rigid reserve requirements, as well as a finite supply of reserves, meant that banks shut their doors and stopped providing currency or funding to individuals and businesses that came in to get it during runs, even when the bank still had very high supplies of those things, out of concern that they would not meet their requirements.

The Fed was created in part so that banks could discount commercial paper, or borrow against collateral of business loans, to replenish their reserve supply so that they knew that they could replenish it and could continue to provide funding even when demand was very high. In addition, it was recognized that access to that additional source of funding made it possible, made it safe, to lower reserve requirements so that banks could provide more of their balance sheet towards lending to businesses rather than holding reserves. This next slide is several quotes that support what I just said. For the sake of time, I'm going to skip over them, but I do hope you get a chance to read them.

Another bit of history now, more recent history: During the global financial crisis, there was a sea change in how a banking institution's liquidity assessment was made by supervisors. Before the global financial crisis, what the examiners looked for to judge liquidity of a bank was diversified, reliable funding, and diversified, reliable contingency funding, and access to the discount window.

After the global financial crisis, the focus was almost exclusively on having large stockpiles of HQLA [High-Quality Liquid Assets]. I note that prior to the global financial crisis, an asset-based liquidity strategy was considered to be something that a relatively unsophisticated community bank would do.

Two weeks before SVB failed, I wrote a note drawing on a lot of things I've written in which I argued that this shift was a mistake and that we needed to have a holistic review of how liquidity requirements are designed. Two later, SVB failed awash in HQLA but funded with nondiversified unreliable liabilities and unable to borrow from the discount window.

For the sake of time I'm going to skip over this discussion of the laws the Fed uses to lend against, except to note the last bullet, which is that by law the Federal Reserve has to publish the names and identities and terms of borrowers from the discount window—something that I was surprised to learn from treasurers actually really contributes to stigma and that Susan will discuss—and after emergency loans, after one year, so some of the banks that borrowed from the Bank Term Funding Program. I hope they know that. It was a 13(3) facility, so they might be surprised by that disclosure.

There are three types of regular discount window credit. Seasonal credit is provided to very small institutions that have seasonal swings in their funding or loan demand, and I mention it in particular because its purpose is very old school. It's very much aligned with why the discount window was created. The objective is so that these small banks can continue to lend into their community year-round, rather than having to hold large stocks of HQLA to run down and buffer the swings in their deposits or loan demand.

When somebody mentions the discount window or the discount rate, what they're talking about is primary credit. Primary credit is provided to banks that are sound financial institutions, meaning those that are CAMELS 1-, 2-, or 3-rated, and at least adequately capitalized. It's provided on a "no questions asked" basis, at least in theory—that doesn't seem to be the way it's being done now—and at an above-market rate. Secondary credit may be provided to those banks that don't meet the financial soundness criteria, either to help them recover and to become financially sound again or to aid in resolution.

Three years ago, the Federal Reserve System created another lending facility, the Standing Repo Facility. The facility conducts fixed-rate, full-allotment auctions of overnight repos against Treasuries, or agency MBS or agency securities, once each day at 1:30 p.m. The Standing Repo Facility charges the same rate and applies basically the same haircuts as the discount window. If you're a commercial bank, it's actually not that clear what the advantage is of the Standing Repo Facility relative to primary credit, because primary credit is available until 6:59 p.m.

The Fed's objective in creating the Standing Repo Facility was to create something with a different look and feeling from the discount window, hoping to reduce stigma so there was no stigma attached to borrowing from the SRF. It also was intended in part to limit volatility in the repo market but many people are not sure that it would be able to do that because the biggest borrowers in the repo market, hedge funds, are not included as counterparts.

All discount loans are collateralized. The Fed seeks to accept all bankable assets as collateral. It has always done so. The limits on the Fed's ability to accept collateral are that it has to be able to value the collateral and it has to be able to get a perfected security interest in the collateral, meaning that if the bank fails the Fed will know it has priority in seizing the collateral.

To come up with lendable value, the Fed applies conservative haircuts against the market value of the collateral, securities, and loans. It comes up with an estimate of market value. There's been a lot of talk lately about the importance of prepositioning and the value of prepositioning collateral. All discount window collateral has always been prepositioned. Banks maintain pools of collateral at the window to borrow against if they need to. As Vice Chair Barr mentioned, at the end of 2003 there was about $3 trillion in collateral prepositioned at lendable value. BPI's estimates are that 80 percent of that collateral is in the form of loans, and of those loans 40 percent of the collateral is consumer loans and 20 percent is business loans. This reflects what collateral is useful for the Federal Home Loan banks, which is not consumer loans and business loans.

Almost entirely, loan collateral is maintained in what are borrower-in-custody agreements, or BIC agreements. That basically means that the bank doesn't have to provide the physical loan documentation to the Federal Reserve Bank. It maintains those documents itself.

Turning to liquidity requirements. Basel 3—the original Basel 3, not the contentious Basel 3—the reform package put in place after the global financial crisis, created two new, numerical liquidity requirements. The one that most everyone is familiar with is the liquidity coverage ratio. That requires banks to have enough high-quality, liquid assets to be able to meet their funding needs at the end of a 30-day stress episode. HQLA consists, in the United States, mostly of reserve balances. That is to say, deposits at the Federal Reserve Bank, Treasuries, and agency MBS.

HQLA is calculated at market value, and that's true in particular for held-to-maturity securities, even though those are held on the bank's books at par value, as HQLA they're valued at market. The other liquidity requirement is the net stable funding ratio. This requires banks to have sticky liability sufficient to fund their core assets over a one-year period.

These two requirements are not actually the most binding requirements for banks, including the LCR. The most binding requirements are actually two US-specific liquidity requirements: internal liquidity stress tests and the resolution-related requirements. Large banks have to conduct internal liquidity stress tests at the overnight, 30-day, 90-day, and one-year horizons and report the results of those tests to their supervisors at least once a month.

The largest banks must satisfy two liquidity requirements that are part of their resolution plans: the Resolution Liquidity Adequacy and Positioning requirement, or RLAP, and the Resolution Liquidity Execution Need requirement, or RLEN. I'm not going to discuss those further, but I have written about them. At the end of my remarks, I've included a set of writings on this topic that I've done over the last several years.

The table having been set, I'm now going to turn to some feedback that I've gotten from bank treasurers. So far in 2024, I've visited with 17 of our BPI banks' treasurers and their teams to discuss the topics of today's panel: how they manage liquidity risk, their views on liquidity regulations, how they use the Federal Home Loan Banks, and their views of the discount window. These consisted of regional banks, G-SIBs, FBOs, and custody banks.

In terms of dealing with liquidity stress, we've had a couple of great episodes to talk about over the last several years. In 2022, as you all know, there was a large shift of deposits out of the banking industry as the Federal Reserve raised rates sharply and deposits moved into the money fund industry, largely. That wasn't a stress event, but it was a profound event. Banks dealt with that by borrowing from the Federal Home Loan Banks and, to some extent, borrowing from other wholesale sources of funding.

In 2023, many banks needed to face the prospect that they might be subject to a run. To prepare for that episode they borrowed more from their Federal Home Loan Banks and from other sources of funding to build up their cash supplies to make sure that they had them if necessary.

One other thing I would note is that, and this is relevant to the potential limiting of held to maturity securities, there's a perfectly good case to be made for limiting held-to-maturity securities. However, I don't think it should be done through liquidity requirements because held to maturity securities are liquid. They can be repo'd just like available-for-sale securities. Nobody's going to sell a security in a stress event, and they can be used as collateral for the Standing Repo Facility and the discount window.

Generally, it's bad policy to use one type of regulation to solve for another problem. Another thing that I discussed with the banks, because it's of so much importance right now, is deposit stickiness. What deposits are sticky, and which ones aren't, in particular uninsured deposits. What I heard back was that the uninsured deposits of customers that have multiple lines of business with the bank, collateralized deposits, and deposits used by the borrower for their normal business operations are sticky. At the other end of the spectrum, the deposits that flowed into the banking system during the COVID crisis and its aftermath were seen as quite flighty and unreliable.

Another useful point is that several banks told me that if a large depositor was seen at risk of leaving, they might be offered IntraFi, which is a service that breaks up the deposits and makes them entirely insured, or higher deposit rates.

In terms of the discount window, we talked about the discount window and the Federal Home Loan Banks. One of the treasurers grew up in Bulgaria, and she commented that working with the discount window brought up a lot of memories of dealing with institutions in her childhood. [laughter]

The collateral procedures are antiquated, everyone was uniformed, that the discount people at the Federal Reserve Bank would say, "yes, you should use the discount window," but the examiners, including examiners from the Federal Reserve System uniformly said, "no, you should not be using the discount window" ... that it was to be avoided. One treasurer with a long memory indicated that the 2003 revisions that created primary credit and secondary credit were actually working. Fed staff visited banks and encouraged them to incorporate it into their liquidity assessments, met with examiners, wrote an SR letter to encourage examiners to deal with that positively, and over time that was working.

For one thing, Bank of New York Mellon was stepping into the federal funds market when rates were above the primary credit rate, even when they didn't need funding. They would step in and they would borrow and then lend to whoever it was that was paying up. But the Global Financial Crisis, and the recriminations against banks that borrowed, and the treatment of borrowing as bailouts, just supercharged stigma, got rid of all of those gains.

One thing that I heard over and over again that surprised me a bit was that one of the main sources of stigma is that banks are very concerned about how investors would view borrowing. They're concerned about the impact on their stock prices. I'm still not completely sure how they think that that information would get to investors, but it is a real concern.

On the other side of this, in terms of the Federal Home Loan Banks, I'd lose my economist's secret decoder ring if it weren't true that I am very concerned about the moral hazard and financial stability risks presented by the Federal Home Loan Banks. What I learned was that at the current juncture Federal Home Loan Banks are very important for banks' normal, everyday asset liability managements. Their operations are also seen as more user-friendly than the discount window.

In terms of collateral competition, Federal Home Loan Banks only assign lendable value to mortgage-related collateral, but normally if a bank borrows, or if it wants to borrow, the Federal Home Loan Bank takes a blanket lien against all the bank's assets. That of course, competes with the Federal Reserve's need to get a perfected interest. If an individual bank wants to be able to borrow from a Federal Home Loan Bank and from a Federal Reserve Bank, a bespoke collateral sharing agreement is worked out for that bank in which a certain set of assets is carved out for pledging to the Federal Reserve, typically consumer loans and business loans.

A lot of useful feedback in the discussions on the examination process: banks indicated that their internal liquidity stress test, the way they're forced to do the stress tests, are basically divorced with how they would deal with liquidity risks. For one thing, you can't count on using the Federal Home Loan Banks, and as I mentioned that's the primary way that banks currently deal with liquidity risk. I also heard more problematic stories. One treasurer relayed how one of her subordinates had done a lot of work researching how long it would take a municipality to shift its deposit relationship. The answer was about six months. They asked their examiner if they could use that information in their internal liquidity stress test, and the examiner said, "No, we'd actually really rather you use the LCR assumptions of a 25 percent outflow each month." That kind of thing not only prevents the ILSTs from providing some variety in the assessments, but it also is quite dispiriting and makes it less valuable to do your own work and do your homework and figure out your liquidity risks.

One thing I heard consistently from all of the banks is that their examiners have instructed them that they cannot count on using the discount window or the Standing Repo Facility probably, as the means by which they would convert HQLA into cash, into their ILSTs. Vice chair's remarks suggested a different interpretation, and I'm very pleased to hear that there's going to be some clarification on that point going forward.

One treasurer was recently told that he could actually count on using the Standing Repo Facility to monetize his assets, but not the discount window. The reason was the Standing Repo Facility are repos, and repos are allowed. Now he has collateral in three places, at the Federal Home Loan Banks, at the discount window, and at Bank of New York Mellon, which is where the collateral for the Standing Repo Facility is maintained. It takes him about a day to move collateral from one place to the other.

Every bank uniformly said that it's been hammered home to them over and over again by their examiners, including Federal Reserve examiners, that using the discount window is not okay. One treasurer said, "I know it's not okay because I've been told that over and over again over many years."

Lastly, I want to discuss ways that liquidity assessments can be improved by recognizing a bank's capacity to borrow from the Fed. I'm going to talk about two things in terms of internal liquidity stress tests and the LCR. Internally, liquidity stress tests are typically the most binding liquidity requirement for banks. Banks cannot anticipate using the discount window or the Standing Repo Facility to monetize their HQLA. The LCR grants no credit at all to the ability to borrow from the Federal Reserve or the central bank.

Why should this be changed? Here's four reasons why it would make sense to recognize the liquidity value of being able to borrow from a central bank. The first is that it makes the assessments more accurate. When I was in the business of designing regulations, one of the things that I sought was to make them as accurate as possible because that creates good incentives and avoids unintended consequences. We all saw that in March of 2023. It's clearly the case that a bank that is prepared to borrow from the discount window is more liquid than one that is not prepared to borrow from the discount window.

Another thing of value here, and this comes up particularly because everyone was shocked by the speed of outflow rates, and there's consideration of raising those outflow rates, is that by recognizing that the discount window's purpose is to meet those outflows it enables banks to keep lending. Rather than making banks just hold more and more HQLA and become narrow banks, recognizing the discount window allows them to keep lending to businesses and households, pledging those loans as collateral to the discount window and then looking to that capacity to meet their day zero needs.

Another reason is that efforts to convince banks that it is okay to use the discount window—efforts to reduce sigma—simply ring hollow if the regulations basically are designed to prevent them from using the discount window or to make sure that they won't use the discount window. Lastly, it creates an added incentive for banks to be prepared to borrow. This is not an abstract concern. This is a real issue. SVB, about a year, maybe a year and a half before they failed, was interested in signing up for the Standing Repo Facility, but as a relatively small regional bank they didn't use the repo market regularly.

It's expensive to sign up for tri-party repo access. They wanted to sign up so that they could pass their ILSTs, which we now know they were then failing. When they learned they wouldn't be allowed to count that capacity in their ILSTs, they lost interest. We could be standing here right now having a very different discussion if SVB had been able to get liquidity from the Standing Repo Facility and had an orderly failure rather than a disorderly panic.

Let me mention a couple ways this could be done. First, I fully support the idea of adding a new requirement that banks have enough cash and discount window borrowing capacity to meet a run on their deposits, and as the vice chair mentioned this is something that's actively being discussed. I do think it's important that it's not written as a simple fraction of uninsured deposits for a variety of reasons. I wrote a note on that, "10 Pitfalls to Avoid When Designing Any Additional Liquidity Requirements," that I encourage you to read.

One of the most important reasons is that you could end up making the same mistake as rigid reserve requirements that made unusable reserves in a rigid LCR that makes unusable HQLA. If banks meet such a requirement written in that way, then if they borrow they'll fall below the requirement, so it actually could make banks less willing to use the discount window rather than more. Instead, it should be incorporated into a requirement on their contingency plans. Nobody's going to blame a bank for using their contingency plans in a contingency.

One thing that could be done—I find it really unobjectionable; something that we've been arguing for since 2019—is that discount window borrowing capacity could be subtracted from what's called the bank's "maturity mismatch add-on." Let me unpack that. The international standard for the LCR is that a bank has enough HQLA to meet its needs at the end of 30 days; the US standard is that a bank has enough HQLA to meet its requirement throughout the 30 days.

Since banks pay out a lot of money in the first few days and then only get back money over time, for most banks that peak need is higher than the LCR standard, and that difference is called the maturity mismatch add-on. It can be quite substantial for some banks. You could subtract from that need the lendable value of the collateral that the bank has at the discount window, recognizing that the discount window is a perfectly legitimate way to meet that intra-30-day need. Such a change would be Basel compliant with the international standard for the LCR, because the maturity mismatch add-on is a US gold plating of the standard.

Another way which I am hopeful now about is that banks should be able to recognize their capacity to borrow from the discount window or from the Standing Repo Facility. One thing I should note: in their internal liquidity stress tests, banks not only have to have enough HQLA, they have to be able to monetize the HQLA and convert it into cash. That's what SVB was failing. Right now, banks think that they can't point to the discount window or the Standing Repo Facility as the way they would do that. They should be allowed to do so.

One cautionary note I would add here, is that banks shouldn't be granted this recognition if they're not willing to borrow from the discount window. Susan's going to talk about the severe stigma problem associated with it. It's worse than not recognizing it at all, if you recognize it but then in the end banks won't use it and behave as if they're out of liquidity.

If there were to be an international standard, a top-to-bottom, holistic review, I think an important thing to consider as part of that review is committed liquidity facilities, where banks pay a fee for a collateralized line of credit from the Fed. It's something that's in the international standard. It's also something that's in the US final rule, but for time purposes I'm not going to go into it. But again, that's another thing that is provided among these readings at the end. Thank you.

Kelly: Thanks, Bill. Our next panelist, Susan McLaughlin, is a 30-year vet of the New York Fed, with roles in the Markets Group spanning monetary policy, FX [foreign exchange] reserves management, market structure, financial stability, lender-of-last-resort issues, and fiscal agent matters. She was deeply involved in various aspects of the Fed's policy response to the GFC as head of the discount window at the time in New York. She then worked on the post-crisis reforms of tri-party repo and was heavily involved in the emergency lending programs during the pandemic.

Perhaps most impressively, she's joined us now for about nine months at the Yale School of Management's Program on Financial Stability, where we have basically plugged her brain into our servers and are making an LLM, and she can share all her 30 years of experience in the New York Fed. Susan.

Susan McLaughlin: Thanks, Steve. Good morning; I'm very pleased to be here. I appreciate the invitation from President Bostic and the team at the Atlanta Fed. I actually was here about 10 or 11 years ago I think, to talk about tri-party repo, so it's nice to be back.

I'm going to focus my remarks today on the Fed's role as a liquidity provider to banks and what we need to do to ensure the Fed's liquidity tools are effective in mitigating the fallout from bank runs. Back in the day, which I guess means pre-March 2023, bank runs used to occur over a period of days, or even weeks. But last year we saw that a depositor run can now take a bank down in a matter of hours due to the speed of withdrawals that are made possible by digital banking as well as the role that social media can play in accelerating a run.

In this new world of intraday runs, central banks must be agile and effective in providing liquidity to mitigate the risk that a run poses to financial stability. The SVB and Signature failures showed that the discount window is currently neither agile nor effective in that sense. We needed to establish an emergency facility, the Bank Term Funding Program, because banks either would not or could not finance their government securities at the discount window.

To date, discussion of how to make the window more agile and more effective has really focused primarily on ensuring that banks are ready to borrow. This is important, but it's a necessary but not sufficient condition. Banks must not only be ready, but willing to borrow from the window when they need to, ideally before a run materializes when the central bank can be most effective in mitigating run and contagion risk.

Unfortunately, as we know, the discount window is highly stigmatized, and not an effective financial stability tool as a result. We need to take the steps necessary to eliminate stigma in our liquidity tools for healthy banks. I think the Fed is actually trying to take some of these steps now. Vice Chair Barr mentioned some of these, but I think we can still go further.

This panel is billed "Domestic Liquidity Provision during Potential Crises," and I want to note that central bank lending supports financial stability not only in times of crisis but also on normal business days. Every day, central banks provide intraday credit to banks to support payments system functioning, and the purpose of this lending is clear. There's no stigma attached to it.

In the same way, the central bank may on occasion be asked to provide overnight credit to banks to ensure continuity of banking and payments systems. For example, a bank might have an operational error, or might have made an inaccurate cash flow forecast that day that leaves them with an unmet need that they need to meet and it's too late to do so in the market.

A bank might also need funding amid a disruption to market functioning, such as a cyber event or a disaster of the type that we experienced on 9/11. In such cases, the existence of a credible central bank liquidity backstop can ensure that a temporary disruption of funding access doesn't push otherwise solvent banks into insolvency. It mitigates run and contagion risk and stress by assuring depositors and investors that banks have access to a backstop source of funding.

I've just described a couple of cases in which central bank borrowing does not mean that the bank is on its last legs, and yet this is how we have come to regard the discount window in the US. We've really been of two minds about the discount window, pretty much ever since the early days of the Fed. On the one hand, we have the classical Bagehot theory that central banks should lend freely to sound banks against good collateral when they need it to mitigate the risks from bank runs to financial stability. A corollary of that theory is that banks should borrow when they need to, and this is the messaging that we're hearing from Fed policymakers in the aftermath of last year's bank failures. This is very helpful.

On the other hand, we have the long-standing argument that central bank lending creates moral hazard, is bad, and should not be done. I really disagree strongly with this moral hazard argument against collateralized lending to solvent, prudentially regulated financial institutions, and maybe that's something we can talk more about later. As Bill noted, this is the message that many banks still hear from their examiners. It still influences the way that the media covers discount window usage. Look at the stories that get written whenever there's an H.4.1 weekly release from the Fed that shows an uptick in discount window borrowing.

This message is reflected in the bank liquidity regulations developed in the aftermath of the global financial crisis, which don't, as Bill noted, recognize the discount window as a legitimate source of funding and contingency. It's really important that we decide what we're for. We should be clear. I would advocate that we should be clear that the Fed will lend to sound banks when needed in a world where runs happen very quickly. We should do everything we can to eliminate stigma.

Why is stigma a problem? Central bank lending is most effective in stemming risk if it happens well before the bank run materializes, but stigma causes banks to wait until the last minute to borrow. Importantly, stigma actually undermines bank readiness. If banks don't want to use the discount window and they don't intend to, they won't prepare to use it and they won't be ready when they need it.

We saw this last spring, with both SVB and Signature. Both of those banks did not have arrangements to move the collateral that they had parked at their Federal Home Loan Bank to the Fed. For SVB, they didn't understand the operational cut-off times for moving collateral, seemingly a simple thing, available on the public website. But again, they were not thinking in these terms.

Signature hadn't tested discount window usage for five years prior to its failure. In fact, discount window was not even a part of its contingency funding plan. They were heavily reliant in that plan on FHLB lending. Now to be clear, discount window borrowing would not have saved either of these banks, but it could have bought more time for at least cost resolution in both cases. It also could have slowed or stopped the spread of run risks to other regional banks.

How do we reduce stigma? Stigma is a really complicated, multifaceted problem, and it requires a multifaceted approach. I think there are three areas that we need to look at. First, I think we need changes to the Fed's framework and tools for liquidity provision. Second, we need to change supervisory and regulatory practices to recognize central bank borrowing capacity as a legitimate source of contingency funding. And we need to take a hard look at the relationship between FHLB and Fed lending.

In terms of things that the Fed can do, I see four areas of opportunity. First, creating a new framework for liquidity provision that has two very separate and very distinct tools for liquidity provision. One would be a standing facility to provide routine, short-term liquidity to healthy banks on demand. Sounds a little bit like the primary credit program, but it would not be stigmatized, to support the continuous operation of banking and payments systems.

The other tool would be a different, separately administered tool to provide recovery and resolution funding to troubled banks at the Fed's discretion. Sounds a little bit like secondary credit. This lending would be done at the Fed's discretion in coordination with the bank's primary regulator and resolution authority, so not so different from how secondary credit works today.

I think a key issue here is that other countries draw a very bright line in their frameworks between these two tools, but in the US they're combined under a single discount window. I would argue that that has muddied the waters a bit and has probably contributed to stigma.

Secondly, the routine, on-demand lending tool should be priced to minimize stigma. Primary credit was initially priced at 100 basis points over Fed Funds when it was introduced in 2003. With two decades of experience I think we've realized that that's probably too high a spread. Currently, it's priced at 0 basis points over Fed Funds, and that might actually be a lot closer to where it should be. It shouldn't be cheaper than other sources of funding that banks have access to, but if it's too expensive relative to those other funding sources its use will become stigmatized.

Third, I think there's more that can be done to make the process of borrowing from the window easier for banks. Today, banks must apply affirmatively for a Fed account, then take the initiative to execute borrowing documents before they even have access to the discount window. And that's before there's even any collateral pledged. Why not require these steps to be done up front as a part of the bank chartering process?

Today, banks contact their Reserve Bank for a loan, they wait for the staff to check collateral adequacy and get an approval, and only then they receive the funds. Why not automate that process, automate the collateral check for healthy banks, and make borrowing from the standing facility automatic if collateral is sufficient? This would catch up the process a little bit more to the ease that Bill describes, that's associated with the FHLB process. The wait that banks experience seems to signal we're deciding whether to lend to you, and in this sense I think that is another aspect that reinforces stigma.

Today, the process of pledging loan collateral is time consuming and costly. This is really important because loans are about 75-80 percent of all the collateral that's pledged to Reserve Banks today. It's also the main type of asset that smaller banks have to pledge to the window. One regional bank told me that it cost them $10 million to set up a borrower and custody arrangement for their CNI loan portfolio. There must be IT and data management solutions out there that could help us make this process easier and more efficient.

Finally, I'd advocate that the Fed develop and implement a public communications strategy to support a revised framework for liquidity provision that distinguishes much more clearly between routine, on-demand lending to healthy banks against collateral, and emergency lending to troubled banks on the path to recovery or resolution. This is really important because there are long memories out there among bankers, reporters, examiners, investors—anybody who's familiar with the past history of the discount window and were taught for years that borrowing from the Fed is a bad thing to do.

Language also matters. You'll notice that I haven't really said "lender of last resort" in these remarks so far. I think lender of last resort as a term does read to many in the market as help for a failing bank. Maybe we should be speaking in terms of liquidity provision, or liquidity backstop.

I wouldn't call the new standing facility that I'm advocating "primary credit" or "discount window." I think those terms are very toxic, so there's probably some other way of referring to it that would be better. It would also be important to reject this idea of constructive ambiguity when it comes to the standing facility. It really needs to be on demand, so that banks and their regulators have certainty about its availability.

Not for nothing, the Bank of England actually did this kind of overhaul of their framework and communication strategy after Northern Rock and they've been successful, so that is a model to look to.

I've talked about what the Fed can do. There are also opportunities for bank regulators to help, some of which have been alluded to already. Lisa White's recent remarks, acknowledging that banking supervisors should encourage rather than discourage discount window readiness and usage, are a very good first step, but as Bill noted we also need to recalibrate supervisory practices to make borrowing from the standing facility a legitimate source of contingency funding. Why not allow banks to count the standing facility as a source of contingent liquidity in their internal liquidity stress tests, as is done in other jurisdictions?

We could also, rather than simply requiring banks to preposition collateral, why not incentivize them to do so by counting some portion of the assets that they pledge that are not classified as HQLA, such as loans, toward their liquidity coverage ratio? I know some bank regulators don't like this idea, but my rationale here is that these assets can be monetized same day at the standing facility, and it might even be more liquid in that sense than some level 1 or 2 assets that might not have great secondary market depth on a given day.

Instead of encouraging banks to make the standing facility a source of funding in their contingency funding plans and tests, why don't we require that? It was great to hear Vice Chair Barr say that that's actually something that's being considered. Letting banks decide whether or not they're going to be ready doesn't seem optimal for financial stability.

Finally, the Federal Home Loan Bank system, or FHLB system, has a role to play in reducing discount window stigma as well. The FHLBs are cheaper and easier to deal with in normal times than the Fed is, but they're not the lender of last resort in stress. Only the Fed can play that role. We've seen in previous stress episodes that the FHLBs ramp up their lending early in a crisis period and then pull back as the crisis intensifies.

Banks rely on the FHLBs rather than the Fed as a backup funding source. We saw that with Signature. It's really the Fed, though, that needs to take up the slack when the crisis intensifies and the FHLBs pull back. This state of play amplifies stigma in a couple of ways. First, it equates Fed lending with troubled banks, and disincentivizes the use of Fed liquidity by healthy banks in normal times, which creates stigma and undermines readiness.

It's really important that we rationalize the relationship between Fed and FHLB lending. We need an integrated, national framework for seamless moves of collateral between FHLBs and the Reserve Banks. This is complicated but important because most Reserve Bank districts overlap with more than one FHLB district, and also FHLBs operate quite differently from each other with respect to collateral management.

We also need to reduce the economic incentives for reliance on FHLB credit that result from the current dividend structure. I'm going to make a shameless plug here: Steve and I actually coauthored a piece with Andrew Metrick that offers a solution to this problem, so maybe we can talk about that later as well.

In conclusion, we have some work to do to end stigma to make the Fed's tools for liquidity provision more effective and complementary to bank regulatory tools in containing the destabilizing effects of bank runs. It won't be easy. It requires action across multiple agencies, but other countries have done it and we can, too. Thank you.

Kelly: Thanks, Susan. Our final panelist, and our international representation to let us know all the things we're doing wrong here, is Luc Laeven. Luc is the director general of the Directorate General Research at the European Central Bank. He's previously worked at the IMF and World Bank. His research focuses on banking and international finance issues and is widely published in academic journals. He's a professor of finance at Tilburg University, research fellow of the Centre for Economic Policy Research, chair of the European System of Central Banks' heads of research committee, among many other things that you can see in your program. Luc.

Luc Laeven: Good morning. I would like to thank the Atlanta Fed for inviting me to speak on this very experienced panel on such an important topic, and above all in such a beautiful location. Seen from overseas, the collapse of Silicon Valley Bank and Signature Bank last year was a strong reminder of the fault lines in both regulatory and supervisory practices, resulting in the questionable protection of uninsured depositors and record amounts of liquidity provision by the central bank. Steven asked the three of us to comment on last year's experience with the Fed's discount window, the focal point, and to give our views on what improvements could be made that would have led to better outcomes.

I should start with a disclaimer: these are going to be my own views, obviously, and not those of the European Central Bank. Normally speaking, we make a gentleman's agreement not to comment on each other's policies...unless it's at the Fed's conference, so I'm excused. [laughter] I should also say I come in peace, as a friend and not a critic, and I have two benefits. I'm the outsider giving the outside perspective, and I benefit from the benefit of hindsight. Should I err on any of these two, I have the vice chair in my presence to correct me later on.

I will focus on three issues: first, preparedness, by which I mean the operational readiness of the discount window. My dear two panelists have already commented on that. Two, crisis management, by which I refer to how the functionality of the discount window should adjust during crisis times. And three, crisis prevention from both a regulatory and supervisory perspective.

Let's start with the preparedness issue. The first lesson from last year's experience with the discount window is that both the banks and the Fed need to improve the operational readiness of the window. Most banks found themselves ill-prepared, because they had not used the window for years. The application process is burdensome, the process of approval is cumbersome, and all this requires many manual steps.

These operational issues are reinforced by the fact that most of the open market operations in the US take place with only a small number of primary dealers, with most banks making only very limited use of the existing standing facilities during normal times. The ease of access to liquidity therefore needs to be improved by making the application process lighter and the approval process more automatic.

The Fed could also consider taking steps to encourage banks to make more regular use of the window in normal times. A requirement that banks establish and test access to the discount window would make the system more resilient. I understand and I took note this morning from the vice chair that the Fed is currently looking into these possibilities. While it would obviously impose an administrative burden and cost onto the banks, this could be justified by the systemic nature of this type of liquidity crisis.

Obviously, banks also need to do their part to be ready for discount window use and basically take out the insurance. They need to establish access to the discount window already in normal times, familiarize themselves with the rules, set up collateral arrangements, and test the borrowing procedures. In short, every bank should be ready to fully access the discount window when the need arises. A lender of last resort should not be like your house keys, put away mindlessly, unable to find in case of need. That's not convenient when there is a fire.

Let's turn to crisis management, number two. The second lesson relates to the functionality of the discount window during crisis times. When a crisis strikes, the lender of last resort needs to respond to liquidity needs with great speed and flexibility, to make sure that liquidity can reach those parts of the system that are still solvent, and to prevent liquidity problems from turning into solvency problems. Operational procedures, lengthy approval processes, or inability to mobilize collateral should not get in the way of obtaining emergency liquidity support.

This requires a flexible approach whereby the Fed is highly responsive to liquidity needs throughout the system, adjusting operational parameters of its borrowing facilities as needed, including by complementing its existing facilities with new facilities that address specific liquidity needs. Section 13(3) of the Federal Reserve Act grants quite broad powers to the Fed to do so upon approval of the US Treasury, but only in case of systemic events. Last year, the Fed used these powers to create the Bank Term Funding Program and provide emergency liquidity support with loans up to one year in maturity to insure depository institutions.

At the same time, the Fed adjusted margin requirements of the discount window for high-quality collateral, such that banks could borrow up to the full value of such collateral in line with the BTFP. By offering loans up to one year, as opposed to the standard 90 days under the window, the BTFP addressed longer-term funding needs by banks and mitigated concerns of a prolonged liquidity squeeze in the market. According to Bagehot's doctrine, such emergency liquidity support should be temporary in nature, be offered against high-quality collateral, and a penalty rate.

This is not exactly what the Fed did. Compared to other discount window facilities, the BTFP was indeed stricter in terms of eligible collateral, allowing only securities eligible for purchase by the Fed and open market operations. However, the BTFP provided loans of more attractive collateral valuation, at par value and more attractive interest rates than the standard discount window facilities.

Moreover, one could question whether the duration of the program was consistent with the temporary nature of emergency liquidity support. The BTFP provided longer-term funds on highly attractive terms, and much of its take-up took place in the second half of the program, when liquidity stress had already ebbed, reaching a peak and total amount borrowed of $165 billion in January of this year, long after the market stress in March 2023. Although the Fed did raise the interest rate on new borrowing from the BTFP in January this year, one could ask whether the terms and conditions should not have been adjusted and made less favorable earlier, or whether actions should not have been taken to encourage early repayment.

The favorable BTFP terms may have raised more hazard by reducing the incentives for banks to manage their interest rate risk. Of course, such an assessment needs to consider the usual trade-offs between exempted regulation on the one hand and efficiency of financial system on the other hand. Perhaps the more important question here is whether the establishment of the BTFP could have been avoided, had the standard discount window been more operational. Ultimately, the goal should be to have a resilient, operational framework for the discount window that does not require the establishment of new facilities each time there is a market tension.

Third and finally, crisis preparedness. A third lesson relates to how regulation and supervision interacts with banks' access to the discount window. Liquidity regulations, such as the LCR, are the first line of defense against liquidity risk. They discourage banks from taking on positions with excessive liquidity risk. While SVB was subject to liquidity regulations, SVB did not meet more stringent LCR standards that would have been applicable had it been designated as systemically important. Of course, when it failed SVB was deemed systemically important and the systemic risk exception was used to inject record high central bank liquidity into the system.

However, regulation should be consistent over the cycle. Institutions that are likely to be deemed systemically important during crisis should normally be designated as such before a crisis erupts, even if it's not always straightforward to identify it, and which institution will be systemic during a crisis. Had this been the case for SVB, it would have been subject to tighter liquidity and capital regulations, and more intrusive supervision. Even if this would not have prevented SVB from running into trouble, it is likely that problems would have been detected earlier and time would have been bought. In any case, it would have resulted in larger buffers to absorb the shock.

More generally, a stricter enforcement of existing supervision and regulation is needed to contain liquidity risks in the first place to ensure that banks maintain sufficient liquidity, adequately manage their interest rate risk, and do not fund themselves primarily with runnable liquidities. At the time of SVB's failure, there was a large concentration of uninsured deposits in the system. SVB itself had about 90 percent of its deposit uninsured.

Liquidity regulation should discourage banks from funding themselves in this way, for instance, through net stable funding ratios. Supervisors should take immediate action when interest rate risk is not managed properly. Protecting uninsured deposits in full raises serious concerns about moral hazards and risks violating the FDIC statutory requirement to use the least-cost resolution method that protects insured depositors.

Given the sizable amounts of large and uninsured deposits in the banking system that are highly runnable, we should ask ourselves whether it is desirable to have a system where the protection of uninsured deposits seems to have become the norm. Some have argued that we should instead require banks to preposition collateral with the window to reduce the risks that a bank cannot borrow when it runs into trouble because collateral is tied up elsewhere. Today, as the vice chair already mentioned, banks already preposition sizable amounts of collateral at the window. The question therefore is not if, but whether more is needed.

While prepositioning collateral at the discount window against runnable liabilities helps from a readiness perspective, it is not without cost for the banks. Any requirement to preposition collateral with the Fed should therefore strike a balance between these benefits and costs and also consider the alternatives of adjusting liquidity regulation. The root cause of last year's crisis was that banks were allowed to fund themselves with large amounts of runnable liabilities. This is what we should focus on. This is what should be discouraged through regulation and supervision in the first place.

To sum up: I have argued that, one, the operational readiness of the discount window needs to be improved so that every bank can access it when the need arises. Two, that emergency lending should be provided rapidly and in a flexible manner during crisis times, but that such emergency provisions should have a temporary nature and be on terms that are not more generous than strictly needed to end the financial panic and prevent failures of solvent banks. And three, that the implementation of regulations and supervision needs to be strengthened to prevent such financial panics in the first place. Thank you.

Kelly: Thanks, Luc. A reminder to submit any questions into my iPad here. Let's start and talk about the vice chair's comments today, specifically prepositioning—what Luc just mentioned at the end—at the discount window. He mentioned the Fed is considering prepositioning against uninsured deposits. The G30 report from earlier this year mentioned 100 percent of runnable liabilities. We've heard from the Wall Street Journal that the Fed is considering somewhere around 40 percent of uninsured deposits. Do we think this is directionally correct? Is it not enough? Is it too much?

Bill, maybe start with you. You talked about how regulatory thresholds can sometimes become floors. I don't know if you want to start on this policy.

Nelson: Sure. First, I would say I'm very supportive of the thrust of the idea of making sure that banks are able and ready to borrow from the discount window, and that they're adequately prepared to borrow from the discount window. Nevertheless, such a new requirement needs to be instituted with a lot of care and thought to think about unintended consequences. For one thing, as I mentioned, not all uninsured deposits are the same. Treating them all as if they are the same is going to make it unnecessarily difficult for some certain banks, such as custody banks, that have a stable business model but are funded largely with uninsured deposits.

As I said, I'm also very concerned about forming just a simple ratio as the nature of the requirement. It should be a more sophisticated estimate of what their funding needs might be, and it should be built into their contingency funding plans. Nobody's going to blame a bank for using their contingency funding plans.

If you just say, "You have to have enough borrowing capacity that's over some threshold," then it's just like rigid reserve requirements; it's just like the LCR. It's actually going to make the ability of the institution to use that resource less rather than more, because if you are meeting the threshold and then you use it, then you no longer meet the threshold. It's the classic "last taxi at the train station" problem.

Kelly: Susan or Luc?

McLaughlin: I think directionally it's absolutely the right move. As I noted in my remarks, I would just like to see us incentivize pledging of collateral that's not HQLA, which has maybe fewer higher invest uses in the market, even if it's just a small fraction of that that's counted toward a ratio requirement.

Laeven: I guess I'm going to repeat myself a bit, but I think we should focus more first on the root cause of the problem which I believe was the large amounts of runnable liabilities in the system. That should be getting much more attention from regulation and supervision. And then what is left of it? Indeed, the discount window needs to be ready, but it's not the case that seems to be the flavor of many of these proposals, that banks don't preposition a lot of collateral. They already do. They by and large meet the guidance that the Fed is working on. I don't think it makes sense to go all the way to 100 percent. It's very costly for banks. The direction the Fed is going to is the right one, and the focus is not on the 40 percent. The focus is for banks to be ready, and that's the essence.

Kelly: Some have suggested that the prepositioning requirement actually kind of solves for this idea of how a bank is funding itself. If you require prepositioning against certain runnable liabilities, it affects how a bank thinks about its funding mix. Are you thinking this is not the way to go about it, and the Fed and other regulators should be more direct? Or do you not buy that argument that that'll have enough effect at the margin?

Laeven: So again, the market failure here is much more on the liability side of the banks. If I may take issue with one of the points of the panelists, I think Bill was proposing that you should basically get some sort of discount for being prepared to borrow from the discount window, in your LCR or other type of regulation. I don't think that's the right way to go. These regulations serve a very clear purpose. If a bank gets into liquidity issues beyond that first line of defense, the issue is, at the moment we seem to have a market failure there: that banks are either unwilling or unable to make use of it.

The second part very much is the Fed's job: it should be possible for banks to tap into the discount window without all these delays and complications of posting collateral. That's, to me, more important than focusing on the stigma because the stigma is ultimately very much related to the confusion that seems to be there between the primary and the secondary part of the window. Maybe that's more of an educational issue. It would indeed be useful, but Susan was advocating to separate the windows, maybe in name somehow.

In most other countries, there's no stigma, so there's some soul searching needed. To me, it's fixing liquidity regulations first, and not mixing them up with this discount window. To me, these are two very separate tools and they should stay separate.

Kelly: Bill, do you want to respond?

Nelson: Yes, I'd say a few things. First of all, what I proposed was that the gold plating that the US does of the LCR be reduced to the extent that an institution has collateral pledged at the window. That would only bring it down to being similar to the LCR as it is implemented in Europe, so it's hard to see how you could object to that unless you think that the European LCR, as implemented now, is incorrectly being implemented.

On this general point, when you're designing a regulation, I think the first thing that you need to do is to think about what I'm trying to achieve with this regulation. There's a lot of confusion out there right now between using discount window capacity in a manner similar to what Mervyn King has proposed as a replacement for deposit insurance.

It's true. If every liability that a bank has is backed by the capacity to borrow from the central bank, you don't need deposit insurance, you don't need capital requirements, you don't need liquidity requirements. That's what Mervyn King has recommended. I think of the G30 report as being—and he was on that team—as being sort of a hybrid of that, focusing just on uninsured deposits, which again is sort of fighting the last war. It wasn't that long ago we were all worried about the repo market because that what caused the Global Financial Crisis. That's a very different objective from what liquidity requirements are typically aimed at. They're aimed at making sure that solvent banks are able to withstand a liquidity stress episode. You might think of what were the liquidity needs of the Pac West or the other banks that were conceivably going to be run on, rather than SVB, whose capital was zero in a mark-to-market basis. That's not going to be solved by a liquidity requirement. I think it's very important to keep those two things separate.

Kelly: Shifting into the stigma discussion, Susan, you talked a lot about the importance of Fed and other regulator communication. Short of rebranding the facility the way you propose what more do you think is needed—Vice Chair Barr said this morning, "every time I give a speech, I'm talking about how great the discount window is," and Bill's remarks said at ground level maybe that's not the case—what more is needed besides Jay Powell giving speeches on the communication front around the discount window?

McLaughlin: I think it's a multifaceted issue. Communication is great, but the treatment of borrowing capacity in liquidity regulations, and also the relationship of Fed lending to FHLB lending, are also really important. I think you have to address all three of those aspects.

Kelly: Since you mentioned the FHLBs—you broke the seal—let's talk about the FHLBs a little bit. They put out a report late last year on reforming the FHLB relationship, with emergency lending specifically. They talk a lot about, "Let's be in a position to transfer collateral to the Fed; let's work well with the Fed." What we see in crisis is, FHLBs are called the lender of next-to-last resort, so banks go to FHLBs. Then when FHLBs stop playing along, they go to the Fed.

You were at the discount window in 2007, 2008. What was your experience like with the FHLB of New York? Is the FHFA report around communicating between the Fed, the regional Feds, and the regional FHLBs... does that feel helpful to you? Does that feel like it would be helpful?

McLaughlin: The FHFA report was actually very helpful in a number of ways, including with communication. I will say, in three years of running the discount window between 2006 and 2008, I had exactly one conversation with the Federal Home Loan Bank of New York. I think it took us two days to find the right person. That's obviously problematic. These systems are interrelated in the sense that they are potentially both sources of contingency liquidity for banks, so the interoperability we've seen is just really challenged.

I remember trying to get information because I also was responsible for collateral valuation for the system in that role. It was very difficult to get information on what the haircut schedules were at the FHLBs, and they also differ by Federal Home Loan Bank, whereas the Fed has one schedule and everyone manages to that. So yes, absolutely. Anything that can be done to improve interoperability and communication in normal times so that those relationships and the arrangements to move collateral are there in stress would be extremely helpful.

Kelly: Another issue with the FHLBs is pricing. Generally, several of your remarks talked about how pricing contributes to stigma. There are some good questions around pricing at the BTFP about haircuts. Susan, you talked about how even the "penalty rate" at the discount window is now effectively zero, still much more expensive than deposits. Is there more progress to be had on pricing? Does the BTFP tell us anything about pricing? What can we do on that front to reduce stigma, or is it all going to be other channels? That's for everybody.

Laeven: This is more of an outsider's perspective, but it seems to me this issue of stigma is a bit overdone. If I just look at the comparable facilities of the ECB last year during the same period of market stress, we had sort of the same amount of counterparties, ballpark figures, actually much less borrowing. The discount window was used. Larger number of counterparties in the end, with all the hiccups associated with it. The issue to me is more, why was this emergency facility needed? I also think it was unrealistic to think, as some have said, that any of these banks would not have failed anyway. We're talking about stigma, but I have yet to see a good analysis of the root cause of this and how severe it is.

McLaughlin: I'm going to take a crack at responding to that. I think that stigma is a very real issue in the US. My view is that the BTFP was needed because the discount window was, as you noted, not flexible enough, not able enough, to modify operational parameters. You could have done a term lending program through the discount window against government securities. I think the statutory limit on the term is four months. The current limit that the board has established is 90 days. You could imagine doing a program where you're lending 90 days with, maybe it's X number of opportunities to roll over, or the program is in effect until Y date. You could have done that with the discount window had the stigma not been an impediment to banks coming to use it.

Nelson: Let me make two comments, one on pricing and one on the need for the Bank Term Funding Program. Starting with pricing, I have to say that this was actually the only thing that I disagreed with out of everything that Susan said. The primary credit program was created to replace the adjustment credit program. The adjustment credit program was also created to reduce stigma by giving banks a clear borrowing privilege. For historical reasons that are an interesting story (but I won't tell now), the discount rate ended up being below the federal funds rate and it typically was about 50 basis points below the federal funds rate.

Because it was below the federal funds rate, there were all these rules associated with borrowing. You had to go out in the market first and shop around your need, which contributed to the concern that you'd be seen. You had to ask for it, and you only had so much that you could use, and it just made it very confusing. Our belief at the Federal Reserve when we created the program in 2003 was, it was very important to make it a "no questions asked" facility, ideally one that would be a guaranteed source of funding.

To achieve that, we concluded it needed two things: it needed financial soundness criteria, although, as I noted, actually quite a slack one, and it needed a significant spread. We came up with that spread by going out and talking to banks, and figuring out what the alternative cost of funding was that would cover the whole banking system. Remember, there's thousands of banks, and there's thousands and thousands and thousands of credit unions, all of which can borrow from the Federal Reserve, and the Federal Reserve did not want to be an ongoing source of funding.

It needed to have a spread. That spread alone would lead all these institutions to choose to use it when it was appropriate, but not to use it as an ongoing source of funding. It's simply not politically feasible to charge a higher rate to smaller banks, so that's just not a course that could be taken. I think that the Fed's got a problem right now, basically with the discount rate of like 12 basis points over the federal funds rate. Over time, if they stick to that, a lot of the smaller banks will say, "Hey, well, I'll hit the links and just let my liquidity problems be solved by the discount window," and they'll become an ongoing source of funding to those institutions.

Stigma is not something that each and every institution has as a problem, but you only need a certain number of very small institutions to start borrowing regularly when you then have to start once again putting in rules about use, and that's what we were trying to get away from, because that's what we concluded caused stigma before.

The BTFP. I'm not a fan of the BTFP. I mean, I recognize that a lot of work went into it; they were working under stress, and it was important that they did it. It really violates all the Bagehot principles, basically, and I won't go into that. I think the BTFP would not have been necessary, nor would invoking the systemic risk exception have been necessary, if banks had had enough collateral at the window.

It looked to me like the Fed looked around on that weekend and they said, "When we think about the banks that could be run on, and when we look at their cash resources and discount window capacity, a lot of them are going to have to shut their doors, and then we're going to have a serious banking panic." If all of those institutions had had collateral, that wouldn't have been necessary, and we could be sitting here celebrating the first safe bank failure, which is what we all were looking for, instead of basically looking at the failure of the post-GFC regulatory regime.

Kelly: Vice Chair Barr talked today about the intermingling of liquidity capital resolution. What does the discount window need to be in a better place, if we're just thinking about, "Okay, a bank is going to fail." He mentioned solvent banks. There are challenges written into the law. It doesn't explicitly say it must not be failing, but there are challenges in the law about who the Fed can lend to in that respect. If the Feds collateralize, there's a certain trade-off there. How do we get the window in the right place for resolution?

McLaughlin: The hardest thing about liquidity provision is making that distinction in the moment between solvent and insolvent. We, like many other countries, rely on the central bank to use the bank supervision intelligence that they have to make those judgments. but that's also very challenging for supervisors because exams are not happening on a continuous basis. Solvency can actually shift very, very quickly. I do think that that's something that I have yet to grapple with and figure out in my own mind—like, how do you deal with that?—because I don't think reliance on CAMELs alone really works. You also want to be careful not to rely too heavily on market-based indicators because that can be very pro-cyclical. I guess I'm answering your question to say that I don't have an answer, but that I think that that is something that still needs to be addressed.

Kelly: Just to jump in before Bill here; this is sort of related. Laurie Logan has raised this idea of just purely collateral-based lending. You're talking about determining whether a bank is solvent or not, but you can think about it as "Do they have enough collateral? Do we care if they're solvent, as far as the discount window is concerned?" That's another issue and contributes to some of the stigma at the window, which is you have to wait for approvals to come through and it's not as automated.

McLaughlin: Unfortunately, I think we do care, because we have FDICIA. There are actually legislative requirements or prohibitions on lending past a certain level of two banks that are not well capitalized—and you know more about this than anybody, probably, Bill—but that's a real reason that we do have to think about lending differently with solvent and insolvent institutions, at least in the current legal framework.

Kelly: Bill, do you want to...

Nelson: Sure. First of all, I have to confess that one of the reasons why I didn't discuss resolution liquidity requirements is because, as you know, they are probably, as Susan said, the most complicated to consider. There's two: there's RLAP, and there's RLEN. RLAP is actually not meant to be sized using a failure scenario. You're supposed to use one of your stress test scenarios. In my judgment, if you are sufficiently solid now that in your stress test scenario you remain qualified for primary credit, then it seems appropriate for your RLAP to recognize that capacity to use primary credit. I note that that's a higher standard than being qualified. You have to be super qualified because you need to remain qualified in your stress scenario under which you might have a three-notch down group.

RLEN is meant to deal with liquidity in the event of failure, and that's much more complicated. One of the dirty little secrets about managing and operating the discount window for 20 years is that every bank is systemic in its own market, and lending to a bank to get it to the weekend to have an orderly failure is a very traditional function of the discount window. Even though that's not consistent with the lender of last resort rules, lending to insolvent banks is an important part of preserving franchise value and bringing about an orderly resolution. Recognition along those lines seems appropriate, but it is a tough issue.

Kelly: All right; we are at time. I'm going to give one last question to Luc. It's been coming in from the audience as well. NBFIs: the UK is basically building a standing facility for NBFIs. Should the Fed think about something similar? How realistic is that? We see a lot of challenges with getting banks to onlend in crisis; we find that NBFIs are helpful. What are your thoughts on that?

Laeven: Okay; I'll keep it brief. I think the Fed has these powers under 13(3). The usual issue is that if these NBFIs are less slightly regulated we have an issue of level playing field with the banks. Given the topic we are discussing in this panel the more important question to ask is if you already have all these problems associated with the discount window access for banks, I don't even want to think about the range of problems we're going to have with these nonbanks. Let's first fix things for the banks before we even...let's keep it at that.

Kelly: All right. Thanks to all the panelists.