Steven Kelly on the Silicon Valley Bank Collapse and Its Implications for Financial Policy

Steven Kelly is a senior research associate at the Yale Program on Financial Stability and is a previous guest of the podcast. Steven rejoins Macro Musings to talk about the recent bank collapses at Silicon Valley Bank (SVB) and Signature, the government response, and what this means for financial stability policy in the present and future. David and Steven also discuss the role that interest rate risk and macro policy played in SVB’s failure, the debate over the systemic nature of this crisis, the implementation and use of the Bank Term Funding Program, and more.

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 show.

Steven Kelly: Great to be here, David. I wish it was under better circumstances.

Beckworth: Absolutely. But I can't think of anyone better than you to have walk us through what has happened and what's going to happen potentially going forward. This is kind of your specialty, this is where you have been. In fact, you cut your teeth in 2008 on the previous banking crisis, financial crisis. Now this may be too strong to say this is a financial crisis. It's a limited one, but it potentially could have been had the Fed not stepped in and we'll discuss that later, what was the potential outcome had there been no intervention? Could we have had another Lehman moment or not? But I'm looking forward to discussing this with you because you're very active on Twitter. I know you've been everywhere. You've been cited in the Times, you're doing other interviews, so I appreciate you taking out time. And also, this will be a quick turnaround.

Beckworth: We're recording this on Wednesday, it'll be out Monday. So this is the week after the big collapse of Silicon Valley Bank, Signature Bank. Also, there's the Silvergate Capital shutdown, its institution. And there was pressure on a number of banks. Stock prices went down for a while, so there was real concern. And people like you and others have been really helpful to help us understand what's been going on. So let me just quickly summarize what happened and then we'll jump into the issues at hand here and you can correct me if I misstate anything here. But we basically had three institutions shut down, Silvergate Capital and that was its own decision. They liquidated the bank part of the operation. That was March 8th. And then Silicon Valley Bank was March 10, and that was shut down by state regulators. They literally were insolvent. And then state regulators also shut down Signature Bank in New York and Silvergate and Signature Bank were both tied to the crypto industry. Is that right? They're big crypto-

Kelly: Correct.

Beckworth: So this has been a big blow to the crypto bros and people in that industry, whereas Silicon Valley Bank was closely tied to Silicon Valley and venture capital so that was a blow to them. And what do we know about them? And let's focus on Silicon Valley Bank, Steven, because that's kind of the ground central, the key point there. And I think a lot of the interesting issues emerged from that particular institution, so just over $200 billion in assets, second largest bank failure at the time, the 16th largest bank. It was a large regional bank and they had massive withdrawals on Thursday before they shut down, $42 billion if I have my numbers right here. And they faced an issue of not having sufficient assets, at least mark-to-market value to cover all the withdrawals.

Beckworth: There was some intense rumor mills, social media stuff that really fueled the run. And that kind of gets us up to the point where the state regulator gets in and then the FDIC takes over. And also one last fact, only 14 days before this happened, the auditor of the bank, KPMG, gave it a clean audit. So of course, we'll talk about that. We'll also talk about the Federal Reserve as its main regulator, where was it? But let's stop there and let's look at the bank. So walk us through, what were the big issues at the bank that led to this crisis?

Silicon Valley Bank: Setting the Scene for the Crisis

Kelly: So I think the biggest thing is really the business model of Silicon Valley Bank, which is right there in the name. Their value prop was, "We are going to bank tech and we're going to do so in a way that the large banks maybe can't or won't," which on its face is not a bad business model. If big banks are failing to understand an industry, to some degree it makes sense to start a bank that serves that industry. But the Fed is on a huge tightening campaign. So we've seen frothier industries sort of get washed out in interest rate hikes. The first to go was crypto, the frothiest of the froth. And we've seen other industries come under pressure as well, tech being a big one, tends to have high-flying valuations in a low interest rate environment, thinly capitalized mortgage lenders, stuff that's really thinly capitalized finance, so Carvana, these sorts of firms have been getting washed out.

Kelly: And when you bank an industry that's highly cyclical, you're going to be the first bank to go as financial conditions tighten. So the Fed has very explicitly been on a crusade to tighten financial conditions and banks are a big part of the financial sector. And so it's not wholly unsurprising that they got washed up in this, but basically with Silicon Valley Bank, their depositors were venture capital folks, people in Silicon Valley, and they've just been running out of cash as the economic environment has gotten harder and it's been harder to raise funds. So that runs down SVB's deposits and they were forced into some asset sales and then basically revealing those asset sales to the market when they tried to raise capital set off a run on the business. So it sort of made the cyclicality of the business model plain when they revealed those losses. I mean, those losses are throughout the system. There's been a lot of discussion about these unrealized losses and the accounting, and I'm sure we'll get into all of that, but this was largely a known issue. The FDIC two weeks ago released a report that had a nice chart showing unrealized losses throughout the whole system. And we didn't have a run on the whole system, right? We had a run on SVB because they had a business model that doesn't look viable into the intermediate future.

Beckworth: So is it safe to say that SVB was truly a case of insolvency versus, say, a liquidity crisis?

Kelly: Yeah. And I'm going to qualify that because-

Beckworth: Okay.

Kelly: And we can get into this too, but too often, and I'm sure we'll start to hear from people involved in SVB, the solvency question comes down to, "Well, look, our assets really were worth more than our liabilities," but that doesn't make you a bank. If your balance sheet no longer is viable, you are insolvent. If you can't get counterparties to do business with you as a bank... that's what your service is. You can think about Covid as sort of the perfect example. So we start lockdown at the beginning of Covid and businesses are shut and there's serious worry about a banking crisis and banks, their product is balance sheets. So the Fed rolls out about 95 emergency facilities to protect the banking system at the beginning of Covid, and there was a run that started. But what about every other kind of business? Well, we said, "Hey, make sure to order takeout from your favorite restaurant." We didn't say, "Hey, run from your favorite restaurant because they might go insolvent." But a bank's business is their balance sheet. So if they can't get anyone to do business with them, you're out of business.

Beckworth: I'm glad you mentioned Covid because that was one of the big issues related to this particular bank, that during Covid the tech industry really exploded. We're working from home, we're depending more on our computers, the internet, and this is the bank that serviced them. And so they also had a rapid increase in the deposit base. So they rapidly grew their deposits and most of the deposits were uninsured. Now, I guess, is that typical or is that just unique to this particular business model?

Kelly: That was very unique to SVB. Over 90% of their deposits were uninsured. And we can get into this too. There's probably a larger conversation to be had about exactly what's insured and not insured because we know that in practice, a lot of stuff gets guaranteed in a crisis. The deposit insurance limits drive a lot of macro things that we see as sort of inefficient or risky… I mean the demand for money market funds for instance, and the demand side of all the short term financing is partially driven by shortages of deposit insurance. So there's probably a bigger discussion to be had there.

Beckworth: So they had an unusual deposit base, largely uninsured. It raises the questions, in my mind, for example, were these depositors not cognizant of the fact that they were putting a lot of money into a bank where they could lose funds? Now of course, they're being fully redeemed and they're getting their funds back but I guess the question is, is there any discipline mechanism for depositors? I mean, clearly there is for shareholders. They can lose their wealth, banks can shut down, but I have no incentive myself to look at my bank's balance sheet. I'm well under $250,000. I know I'm good, but maybe I should. I don't know. Is that a feature or a flaw?

Kelly: Yeah, I mean it sort of depends how you want to operate. If you want to say, "Look, depositors might get bailed in and might have to rescue a bank, they might have losses," then I think we can go forward with the current system, but you have to stop rescuing it. It's sort of one or the other. If you really want uninsured depositors to be monitors, you can't keep rescuing them. I don't think that's particularly practical or useful to have every corporation that has more than $250,000 being a bank analyst and having to dig into the 10Ks. So there's probably a lot of room here to move. Whether it's realistic or not, I don't know that it's socially useful to have every small business with a payroll account having to do analysis about their bank.

Beckworth: There's a lot of expectations already on small businesses. This would just add another one on top of that.

Kelly: Right, right.

Beckworth: So again, going back to Silicon Valley Bank, they had an unusual liability side of their balance sheet, so a lot of uninsured deposits. The asset side was also unusual. As you mentioned, they had a lot of long-term bonds, mortgage-backed securities as well as treasuries. And they got burned because they expected rates to stay low, I guess. And that's a question… and they also didn't hedge... So two things, I guess, the expectations that rates would come back down or stay low, and then secondly, they didn't hedge against the possibility that the rates would stay up. So they didn't have insurance against interest rate risk. So let's do that last point first. Don't most banks think about this? Aren't they supposed to be getting to hedge against interest rate risk and why didn't SVB do it?

Interest Rate Risk and the Role of Macro Policy

Kelly: Yeah, I think we don't know as much as we'd like to, and maybe some of the Fed review will shed some light on this, but it seems from what little we know, they were previously hedged and basically got rid of them. They wanted to ride, basically, a wave of downward rates. And that obviously wasn't what came to fruition. As recently as six days ago, we were talking about the Fed going to 6%. I don't know how much that's realistically on the table now, but they just got burned. I mean, it's stupid but the other thing to note about this, and this is why crises get sort of misdiagnosed or mistreated after the fact, is there's only big investigations into bank mistakes in a crisis. When you do a big investigation, you find idiocy amongst managers, you find bad incentives, you find fraud, you find control shortcomings but those things exist throughout the cycle.

Kelly: We haven't had a bank failure since 2020. I guarantee there was just as much idiocy in 2021 and 2022 as there was today. So this is where sort of the macro picture and the bank structure picture and whether or not there's enough FDIC insurance and all these things sort of get lost because you're right, the story of, "Oh my god, this CEO took off these hedges because they wanted to make an extra million bucks," which, yeah, it's stupid, but that's not how you solve systemic crises. So we'll see how Congress diagnoses this, but I'm not super confident.

Beckworth: Okay, let's go to the other point, the expectation that the bank had that rates would remain low. And here's where we get into discussions and I would say a debate. I've gotten into debates on Twitter about this. To what extent did macro policy play a role in this versus bad decisions at the bank? And I would argue both of those played a role. Obviously, we've talked about the bad decisions at the bank, but to me, it's understandable that many investors, not just banks holding bonds, but many investors expected rates to stay low for a long time and at least come back down, that, "Okay, we'll beat inflation, they'll come back down," and there's good reasons. We've lived through a decade of low interest rates. And I was looking back at the FOMC's own interest rate projections in that Summary of Economic Projections, SEP, back in March of 2022.

Beckworth: And they had it barely getting to 2% in 2022, a little above 3% this year. I mean, the Fed itself was surprised by what it has done. And in fact, the Fed itself is suffering mark-to-market losses. Doesn't really matter for them, but everyone is surprised and everyone is bearing some of this cost or loss from bonds losing their value. So I guess my question is this, had there not been these high rate hikes, maybe SVB could have survived its risky model.

Kelly: Sure.

Beckworth: But it takes two to tango here so I don't want to put the blame all on the Fed's macro policy, but any thoughts that you have on that?

Kelly: Yeah, this is another case where sort of the left hand and the right hand of the Fed are a little bit in disagreement. The Fed is in a weird place because like I said, they've been explicit of, "Our policy works through financial conditions, so we're going to talk up financial conditions, we're going to tighten things, we're going to raise rates." And so when this started, it was like, oh, the S&P fell and credit spreads went up in bond markets. And everyone's like, "Oh my God, the Fed's doing a great job." And the Fed is like, "Wow, we're doing a great job. Employment is still rising, unemployment's going down, and we're tightening financial conditions." But the Fed is also responsible for the safety and soundness of the banking system. And it's sort of a weird thing to say for the Fed to be like, "I'm going to go eviscerate financial conditions and you think there's a Fed put, but we're going to prove to you there's not."

Kelly: But that also applies to banks. They are a huge section of the financial sector. So all of a sudden, we get some bank blowups, and now everyone's pointing fingers at Fed officials, saying, "Where were you on this?" And now they're doing a review of supervision. And there were probably some shortcomings there, but also it's almost a meme. I mean, it was them doing this. They explicitly said, "There's excess finance out there, let's get rid of it." And now we have a bank that's banking tech and VC and speculative stuff, and it goes away and now we're going, "Oh my God, they went too far." I mean, there's just something about banks failing that is a little bit different than the other financial conditions.

Beckworth: Sure thing. And you raised this great point that Fed on one hand has a price stability mandate, on the other hand is a financial stability mandate. I guess you'd have a third hand, the full employment mandate. And right now, full employment's not an issue. We have low unemployment. It appears, if anything, to be at full employment and running a hot labor market. But there is the financial stability part of the Fed's mandate, which it's manifested in the words about keeping long-term rates stable. So there's that financial stability element and many people are saying, "Hey, Fed, you also have this. What are you thinking?" And as you noted, but the Fed has to tighten financial conditions to get to the place where it gets price stability. So it's a tough balancing act to have all those things. It's juggling three different balls at once.

Kelly: And it makes its supervision look bad. We could argue until we're blue in the face about whether financial stability still existed if SBV went down or not, whether it was really systemic, and we could talk about the designations they've made, but there's also the supervision aspect of safety and soundness. So even if it's financially stable to let a bank go, there's still going to be questions on the Fed of, "How did you let this bank fail? What's going on with these depositors? Where were the supervisors?" So, there's just a lot of strings in the Fed right now, and they're pulling in different directions.

Beckworth: One last thing on the bank run and the collapse of SVB, and I want to quote Jeanna Smialek from The New York Times. She says, "And this is not your grandmother's bank run. Depositors have cell phones, they have Twitter, they pulled 42 billion just on Thursday. Suddenly we have a solvency problem." I also heard Slack channels and other emails and other things are being used. So this is a new element, right? Something new to add to the mix.

Kelly: Yeah, it's interesting. First, I'll say poor Northern Rock is losing a little credit over this because Northern Rock in the UK, rescued in 2007, 2008. I would call that the first major digital bank run. They had a lot of online deposits, and when that basically just disappeared, and then when that became known, Northern Rock suffered a huge run, or suffered a huge solvency crisis, effectively. But yes, this is a new element. I'm not super convinced. This played a little bit of a role in Credit Suisse too, started on Twitter. That reporter said, "Oh, I'm hearing rumors that some bank is facing a crisis." So these things matter at the margin for sure, the way information can move. And of course it's going to move fast when you have a Silicon Valley Bank. These are the most tech savvy people in the world in theory. But again, it's tough to say there's not a fundamental case, both for Credit Suisse, for SVB. There was real things that happened here. Looking back in hindsight, there wasn't no problem with the business model. There wasn't something first that sparked it. And that's the age old banking crisis. Information has always gotten around and people run, but something typically causes it, and it was a broken business model in both cases.

Beckworth: Okay, so the bank shuts down on Friday, taken over by the FDIC, and then Sunday evening we get an announcement from the Federal Reserve, the Treasury, the FDIC. They have a joint statement they release on how they're going to support and help out, and the Fed itself introduces a new facility. But before we get to that, the Bank Term Funding Program, walk us through, what were the options on the table, at least legally, and based on what had happened in the past? You had a nice Twitter thread and we'll provide a link to it. But walk us through what were the possibilities before we knew what they were actually going to do on Sunday evening?

Initial Options for Fixing the SVB Crisis

Kelly: So once the bank was in receivership on Friday, the options were particularly limited, because then the conversation shifted to our uninsured depositors, which was basically all their depositors, and most of their liability structure. They were very heavily deposit funded. Are they going to be protected? And if so, how? So there was a big discussion of, "Oh, they need a bailout." The Fed, the FDIC, the Treasury, they've done this before. They did this in 2008, all these bailouts that we heard about. But their authorities have been majorly curtailed since then. So just to walk through the three crisis fighting institutions. The Fed, for one, basically can't do anything for a single institution anymore. So the rescues that it did, for instance, in 2008, of Bear Stearns, of AIG, it cannot do that. All of its liquidity assistance has to be broad-based. They define that this is by rule, not set in stone, as available to at least five borrowers, but they have to have a broad-based facility.

Kelly: So to set something up when the bank's already in receivership for SVB would've been virtually impossible for the Fed. In theory they could have said, "Look, we're going to rescue all of Silicon Valley.” This has kind of been the conversation in the UK, actually, with the UK version of SVB, is they were worried about their tech sector. So the Fed could set something up and say, "Look, we have a Silicon Valley emergency liquidity facility. Anyone in Palo Alto, you can borrow." Okay, that's probably available to five people. But the problem was that this bank was already insolvent. So that was also a new rule post-2008 is the Fed can't lend to either, basically stave off bankruptcy or to a firm that is insolvent. And the California regulator put out for the whole world to see that this bank is insolvent. It put that up on paper on Friday. The Fed's not going to ignore that. Again, we can talk about the definition of solvency, but the Fed's not going to ignore a banking regulator saying, "This bank is insolvent." So the Fed was effectively out.

Kelly: The other piece is the Treasury, nevermind the debt ceiling. The Treasury can't fund anything that hasn't been legislated. So it has basically one pot of discretionary money, which listeners may be familiar with, comes up in crises typically, the exchange stabilization fund. This fund now has... It had about $50 billion in 2008, and it was used to guarantee the $4 trillion money market fund complex. It now has over $200 billion. So you can imagine a world where this starts to make a difference for capital injections and things like that. It has a weird mandate. It started during the gold standard as a foreign exchange vehicle. Obviously we're not on the gold standard. It evolved a little bit with the IMF. And basically it exists for... It's in the discretion of the Treasury, in consultation with the president, to use for stable exchange rates and, broadly, a stable and avoid an erratic disruption of the financial and economic system. So that's some vague standard.

Kelly: But we know from 2008 that Treasury lawyers said, "You can't use this money to rescue Bear Stearns," which the Fed wanted. And then the same thing happened with AIG. The Fed wanted the money for AIG, wanted some Treasury backstop. Treasury lawyers said, "No, this doesn't fit the purpose of the fund." But then they used it two days later to backs stop money market mutual funds. $4 trillion, whole industry, clearly and important dollar market. Okay, so we get a sense of where the line is. But if Treasury lawyers are saying no to AIG and Bear Stearns, they're not going to say yes to Silicon Valley Bank. I mean, they're different lawyers, but still there's some precedent there.

Kelly: Okay, that leaves the FDIC. And the FDIC did a lot in 2008, mostly reluctantly, after being strong armed by the Fed and Treasury. But they did a lot. And their powers got significantly curtailed as well. So they can no longer do open bank assistance for a single bank. So there was some guarantees and agreements done between the Fed, the Treasury, the FDIC for Citigroup, for Bank of America, some unique structures they set up to protect their balance sheets in 2008. The FDIC can't do single bank open bank assistance. They can do broad-based debt guarantees still, but now they need the approval of Congress. They can assist a single bank if the bank is in receivership. And that's where we were on Friday. So then the FDIC has two choices. Its mandate is to resolve a bank using the least cost standard. The least cost to the deposit insurance fund.

Kelly: So sometimes you can qualify uninsured deposits as least cost if you're thinking of maintaining the going concerned value of the bank, you want to keep the franchise value. So you stand behind uninsured depositors and you get a better bid when you go to sell the whole bank. That option was probably off the table based on the announcement they made Friday. If you walk into a bank in the middle of the day and you say insured depositors are protected and you don't mention uninsured depositors, you've already eroded the franchise value. So that would've been a hard designation to make, but they've made it before. So then the only remaining option was to invoke the systemic risk exception, which allows them to avoid this least cost standard, and they can think about systemic risk, and we can talk about how systemic this actually was, but it requires the approval of the FDIC board, the Fed board, and the Treasury Secretary in consultation with the president. So that was the outcome we got over the weekend to ultimately protect uninsured depositors of SVB.

Beckworth: Well, let's go there. Let's talk about what would've happened had there not been this package, this relief to the banking system. Would it have been a systemic event? What are your thoughts? I mean, clearly a lot of people on Twitter were screaming "systemic event," and of course a lot of those people were tied to the bank or tied to Silicon Valley, some self-interest there, but there were also other people who were genuinely worried this could turn into something more. So thoughts? Is it something we should have been careful about?

Would SVB’s Failure Have Been Systemic?

Kelly: Yeah, I think the narrative has shifted probably at least twice. Early last week it was, "This is not systemic." By the weekend it was, "This is definitely systemic. This is a banking crisis." And now it's Wednesday and things have calmed down a little bit. So now it’s, “They acted too fast, they did too much, too fast. They should have let this thing go.” So some of the way I'm seeing this being characterized in the media is that the government declared these banks to be systemic… maybe going to do some hair splitting here, but I think that's not exactly what was done or the right way to put it. My personal view was that this was probably not systemic. To me, this was a very much a run on SVB's business model. That's why the runs that we saw were not on everybody with available for sale securities losses.

Kelly: And yes, SVB's were worse, but it was not a run on the banking system, it was a run on SVB and the banks that looked like it, First Republic, PacWest. So to me that's well within this fringe that the Fed is trying to burn off. It's not a run on the business model of Wall Street like 2008 was, where the whole world said, "We don't like repo, we don't like commercial paper." That's a run on the whole street. This was not that. And now what we're seeing today is a lot of funding flowing into the large banks, actually. So I doubt it was systemic, and in a world where policymakers had infinite authority to fight crises, infinite discretion, it probably would've been optimal to let the fire burn a little bit longer. Even if you're going to protect uninsured depositors, some of the stuff the Fed did, maybe not quite necessary yet.

Kelly: However, given that their powers have been so curtailed to assist open banks post 2008, I think this was probably the right risk management play, because you're doing it when SVB is in receivership. You don't want to roll this out when a bigger bank… I'm just going to pick like PNC Bank or US Bank. Nothing's wrong with those banks, but you don't want to size up and wait until that bank is in receivership. It would be different if you still have the authority to assist open banks, because then you could wait and see and you go, "Okay, now PNC and US Bank and Truist, these banks are starting to look bad. Let's react." But the authorities don't have those authorities anymore. So to me, this is the right risk management play of let's cut this thing off at the knees now. These banks aren't systemic, but relative to the powers we have to do more is going to be difficult.

Beckworth: So it was a reasonable response given the institutional framework, the tools at hand, and possible risk going forward. So let me play off of that and talk about maybe my views on the systemic outcome, and this is definitely Monday morning quarterbacking here, so guilty as charged on that count. But what you just mentioned, there's now flows into these bigger banks, the G-SIBs, because they are more regulated, they're also too big to fail, as we know.

Kelly: I'm going to jump in there quick. I'm going to say they have a better business model too, which is they have six businesses. I'd argue the only reason Credit Suisse is still alive, and it might not be for much longer, is that it's got a great wealth management business. So the run was on the rest of the company, but you have a diversified business model.

Beckworth: No, that's a good point. It's a great point. So many other banks that have a better business model, they also have maybe better regulation. So they were better equipped and they're now receiving the funds that were previously in some of these mid-size banks. The other point I think you're alluding to is just the expectation of these broader liquidity facilities are out there. I mean, should things get really bad, the Fed could turn them on again, right? 2008, 2020, so there's that notion that we know the Fed won't let the financial system collapse and it has more tools. So there's two things there. And the third thing I would add to your list then would be, we're in an environment of robust aggregate demand growth. I mean, it's not like we're in a weak economic environment. If anything, the economy's hot. So to me that's a key piece of the puzzle, financial stability. It would be one thing if spending were collapsing, if incomes were collapsing, but we actually are at the other side of that issue.

Kelly: Yeah, that's totally right, and it's making the Fed's job harder, but it's also... It absolutely weighs into how systemic this thing is.

Beckworth: Let's move on from the systemic considerations onto the actual rescue plan of the banking system and in particular those mid-range banks and what's available to them, and the big news was the Bank Term Funding Program, a new facility. And maybe you could argue it's just a tweak to the discount window. But before we jump into that, though, one of the things that's been interesting and a lot of debates surrounding it is the justification for it. And 13(3) was used, is that correct, to authorize this?

Kelly: That's correct, yes.

Beckworth: Even though this is through the discount window apparatus, which was another part of the Federal Reserve Act. So explain that tension and should we care that they're invoking 13(3) here?

The Bank Term Funding Program and the Use of Section 13(3)

Kelly: We don't totally know why 13(3) has been invoked. That remains a big open question. So there are terms to this facility that you can't do under the discount window. And the main one is just... The loan term is for a year for this facility. And the discount window has a statutory max that they can only do four month loans. But it's sort of a distinction without a difference because you can roll loans at the discount window.

Beckworth: Okay.

Kelly: So you could just have three, four month loans. I'm skeptical that the effectiveness of this program would've been any different if they would've said four months renewable maturity. So it remains a mystery why this isn't run out of section 10B, which is a discount window section of the Federal Reserve Act. And there’s speculation that there's a lot of stigma to the discount window and it's hard to get people to use the discount window. But it doesn't strike me as much easier to get people to use 13(3) either. I mean, this is an emergency, so we don't totally know why 13(3) has been used. This is now an authority that the market is very familiar with, that the Fed is very familiar with. So there's probably value in that of not explaining some new section of the Federal Reserve Act.

Kelly: It maybe gives them flexibility if they want to tweak these terms too. But we don't totally know why. Because the eligible collateral is actually all open market operation collateral, which is Section 14 of the Federal Reserve Act. So you can think about the standing repo facility which has been onboarding banks. And so you kind of go, "Okay, well that kind of looks like that too." So it's not totally clear why Section 13(3) is being used yet, but, I mean, the other thing to consider, I guess, is that the Fed’s signed off on the FDIC'S systemic risk exception. So the Fed is sort of already de facto saying, "Okay, we're in unusual and exigent circumstances."

Beckworth: To be consistent with that interpretation.

Kelly: "This is consistency." Yeah, right.

Beckworth: Yeah. That makes sense. Yeah, I was wondering about the standing repo facility, how different is that than this new Bank Term Funding Program? I mean, they could bleed into each other in some ways.

Kelly: Yeah, right. I mean, the only difference is that, as far as banks go, is you got to be in the repo complex, you have to be in the tri-party repo system, which a lot of banks are not and don't have a reason to be. And every bank has access to the discount window. So that's easier operationally.

Beckworth: Okay, Steven, let's move on to the specifics of the Bank Term Funding Program. And walk us through. What are the specifics of it and what's novel about it?

Kelly: Yeah. So as we sort of alluded to already, the collateral is the open market operations collateral. Typically... I mean effectively this is agency MBS and treasuries. Technically there's agency debt, FHLB collateral is in there. Munis up to six months are technically allowed. There's sort of a weird historical list of things that are technically open market operations securities. But for all intents and purposes, we're talking about treasuries and MBS here. So that's the eligible collateral. It's all bank counterparties. So everyone with access to the discount window, it looks like it's being run on the discount window rails. The biggest thing... We have a one year program funded at effectively the market rate, OIS plus 10 bps. And the biggest, most notable thing that's provoked a lot of discussion is that the collateral that the Fed is taking will be valued at par.

Kelly: And so the other piece of this is that they got $25 billion of credit protection of an equity, first-loss layer, from the Treasury’s ESF for 25 billion. This is very akin to what they did for a lot of… most of the facilities in 2020. But it's a little bit peculiar for a few reasons. One, the Fed talks about it differently. They used very hedge language in 2020. We have this first-loss layer from the Treasury, the Fed does not expect losses. The way they announced it now, they said, "We don't expect to resort to that $25 billion, we have it." But they sort of distanced themselves from it, which I think is fine, but they probably could have distanced themselves all the way from it in not taking it. You're taking treasuries and the MBS. As you noted, David, the Fed is already sitting on over a trillion dollars of losses on those very securities. So they've shown the capacity to carry those to maturity as needed.

Kelly: There's not going to be any credit losses on these assets. It's not clear they needed Treasury money. Maybe they wanted to look like a united front and have Treasury support, but the secretary already signs off on 13(3). So it's not totally clear to me why that money's there. It may become relevant if this goes a little further and they go, "Look, you know what? We need to start valuing corporate bonds at par too, and we're going to start taking loans." It'd be nice to have the 25 billion there, but I don't know why it's 25 billion. I don't think there's going to be any losses on the current collateral. Why not zero? Why not 5 billion? Why not 200 billion? I have no idea. But that's sort of another question mark at this point.

Beckworth: So the Treasury's providing an equity cushion should there be a loss. I guess, what would that take? So some of these banks go belly up and they default and the Fed's left holding the collateral and they bought it at face value, and it's worth less. Is that the risk?

Kelly: Yeah. So some weird things would have to happen. Because you have treasuries and MBS. And so first, the bank would have to fail for the Fed to get stuck with it. The Fed has recourse. They've done non-recourse facilities where you can walk away; this is not one of those. It's like the discount window, it's a recourse facility. So the bank would have to literally fail for the Fed to have to take on this collateral. It's treasuries and MBS, which it's already willing to bear losses on, it's going to hold to maturity on the rest of its portfolio. But also they're probably going to cut rates in the next year. So they'll probably get an asset gain there. So, okay, there's another thing that would have to be overcome. And then they have to basically not carry it to maturity. And then you have this weird thing where you would be taking... As you know better than anybody, we're talking about a combined government balance sheet and we're talking about the Treasury recapping the Fed because of losses on Treasury securities that are marked to market. And for some reason, the Fed had to sell them because inflation got so out of hand that it couldn't raise interest rates to a thousand percent for some reason. And then the Treasury's recapping it, which increases the problem.

Beckworth: Unlikely, highly unlikely.

Kelly: Yes. It's hard for me to envision a scenario where there's losses here. The other thing to mention here, actually, is the Fed would be senior in the capital structure. So if somehow they had losses based on their collateral to a failed bank, they they'd be one of the first in line to get their money out.

Beckworth: Okay. So there's a lot of reasons to believe that there won't be any actual losses for the Fed, even worst case scenarios. But in economics we like to say, "There's no free lunch." And so here, let me play devil's advocate with you. I think the economic cost would be in an implicit subsidy to banks. This is the knowledge that in the future this would be available, therefore will take on more risk. And at some point... So even though if there's not an actual recognition today, it's kind of the implicit subsidy going forward. And let me use that as a segue into the, maybe... Some would say critique, but others were more like, "Wow, this is amazing what's happened." So I'm going to quote you from The New York Times, actually. You had a great quote in The New York Times. I'm also going to quote a few other people from... Twitter was amazing.

Beckworth: By the way, Twitter, thank you so much for being around. You guys were just amazing over this past week. I learned a lot. I was able to sharpen my own thinking, having people like Steven and others on there. And dare I say, Steven, that at least for those of us in our part of Twitter, this crisis helped save Twitter. There were concerns about Twitter going down in the toilet with some of the management decisions. But man, Twitter proved its value many times over this past week. Okay, but here's a quote of you in The New York Times. I loved it, it said... This is you speaking, "The Fed has basically just written insurance on interest rate risk for the whole banking system.” Boom. Which is true, right? This interest rate risk which helped bring SVB down, the Fed is like, "Hey, we'll take the collateral at face value which means there is no more interest rate risk for you to bear, at least during the length of this loan."

Beckworth: Daniela Gabor, she wrote something in a similar vein. She said, "Forget about SBV liabilities for a second. The real bailout story is the regime change in the Fed's treatment of collateral. Par value goes against every risk management commandment of the past 30 years. It turbocharges the monetary power of collateral." Another powerful statement. Finally, Tracy Alloway on Twitter, really great statement here, "A bank with subpar risk management and flaky depositors just changed the US financial system forever." Alright, are pretty powerful quotes there, if I may say so. But help us make sense then, what are the big implications going forward, if any, from this change in collateral policy?

The Change in Collateral Policy and its Implications

Kelly: Yeah. I'll say two things and one will continue in this vein. It's a little critical, and the other is more hopeful I hope. So yeah, this is fascinating. It's clever in its easiness, so to speak. But, this is to Daniela's point, we think about effective central banking, especially in a crisis, as writing a floor under the market. The central bank is uniquely positioned to say, "Okay, this market is oversold," or, "the market writ large is oversold on account of a lack of buyers. We're going to go in, we're put a floor under it and we're going to let the market... With that assurance, the market's going to pick back up." Basically what the central bank is trying to do is rip off the left tail of the distribution. If you can get rid of the left tail, you still leave room for downside risk but you're not letting market forces sort of recover and you're solving this bad equilibrium problem.

Kelly: So that's what we like from central banking historically. The point there is like haircuts, you're letting some fire burn, you wait until there's... asset prices have depreciated. So when the central bank comes into assets, it can sort of be a patient investor and let things recover, and the central bank always makes money, things like that. This is interesting in that they've... I characterize it as almost a temporary capital injection, to the extent that that's not internally inconsistent in its own right. Because the whole point of capital is that it's forever. But this is sort of a temporary capital injection that will be paid back over the course of this year that this facility is in place. Because you can think about the way a bank funds itself, and some share of its assets are its own capital, right? Even if you're repoing a treasury, maybe you get 99 cents on the dollar. Or you take it to the discount window, you get 99 cents, you got a dollar of your own capital at play.

Kelly: So the unique thing about this facility is it didn't only just say, "Okay, we're going to write a floor under your losses. Bring us your available for sale securities. We'll give you the market value, don't worry about not being able to fund them anymore." And not only that, it didn't say, "Bring us your held to maturity assets that are valued higher. Don't worry about marking them to market. We'll give you cost accounting. We'll give you amortized costs." They said, "No, we're going to give you par." So you can think about this: If you have a Treasury security that you're holding at cost, held to maturity, let's say $100 face, you're holding it at 90 bucks. Interest rates have surprised to the upside, so fair market value is 80 bucks. You could take it to market and get 80 bucks or 79 bucks, or, now you can take to the Fed and get a hundred bucks.

Kelly: So not only did they say, "Well, we'll give you the 80 and we'll calm things down. We'll replace the 80 that you lost from uninsured depositors being flighty. Nor will we give you 90, which is what you've been carrying your books at, so you don't have to recognize the loss of funding there. We're going to give you 100, which can solve other problems on your books too.” So that's super unique. And Daniela's language is more dramatic than mine and more eloquent than mine, but that's exactly the point she's making. What I would say in a less critical fashion, and I understand why they did this, and I think it gives us a big reason to rethink the discount window, rethink the standing repo facility, make these things more useful going forward… The problem with these facilities remains that they lend against the market value of collateral.

Kelly: So to the extent that you really want to write a floor under something that starts on the asset side… So you can think about COVID, it's major dash for cash, treasuries are being sold hand over fist. So you can take that to the discount window, you're trying to replace repo funding, but you're going to get basically the same thing as the market because it's sold off. Your treasury that was worth a hundred dollars yesterday is worth 97 today, and the repo market's giving you 96, you can take that to the discount window and they'll give you 96. You still have to come up with the other three bucks that you were getting yesterday. And that's sort of the ultimate failure of the discount window and the standing repo facility, is we don't have a way to say... We talk about penalty rates like this.

Kelly: We say, "We'll give you a rate that's… it's slightly penalty relative to normal times, but it's not a penalty relative to the current market." So it'll be cheaper than the current market rates, but slightly more expensive than normal. And we don't do that with price. We don't do that with collateral value. We don't say, "Okay, it was worth a hundred bucks before this crisis. It's worth 90 bucks now. We'll give you 99." And so that I think is a big shortcoming of the current system. I don't have a perfect design in my head yet for how we reform that, but that's something we're thinking about here at the Yale Program on Financial Stability. We're trying to think about how we can basically make the discount window great again and make these classic facilities lose the stigma and make them effective in a modern crisis.

Beckworth: So I like the framing you just made, that this is effectively a capital injection, because they're paying above par. Well, they're paying par which is above market value, so it's effectively a capital injection. I also heard someone else put it this way on Twitter, "The Fed is effectively doing unsecured lending." Is that also another interpretation?

Kelly: So it is if you think, okay, you're getting 90 bucks of collateral for a $100 loan. I mean the Fed's not totally bound by those values. The Fed's legal mandate, especially with 13(3), is that it does not expect losses ex ante. There's really no reason to expect loss if you're the Fed. You're talking about treasuries… There's no reason to expect a loss if you hold this stuff to maturity, which is typically how they think about it, and they have 25 billion from the Treasury, I would argue unnecessarily, but I don't think the Fed needs to think about this as unsecured. It looks like a loan against underwater collateral, but they're going to get paid back, which is sort of their floor for secured to satisfaction.

Beckworth: Yeah. I saw on Twitter, you recently noted that in 2008, I believe, a Fed lawyer said the Fed could do something similar to this, right? They could buy securities at par value. Did I read that correctly?

Kelly: Yeah. The Fed legal division wrote a memo, in 2009, basically describing their thinking at the time that they did the CPFF, in 2008. Things were moving very fast in October 2008, so the memo got written in March 2009, that was intended to describe their thinking. Part of it was explaining this secured to satisfaction and exactly what amount of collateral they have to take, because what was unique about the Commercial Paper Funding Facility, and caused a lot of legal drama, I guess, inside the Fed, was commercial paper itself is unsecured.

Kelly: If you go out and buy unsecured commercial paper issued by McDonald's to make payroll, you have an unsecured loan against McDonald's balance sheet is what you have. You have a claim on their balance sheet and you go, "Fine, call it unsecured if you want, but it's McDonald's, they're going to be here in 30 days when they have to pay it back." That should be a security, and what the Fed did was it took an insurance premium on anything that was unsecured. If you were an unsecured borrower, you had to pay an insurance premium, which the Fed effectively used to capitalize the facility. What it was doing was taking 1% collateral, because that was the insurance premium. It looked like it was getting collateral, cash collateral of 1% of the value of the loan. That's sort of the genesis of this memo. They go, "Look, we don't have to worry about the collateral if there's other things to consider, we can be secured in other ways, effectively. It's not always about the value of the collateral at the time that we do the loan being more than the loan." Because that's the other thing to consider is sometimes collateral is just so destroyed. If it was MBS in 2008…

Beckworth: Right. Right. If you have a firesale…

Kelly: Right.

Beckworth: Yeah. You want to be thinking again, maybe longer term. Alright, let's move on to the reasons why the bank failed beyond its own poor business model, beyond the Fed's rapid rate hikes. That speaks to the regulatory failure part. There are a number of issues one could look at. I mentioned the auditors, too. There's a private sector role as well. The auditors, KPMG, 14 days before, gave a clean bill of health. I think I read 10 days before Signature failed, they also gave a clean bill of health to them. Now, to be fair to auditors, they're looking at last year's financial records, right? They're just saying, "Do these records reflect the reality?" But auditors are supposed to note, forward-looking, any concerns one might be thinking about and un-hedged interest rate risk, you'd think, might have been something they would've brought to the attention of the investors. But in any event, moving beyond the auditors, let's look at the Federal Reserve itself. The San Francisco Fed was the main regulator for the bank, right?

Kelly: Yep.

Beckworth: Okay.

Kelly: It was in the Fed's district.

Beckworth: It was in the Fed's district, and it was like the lead regulator. There's many regulators, but it was the lead regulator is my understanding. One question is why didn't they see this coming? Another regulatory issue would be the changes that were made, I believe in 2018, 2019, where they exempted these mid-range to large banks up until a G-SIB. And I'll just briefly mention a really fascinating article that came out, great timing, March 13th. The title is, *Monetary Tightening and U.S. Bank Fragility in 2023: Mark-To-Market Losses And Uninsured Depositor Runs?* It's by Erica Jiang, Gregor Matvos, Tomasz Piskorski, and Amit Seru. They go through and they look at this issue in real time.

Beckworth: One of the highlights that I just want to stress is table two, page 11 in their paper. They go through all of the balance sheets of the US banking system and they estimate that there's just over 2 trillion in mark-to-market losses right now, so in all of their assets. That includes Treasury bonds, but also mortgage backed securities, other types of loans. They estimate, to be precise, $2.2 trillion losses in mark-to-market. Of that, 1.3 trillion is from the large category, that's like from about 1.4 billion to 250 billion. This would be the range where we had the SVB take place. I'll also mention briefly, it was interesting to see a speech from the now current chair of the FDIC, back in 2019. He was worried about this kind of mid-range group of banks. Again, I think the biggest thing about them was this exemption they got or this deregulation they got in, was it 2018 or 2019? I can't remember the exact date.

Kelly: 2018 was when the law passed.

Beckworth: In 2018. Let's speak to both of those. Let's speak to the San Francisco Fed's role and then the role this change of regulation may have played.

The Regulatory Role in SVB’s Failure

Kelly: Yeah. We don't know much about the Fed's role supervision wise. The Fed itself has launched a review, which again, in light of our discussion about them basically precipitating this with interest rates, is somewhat humorous of the left hand working against the right hand again.

Kelly: They raise interest rates and then they do a review and go, "What happened?" We don't totally know. I would argue that the sort of idiosyncratic things with SVB, the lack of interest rate hedges, things like that, are probably a supervisory shortcoming as opposed to a regulatory shortcoming, which takes an umbrella view, a macro view.  Again, we're talking about a certain business model, a certain kind of bank that was run on last week. These are just management mistakes, stuff that would come up in the supervisory process that doesn't necessarily need to be overly specific in the regulatory process. We don't know. Part of this is just the secrecy of the supervision process. If Congress does an investigation, we may get some stuff publicized. I doubt the Fed will publicize much, but we'll see. Unfortunately, that's a big question mark still, is where exactly the supervisors were.

Kelly: As far as regulation, I guess I would say one thing upfront, which is we're talking about a bank that had a bunch of unrealized losses due to the Fed's interest rates hikes. You don't need one single regulation to see that as the market. If you see their balance sheet, if they have to report their balance sheet, that's the easiest thing in the world to impute if you're the market. That's exactly why we've had this ongoing known problem and there hasn't been a run the banking system, right? Because the market is not concerned with these unrealized losses in available-for-sale and held-to-maturity assets.

Kelly: The idea that it's 2018 that caused it is a little goofy to me, because this is just a widely known thing. Like I said, the FDIC put out a report two weeks ago showing all the unrealized losses in the banking system. This is the easiest, most transparent thing. That goes for the market and the Fed as well. The Fed could cut rates 300 basis points and recapitalize the whole system right now. They're not going to do that, because of inflation, but it's not the same thing as, "Okay, we have a bunch of toxic assets around the system. We don't know what the floor on those assets is. We don't know exactly who holds them or how they're funding them," that's a more runnable situation. But yes, so this is a question. SVB sort of got out of some regulatory burdens post-2018. They were sort of in that group that got exempted.

Kelly: Generally, again, I'm going to push back on the notion that… and we can talk about how appropriate it is to tighten or loosen regulations, I think that the direction was already going to be a little bit tighter under Vice Chair Barr. It looks like this has probably increased that impetus, at least in Congress, taken the air out of some of the pushback from Republicans, but I'm skeptical that stronger liquidity or capital would've prevented SVB’s fate. We're talking about $42 billion of deposit loss. Well, I should say the first very first thing that happened to SVB was they had a slow burn deposit run, because their clients were just burning through money.

Kelly: It started on the liabilities side. Nobody cared about their losses, until they started losing liabilities. Then the market goes, "Well, what's your funding plan? You bet it all on these depositors and they're running out of money." That's where it's a run on the business model and you talk about $42 billion of deposit runs in one day last week, that's 20% of their balance sheet. You're never going to write a liquidity regulation or a capital regulation that says hold 20% of your balance sheet in addition to what you're already holding.  You're just never going to write that regulation. You're no longer a bank. You collect people's cash and put it in a cookie jar if that's how high your liquidity capital regulations go. This doesn't help a run on a business model. It goes back to the point I made about Credit Suisse, great capital ratios, great liquidity ratios, and the market's saying they're undercapitalized. You go, "How?" Look at their tier one, look at their Basel ratios, why is the market running?” That's been sort of some of the confusion about the contagion. What little contagion we saw in the last week is, why is the market running on Charles Schwab? It's a concern over a business model, and the Charles Schwab one is probably overdone, but it's exactly that. Banks are probably running on too much capital normally, as in, it makes sense most of the time to have two, 3% capital. That's the incentive of the system. Sometimes you need 15% capital, so it's good that we do that, but it's not always going to be enough. If your counterparties are no longer willing to do business with you, you're out of business no matter how much capital you have.

Beckworth: Great points, and I really like this point you made that there are many other banks in the same size range as SVB that would've been affected by the change in regulation that aren't in trouble. They also have mark-to-market losses, but they hedged, they had a better business model, they did things differently. It's not the regulations per se, it's the business model that's really the culprit here. But it will be interesting, as you note, to see what happens to the Republicans' push against the "holistic review" by Michael Barr. I guess that's kind of put a damper on what they will be saying the next time he or Chair Powell comes before Congress, for sure. One last thing to wrap this up, and this is… we're going to get more deep into the theoretical weeds here, but the famous Diamond-Dybvig Model on bank runs, just won a Nobel Prize… At this juncture, it may not be that useful, but does this situation fit it or not? I think what we've been saying is it doesn't, because the Diamond-Dybvig Model is much more just about pure panic and running based on what you think someone else is going to do as opposed to solvency. Did I get that right? It's more of an irrational response?

Kelly: Yeah. They talk about sunspots causing the run. Far be it for me to critique the Nobel Prize winner of the last year. But yeah, I don't know how terribly useful it is. It's a key insight that existed in other fields as well. But what we know about banking runs is that there are precipitating factors. If you look at historical banking crises, there is some turn in the business cycle. There is recessionary news that comes in.  As you alluded to earlier, maybe we're living in a new age of where a Twitter rumor can start it, but even rumors, you typically have had some foundation in the event that they're actually effective. I don't know how useful it is. Like I said, I see this as a run on the business model, and a business model that, in itself, is not systemic. I think that that differs from what we saw in the 2008 financial crisis, for instance, where the business model that was being run on that could not be bought by the rest of the banking system. It couldn't just immediately be replaced. You had to inject new capital from the government and you had to restart the system that way.

Beckworth: Okay. With that, our time is up. Our guest today has been Steven Kelly. Steve, thank you so much for coming back on the show.

Kelly: Thanks, David. Great to be here.

Photo by Rebecca Noble via Getty Images