March 14, 2023

Every Complex Banking Issue All At Once: The Failure of Silicon Valley Bank in One Brief Summary and Five Quick Implications

Financial crisis nerds have had quite a weekend. In short order the one-of-a-kind Silicon Valley Bank began to reportedly be experiencing losses, and large uninsured depositors began loudly and aggressively “pulling” their deposits.

Every Complex Banking Issue All At Once: The Failure of Silicon Valley Bank in One Brief Summary and Five Quick Implications

ANNOUNCEMENT: I have a well timed and previously planned trip to the San Francisco Area coming in mid-to-late May. So if you would like me to speak to your organization about uninsured deposits, the Federal Reserve or any other topic, please email me at crisesnotes[at]gmail[dot]com

Financial crisis nerds have had quite a weekend. In short order the one-of-a-kind Silicon Valley Bank began to reportedly be experiencing losses, and large uninsured depositors began loudly and aggressively “pulling” their deposits. Following that, the bank predictably experienced liquidity issues, and then was taken over by the Federal Deposit Insurance Corporation Friday morning. A haze of uncertainty about what the governmental response would be emerged with the issue of significant losses for uninsured depositors — reportedly a raft of tech start ups —on the table. Recall that any account with a balance at or below 250,000 dollars in a FDIC member bank is “insured”, and thus the government will promptly make sure that a depositor has access to that money — “dollar for dollar”. Above that amount, the deposit account is “uninsured”, and could experience losses if the bank does not have enough assets of equivalent value.

As debate raged over whether the tech industry’s bank account large balances should be fully saved by government action, unofficial details of a government response emerged. These details were confusing. Initially it was reported that the FDIC and the Federal Reserve would create a “joint facility”. What would a “joint facility” between the Federal Deposit Insurance Corporation and the Federal Reserve look like? What would that entail? Since the details didn’t make sense, it was hard to tell whether this was an incoherent press rumor, or early reporting on a real plan that was in the process of being executed. It turns out it was the latter.

However, what actually emerged made more sense, but was still shocking. The Federal Deposit Insurance Corporation announced it was covering all depositors, whether they were insured or not. At the same time, but separately, the Federal Reserve is launching a new facility called the “Bank Term Funding Program”. This facility is an expanded form of the discount window, which has a few notable features. It provides loans of up to 1 year to any “depository institution” against any collateral which the Federal Reserve buys in open market operations. In other words, the collateral can be a treasury security, a government guaranteed mortgage backed security, or similar security. This facility’s further notable feature is that, for the purposes of securing a loan from the Federal Reserve, these bonds are valued quite favorably. I will also discuss this in more detail later on. But you can also read my basic 2020 explainer on the neglected area of central bank “collateral policy” here.

At first, it wasn’t clear what legal authority this facility was being launched based on. But then it was announced that they were using 13(3) authority which requires “unusual and exigent circumstances”. I will discuss the legal implications of this below, but for now it suffices to say that it is very surprising for 13(3) to be used under these economic conditions and in this circumstance. That does not mean it is illegal or that it was unwise to invoke this legal authority. It is simply surprising. There were concerns that there would be a “run” on other regional banks. But it seems that this action was at least equally motivated by the knock on effects on the companies that banked with Silicon Valley Bank — and especially two payroll processing companies. In other words, companies that responsibly manage their own banking relationships might see their workers not receive their paychecks because of disruption to the company they pay to send out individual wage and salary payments to their employees. Yesterday’s Politico article is well worth reading to get some sense of what the internal Biden administration discussions were like over the weekend.

Regardless of the reason, they did it. Monday passed without incident, though there are anecdotal reports that deposits will continue to flow out of regional banks, and towards the largest banks such as Goldman Sachs or Citigroup (an issue I will turn to next). The stock prices of other regional banks did go down by, in some cases, more than 50%. However, stock trading should always be looked at skeptically when analyzing financial issues. Sometimes financial institutions are in very bad shape, and the stock price declines reflect larger liquidity and/or solvency concerns. Other times, stock prices reflect the unique and distinct opinions — and herd mentality —of stock traders (who rarely talk to bond traders, let alone detail oriented financial analysts). These traders simply bet that lines will go down because the vibes are off, but don’t have much impact on other financial markets. My inclination, given the evidence I’ve seen, is that Monday’s stock trading experience reflected the latter, rather than the former. However, it's too early to tell. Though as of this writing these same banks have experienced stock price gains, so it seems even stock traders today are agreeing with my assessment of their behavior yesterday. Even still, I am not anywhere near confident enough in that hunch to commit to it, and then see regional banks experience fatal runs.

Readers should be able to tell by now that quite a number of issues have been raised by the failure of this one quixotic bank. One piece can’t comprehensively summarize and dissect all these issues. Indeed, I suspect I will be dealing with the events of the past weekend for the next four months, at the very least. However, I can briefly touch on the five issues which have been on my mind the most for the past 100 hours. Some of these observations I will return to at much greater length in the coming days, and weeks.

1) Bank money is money unless it is not money

Oftentimes in financial writing, financial writers shy away from the biggest topics. It’s hard to tie the big picture issues to current events a lot of the time, and they can seem abstract or abstruse. It’s hard, even for the clearest writers among us, to make these issues seem lively and important. Strangely enough, in this respect, I am very grateful to this mini-episode. The reason is this episode has dramatized my favorite and most abstract topics, three of which I will quickly discuss in this piece.

The first one is this topic —is a bank deposit money? This might seem like a silly question. Outside of physical cash, bank deposits are the most common form of money we use. We pay off our credit cards, we pay rent or mortgages, and we receive our paychecks in the form of bank deposits. Not only do we use our bank account balances as money, we often use it at par. That simply means that if our credit card bill is 1087 dollars, the credit card company debits our account — at our direction —for the amount of 1087 dollars and the amount is paid off. This point is so basic and simple, many readers will be confused by it out of an assumption that I must be saying something more complicated.

Of course, sometimes making payments out of our bank account involve fees. This interferes with the exact equivalence between different dollar amounts that exist in different forms. However, a fixed fee only minutely separates equivalence between the balance in our bank accounts, and the payments we must incur a fee to make. On a percentage basis, this divergence from “par” may be insignificant. Even the percentage based processing fees charged by many payment processors and credit card companies are small and stable. For example, Stripe is the payment processor which lies behind this newsletter. It charges 2.9% +30 cents. This is sand in the wheels, but it does not disrupt the whole enterprise.

This all might seem far afield from the failure of Silicon Valley Bank, but it is not. In some fundamental sense, it's the critical issue. Because bank deposits are not money (at least not fully) to everyone. To you or I, who can’t dream of having anything close to 250 grand in the bank, for all intents and purposes, bank deposits are always money to us. (Rich people, I hope you can forgive me for not writing as if you are my audience. Speeches can be more easily tailored to your perspective and concerns). However, to businesses, bank deposits carry risk they have to manage. To businesses, they need to put their excess cash in “alternative” financial assets if they want to be completely assured of safety.

This is why short maturity treasury securities are such useful instruments. You can hold billions of dollars, without taking on the risk of a bank failing. The issue of “shadow banking” also emerges because of uninsured deposits. The siren song of “good collateral” which could back “shadow money”, and thus make it more secure than an uninsured bank deposit calls out to these big depositors. It helps when they are offered a better return. I can’t cover this issue at the length it deserves, here. But it suffices to say the seemingly small issue of uninsured deposits have big implications for the financial structure of the entire global economy. Tackling that issue was one of the things I took on in the monetary policy report I released early last year.

Some wish that the big Silicon Valley depositors had taken losses. It is true that these balances could have been better managed and these tech businesses seemed to have done little to no due diligence. One of the options available to them was actually a financial technology (“Fintech”) option, that should have excited tech executives. With a system called “Insured Cash Sweep”, a large uninsured deposit can be transformed into quite a large number of insured deposits of 250,000 dollars in dozens, perhaps even hundreds, of other banks. All it costs is a fee. This option is available in amounts ranging between 50 million and 150 million dollars. If Silicon Valley Bank had run this system, it might have earned a reasonable amount of fee income, those deposits would have been safe. It then could have avoided holding a large quantity of long maturity securities which it would come to take losses on.

Yet, look at this from another perspective, what purpose does that convoluted system for backing uninsured deposits with insured deposits serve? In some ways, these “brokered” deposits look more like the shadow money issued outside the banking system (or more accurately, outside the banking subsidiary). In other words, this is just another way of reassuring those with large balances that they have good enough collateral to treat these balances as unconditionally money. What’s unique in this case is that the collateral is simply insured bank deposits. All this nonsense can be eliminated by simply removing the cap on deposit insurance, and admitting that a bank deposit is money, and should just be treated straightforwardly as such. The alternative is treating bank deposits as “fully” money most of the time, except when a crisis hits, or you’re engaging in risk management at a corporate treasury. Meanwhile that ‘risk management”, has all sorts of negative knock on effects.

Of course, the other reason to avoid fully and unconditionally declaring bank deposits money is that this has big implications. The other situation people like to pretend that bank deposits are not money is when they want to treat bank deposits as a matter of private initiative, and discipline. Rather than acknowledging that banks acquire assets based on the powers granted to them by a bank charter, some would like to pretend they acquire assets on the basis of the discipline, and oversight of uninsured depositors. In this vision the regulations we have are supposed to prevent financial crises, and protect the FDIC’s “insurance fund” which covers insured deposits. But they are not designed with the idea in mind that every asset a bank acquires happens on the back of the full faith and credit of the United States government.

Before this weekend, most people did not see things the way I and my closest colleagues did even a number of banking law scholars I am friendly with. I still think that my demand management report from  last year was unique in tying together this perspective with a fully fleshed out framework of just how much we should tighten the rules regulating the quality and quantity of assets that banks can acquire. In the most comprehensive version of this program (as opposed to an alternative “minimal” program), the implications were far reaching and involved bringing those financial institutions deeply reliant on loans and lines of credit from chartered banks into the chartered banking system. Few have gone as far as that “comprehensive” proposal. Yet after the events of this weekend, opinions are clearly quite shaken. The prospect of unlimited deposit insurance, whether de facto or de jure, is leading to a large-scale reconsideration of views among even “moderate” banking scholars. With the remaining embers of the dream of liability side discipline of banking being extinguished, proposals for far tighter asset side regulations will clearly proliferate.

Bringing things full circle, such a comprehensive program could tighten financial conditions throughout the economy. This could reduce aggregate demand without direct effects on interest rates, and certainly with leaving governmental “default risk free” interest rates low. From my perspective, relying much more heavily on direct credit regulation solves the problems posed by fully and unconditionally treating bank deposits as money, and the issue of providing alternative monetary policy tools for managing demand conditions. Two birds, one stone. Uninsured deposits have been a nagging issue hanging over from the last 40 years of banking history. This moment may be just the thing to resolve that issue forever, simplify banking and put an alternative vision of banking and its place in society on the table.

2) Central bank collateral policy makes a sudden, splashy return

I’m not going to be able to devote the time to this issue today that I would like because it's one of the most fundamental issues to come out of this crisis and is an issue I’ve tried to hammer over and over. The flipside of bank runs is the Federal Reserve’s role as a “lender of last resort”, or even simply a lender to banks. This lending typically happens through the “discount window”. As I wrote in 2020, a number of the facilities that the Federal Reserve launches in crises are simply expanding the discount window in ways that they are not authorized to by the part of the Federal Reserve Act which governs the discount window (more pointedly, they may also simply want to keep such temporary expansions separate from the discount window all together). This is one of the most technical and neglected areas of central banking and, until this weekend, it had receded into the background as 2020 moved further into the past.

One of the major issues that discount window discussions center around is the issue of collateral. Central banks have all sorts of choices in what value they can place on the asset used to secure a discount window loan. Recall that collateral is simply an asset that an entity can take ownership of if its debtor defaults. For more discussion of collateral and central bank collateral policy, read this introductory piece I wrote in 2020. For the purposes of this discussion all that matters is any asset has three prices, its acquisition price, its market price and its collateral price. To avoid taking losses lenders like to set collateral prices significantly below market prices. The flipside of that choice is that even if there is no stigma to borrowing from the Federal Reserve, there is only so much of your funding you can get from the Federal Reserve. If an asset was acquired for 100 dollars but its collateral price is only 70 dollars, then you need 30 dollars of funding from elsewhere even if you pledge the full value of the asset as collateral for your discount window loan.

Silicon Valley Bank experienced this issue in spades as the market price of its assets had fallen below their acquisition price. The collateral price was, in turn, significantly below that market price. Thus, even if Silicon Valley Bank had begun borrowing from the discount window in large amounts it would not have been enough to cover deposit outflows. Nonetheless, it's still unclear why they did not go to the discount window, though many have pointed out that they began borrowing from the Federal Home Loan Board, which is a kind of “lender of second-to-last resort” and which needs discussion in its own right. What was extraordinary about this weekend is that the announced Bank Term Funding Program is setting collateral prices for securities at acquisition price. This means that you can borrow 100 dollars against an asset you acquired for 100 dollars but which now is worth 50 dollars. If this had existed Wednesday, Silicon Valley Bank would have been able to cover all of its deposit outflows. It’s hard to emphasize how good of a deal that is relative to past conventions around the discount window and Federal Reserve collateral policy.

In a way, this topic is simply another way of discussing the last section. Banks, through their bank charters, are allowed to acquire assets by issuing bank deposits that are treated as money. Any individual creation of bank deposits may lead to a payment outflow that must be covered by the account that bank has, in turn, with the Federal Reserve. But in aggregate, the banking system holds trillions of dollars of assets which are not matched by borrowing at the discount window. Instead, they are matched by the banking deposits that the banking system has issued. In normal times, this is basically irrelevant. Yes, banks periodically have to “compete for deposits” to find funding that is cheaper than interbank borrowing. But the idea that bank asset growth is limited by trigger happy depositors looking for the exit is clearly untrue. Bank runs happen, as we have just seen. But bank runs are not typical and they are the textbook example of when a central bank is supposed to step in. A bank experiencing rapid asset growth may have higher funding costs than banks growing less fast, but it is their high costs which will lead them to slow growth (outside of control frauds), not some conception that they “can’t find funding”.

So let's step back for a moment. As I said on Saturday on twitter “A lot of the discussion this weekend is coming close to accepting @rohangrey's argument that any asset a bank is authorized to make should be acceptable collateral at acquisition price at the discount window.” This point was quickly illustrated by the Federal Reserve’s actions the following day (and quoted by Matt Levine at Bloomberg on Monday!). When I said this, I didn’t think the Federal Reserve would follow our recommendation in less than 24 hours. If the discount window only has unattractive collateral prices and limited availability when banks mostly don’t need it but expands- either directly or through crisis facilities- when they do, isn’t it kind of like the Federal Reserve is always backstopping banks acquiring and holding those assets?

The difference then, which again goes back to the previous section, is that the “tight” discount window in good times facilitates the fiction that these assets are acquired and held on the back of the volition of private actors, rather than the backing of the United States government. This, in turn, justifies far looser asset side regulations. The alternative, which my colleague Rohan Grey at Williamette University has proposed in detail in a book chapter entitled “Banking in a Digital Fiat Currency Regime”, is for all bank assets to be automatically matched with a discount window loan of the same maturity which treats the collateral price as the same as the acquisition price, just as the Federal Reserve did this weekend with its new crisis facility. The chapter is about “digital fiat currency” because proposals for a new digital currency are typically met with concerns over deposit outflow. Many more people holding these balances, far fewer bank deposits and thus less funding for banks.

If you look at the world the way Rohan and I do, deposit outflow as a result of digital fiat currency, bank accounts directly provided by the Federal Reserve to the public and/or postal banking is a feature not a bug. Once we separate and clearly distinguish payments processing from bank lending, we can see that all chartered bank deposits do is paper over the fundamental contradiction between the assets we let banks acquire through their own money creation and the lack of official backstop in good times for holding those assets through direct borrowing from the Federal Reserve. It is absolutely perverse to intentionally weaken new technologies and better payment processing options out of fear of banks “losing deposit funding”. The gap between the terms and quantities that banks can access the discount window and the terms and quantities banks can create bank deposits is a gap measuring the extent to which our financial regulatory and banking regime is lying to both ourselves and each other. Why would we want banks to control payments processing and credit to the non-financial economy? If we can separate these businesses, and we can, then we should.

It should be obvious from the far reaching implications of what we are saying why there are few who fully share our views. Yet, this weekend was again a turning point. It wasn’t Rohan and I who tied fully protecting depositors,  facilitating the holding of bank assets and increasing the collateral price, maturity and availability of Federal Reserve lending. The joint statement from the Treasury, FDIC and Federal Reserve did that. They have hammered and highlighted the contradiction between what you can fund with uninsured deposits and what you can fund directly with the Federal Reserve and they choose to shrink that gap by expanding the discount window. This weekend has been an almost unbelievable gift to the research agenda we are engaged in.

3) Is there really systemic risk in the failure of Silicon Valley Bank?

When systemic risk is being discussed, usually it is assumed that we are talking about very broad implications. The Great Financial Crisis was a classic example. What the public was told was “if all the biggest banks in the world fail, the entire global economy is going with them”. Yet it should be obvious that this term is fairly slippery. What exactly does systemic mean? How wide are the consequences of “systemic”? After all, we all exist in a great chain of being, and something that affects one of us affects us all. If you use the definition proffered by the Three Musketeers, any risk at all is a “systemic risk”.

This circumstance is clearly not quite that extreme, but it is also certainly not a cut and dry case of “systemic risk”. Indeed, the prospect that uninsured deposits would be covered in full by the Federal Deposit Insurance Corporation seemed relatively dim. At least, if you followed the debate over the weekend among the small set of experts who have both financial system knowledge, and expertise in the legal nuances of the U.S. financial regulatory apparatus. This was precisely because of the legal limitations the FDIC had operated under in the aftermath of the Savings & Loan Crisis. Known as the “least cost” test, it restricts the FDIC from taking actions which will involve greater drawings on the Deposit Insurance Fund. It will not surprise readers to find out that I think this whole schema is a silly “accounting gimmick”.

Regardless, the main exemption to the “least cost” rule is known as the “systemic risk” exemption. If not covering those uninsured deposits is legally designated a “systemic risk”, they will cover them. I personally thought that there is more leeway to work around the “least cost” rule in covering uninsured deposits than others expressed, rather than invoking this exemption. But that is more of an irrelevant detail now. They invoked the systemic risk exemption, and here we are.

What is still not clear is what exactly they identified as a systemic risk. In invoking the exemption, they don’t have to say. More to the point, as I am always quick to point out in these circumstances, who has legal standing to challenge their legal interpretation in court? As you can see from the summary above, they didn’t think the entire banking system was in crisis. Does the liquidity situation of a number of regional banks, which may be able to weather this storm in a way Silicon Valley Bank was not, count as a “systemic risk”? The Politico piece also discusses concerns over the fate of two payroll processing companies.

That clearly matters, but if that is enough to have “systemic risk” then we are a long way down from the interconnections feared in the Great Financial Crisis. As we’ll see in the next section, this defining down of systemic risk, or its Federal Reserve Act equivalent, has major implications for how these powers will be used in the future. In this respect, it can be a mistake to get too bogged down in the details of what actions they’ve taken and what knock on effects they discussed over this very busy weekend. The big picture here is that these powers can always be invoked as long as there is a possible effect that can be found which can be characterized as systemic and has a plausibly decent chance of happening. That legal standard is not a standard at all. The restraint is just what actions congress will sanction if taken in the future. That is a very loose standard, to say the least.

4) 13(3) is no longer an emergency authority

Columbia law professor Lev Menand likes to tease me over extremely nerdy financial regulation dinners about what he sees as my extremely, ahem, “creative” and “expansive” interpretation of the Federal Reserve Act. If you read his recent book, “Fed Unbound”, you will understand where he’s coming from. The discretion the Federal Reserve has claimed for itself without much debate is truly staggering. In our conversations I like to emphasize the selectiveness of the Federal Reserve’s legal creativity, rather than its expansiveness. To me what’s far more concerning is who the Federal Reserve gets creative for, rather than the mere fact that it interprets its legal authority widely (rather than narrowly.) A Federal Reserve which pushed the limits of the Federal Reserve act on behalf of state and local governments would have produced a more well functioning and less unequal economy over the past forty years.

Yet, this episode has pushed even my limits. If you read section 13(3) of the Federal Reserve act, which is the basis for virtually all emergency lending that the Federal Reserve has engaged in since 2007, the key phrase that debates focus on is “unusual and exigent circumstances”. Thus, almost every legal question about the limits and scope of 13(3) hangs on what the legal interpretation of this phrase is. Conventionally, this phrase is taken as a flowery way of saying “emergency”. So the debate shortcuts to what exactly counts as an emergency, which is a familiar issue from other areas of law. One obvious recent example is whether, to what extent and for how long does the global Covid-19 pandemic count as an “emergency”. How big of an event does something have to be to be an emergency?

As we’ve seen in the last section, this latest event has shaken these interpretations to their core. If this is an emergency, then basically any moderately sized event with a knock on effect on the economy is an emergency. In other words, now either everything is an emergency or this is not an emergency power at all. Which interpretation is more accurate is more of a philosophical question than anything else, and has little practical significance. “Unsual and exigent circumstances” seem now to be little more than a day where there was moderately bad weather.

Part of what colors my thinking about this is that I was, as readers know, a huge advocate of state and local government lending by the Federal Reserve throughout the first two years of the crisis. However, over time I stopped focusing on this point (at least in my day to day writing) because it seemed to me to lose practical relevance with each passing day once the crisis facilities were closed. The Federal Reserve is raising interest rates, there is going to be no interest in even listening to a proposal to use state and local government lending powers, let alone to use 13(3) authority to do so. Chairman Powell clearly believes that such a program would even be illegal under today’s circumstances, and would likely say so if asked.

I of course disagree with that legal interpretation, but I have found it hard to muster the effort to argue about it. It seemed pointless to me. But now that the Bank Term Funding Program exists on the basis of 13(3) authority, I am compelled to return to this issue. Because they are invoking 13(3) under these economic circumstances i.e. significant inflation and low unemployment. Not only that, the “emergency” involved here is, at best, very debatable. Before I was battling for a more expansive interpretation of 13(3) which allows use in otherwise very good economic times. Now the Federal Reserve has provided me precedent. Now, they only avoid engaging in state and local government lending because they don’t want to. Because what is an emergency has always been, but definitely is now, simply a matter of taste. Just ask Detroit, or Puerto Rico.

5) Competition law and Banking.

This is both the last topic I will briefly cover. Flouting tradition, it is also the least important. Competition law in banking is more an intellectual curiosity that I hope to return to in future years. There are many unconfirmed reports that Silicon Valley Bank required its customers who wanted access to loans, or lines of credit to hold all their operating cash with Silicon Valley Bank. The implication of these unconfirmed reports is that these uninsured depositors were compelled to take on this default uncertainty, which exploded over the past week. These kinds of arrangements are called “tying arrangements” in competition law, also known as “vertical restraints”. Financial regulators in the United States have jurisdiction over what would otherwise be competition law issues, and there are anti-tying financial regulatory laws. However, there are exemptions to these anti-tying laws. In general this is a good impetus for reexamining those exemptions, and consider narrowing them. Obviously, it's also important to confirm those unconfirmed reports, and find out more details.


That’s all for today, but all these issues need greater examination, and more discussion. In particular, the history of the 1984 failure of the Continental Illinois bank needs to be reviewed and discussed in detail — as well as how that failure changed the FDIC. The Federal Reserve’s discount window and its role in the big picture of banking law needs to be discussed from a wide variety of angles. Financially, the ever-widening meaning of section 13(3) of the Federal Reserve Act will not be going away, not by a long shot. It’s going to be quite an interesting series of weeks as the full implications of this weekend play out. I look forward to covering them for you.

This was a free piece of Notes on the Crises. I will be writing more about the many, many implications of this weekend financial panic over the coming weeks — so please take out a paid subscription. However, even free subscriptions help me and will help you follow along with my free pieces.