From Boom to Bust: How $140 Billion in Deposits Led to SVB's Collapse

>> Amit Seru: So let us think a little bit about this in the context of Silicon Valley Bank. So, all of this has to do with timing of deposits. So what I have plotted here on the x-axis is time in years and quarters, and on the y-axis is the amount of money coming in, deposits coming in.

And the key thing here is between the two vertical lines. What is that? That's basically $140 billion. So remember that 185 billion I showed you before, 140 billion went through the doors between 2019 end of quarter 4 and 2021, okay? So between that time period, we looked at $140 billion of deposits coming in.

Now, this is the time period when, if you remember, we had COVID lockdowns, firms were not investing, there was uncertainty, so there weren't that many investment opportunities, loans to be made. Banks were very cautious, and so the question is, what do I do with this 140 billion? I have to pay them 1%, but where do I get my 2%?

So that's the conundrum that Silicon Valley Bank and other banks are facing. So what do we do? Well, we look around and we look at these securities, so this is treasuries. One way to capture the whole investment curve with vis a vis treasuries is to think about the yield curve.

What's that? If you were to invest in government bonds, how much return do you get at different horizon? That's what this is trying to capture. And there are two lines plotted here, basically between January to December 2021, this is when the deposits are coming in, on the left hand side, as you see.

On the right hand side, this is what the bank is seeing, because there are no loans to be made, so let's figure out safe securities where we can park the money. Now, where am I getting some return here? Well, the rates were pretty low, so if I invested in short term, I get nothing.

You can see the yield curve is almost 0. But somewhere in this middle range, I get something. So it seems like a great idea to take this $140 billion and put it in the middle range, which is like tying in money for three, four, five years. And that's what is known as managing your time.

But this you might call as mismanagement of time, because you're not really covering up the risk that you might be taking. What risk? This is known as interest rate risk. You might have heard of this. What's the idea? That, well, you saw in the US economy, what happened is we had inflation.

Once inflation became pretty persistent, it was clear that we got to raise the interest rates to slow the economy down. We did that pretty quickly, and as a result, what happened? Well, if you look at the interest rate in the economy, it went up. So the yield curve that I have plotted here shifted pretty quickly to that, and that's a three percentage change in the yield.

Now, what does that do? What is this interest rate risk? Well, whatever assets you have, you are getting some return on them because you invested them in 2021. But now their value is much lower because you could actually go out in the market and invest and get 5%, but you were promised 1%, so the value of those assets is low.

That's interest rate risks, assets are revalued to a lower value, okay? So you got deposits in, you invested, it seemed a good idea, but now we have this interest rate risk. So what does this do to the balance sheet? That's something that's important. So when you look at this timing and you feed it to the balance sheet, that's what will tell you what will blow up.

So, the simple and quick way of doing calculations when interest rate changes is something that we teach in undergrad and very elementary parts of our MBA curriculum. Where if you have the duration of the assets or liabilities and you know what the interest rate changes and how much the money is tied in for, you can quickly compute the loss or the gain, okay?

So just trust me when I tell you that when you look at the numbers on the asset side, given the 3% change, you're talking about about $22 billion loss. So 3% interest rates went up, value of the assets in market value terms fell down by 22 billion. So that's on the asset side, similarly, you now owe less than what you thought you did because someone else invested and now the interest rates are higher.

So their effective value of the liabilities is lower. So, on the liability side, you gain, on the asset side, you lose. But the problem is, on the liability side, it's a short-term, short duration asset liability, so you essentially have a very small gain. What that effectively means is, you're talking about a loss, effective loss of about 20, $21 billion for Silicon Valley Bank as the interest rates went up.

Now, why is this important? Well, think about when you buy a home and you have a mortgage, house prices fluctuate, the value of the home moves around, and sometimes you're underwater. Big deal, you don't just go crazy. So, similarly for the bank, these are losses. Yeah, these are paper losses.

These are unrealized losses. So why do we care? Well, when it comes to our home, we don't care as long as we make the mortgage payments. But if the lenders become spooked and ask for some other terms because they are worried about the neighborhood or something, bad things can happen.

In the context of the bank, what that means is what the depositors do is gonna have a big bearing on whether these losses remain unrealized or if they are realized. So they are unrealized losses, but clearly, $21 billion, but they can matter. Why do they matter? Because remember that equity capital that I showed you?

$21 billion, those losses can wipe off the equity and the bank can go under. So they're unrealized, but if they become realized, it could be a problem. How do they become realized? Well, the depositors have to come in and ask for money. Why? Because when they ask for money, I the banker, have to go out and sell these securities.

When I sell these securities, I'll sell them at a loss because the market value is lower. And when I do that, maybe some other depositors get spooked, and there might be a self fulfilling run which leads to the bank going down. So it can happen, but it may not happen.

So, in Silicon Valley Bank, what happened? Well, the first thing was it had 90% of deposits that were uninsured. What that means is they were, basically, there was no FDIC guarantee. So if you lose out, you lose out. So they get spooked very quickly. Typically, we think of depositors as being sleepy.

Think about yourself. I mean, now maybe you are paying more attention, but think about a year ago, you were not paying attention to where your money was in the bank. And so, you're considered sleepy. But the uninsured depositors, who have a big amount of money and who are not insured, might be doing a more thorough job in terms of keeping track of whether the bank is still solvent or not.

Because otherwise they lose money, they are not insured. The other thing is that, okay, they might be monitoring, but really, do we really think that all the companies and all venture capitalists are doing this all the time? No, but if some wake up and they are all connected, which happened to be the case in Silicon Valley Bank, you have a situation where some spooked depositors line up, bank has to sell.

There are losses that spooks other depositors. There is a venture capital network, and the information spreads, and suddenly you have a run. What does that mean? That means $40 billion decided to leave the bank on a given day, single day. When that happens, no bank can really survive, okay?

And that's what really happened with Silicon Valley Bank, so here is a picture of how the deposits look like over the long horizon. In the gray, you see the deposits coming in, and then in the purple, you see the deposits leaving. So the gray is when slowly 140 billion came in, and the purple is in a big chunk the deposits left.

Okay, so this was Silicon Valley Bank, and hopefully, you kind of remember what happened. A lot of stuff has happened. This was March 10th, what happened then? Well, we decided that Silicon Valley Bank, even though it was not a systemic bank before, but it suddenly became systemic over the weekend.

We decided to wipe off some creditors and shareholders took the loss, but at the same time, we decided that uninsured depositors will now be insured because they are too spooky. We don't wanna spook them. We decided we are gonna give the bank some line of credit because they have these assets, their value is lower, so they might be spooked.

The bank managers might be spooked. So we're gonna give them bank term funding, which is, yeah, your assets are now below the par value. Let's say it was instead of 100, it is not $80 because the market value is lower, but we'll give you $100, don't worry. We know these assets will come back.

So you can take $100 and do your lending. So we'll do all that. The problem with all of this, and I'll come back to it towards the end, is that while this is good in the short run because it makes the markets calm and depositors come, the equity is underwater.

Once equity is underwater, bad stuff can happen. And we have enough evidence that bad stuff happens. And we call this future moral hazard. And when there is future moral hazard, nothing good comes out. So we'll talk about how we can sort of think about this in the long run, but I'll come back to it, okay?

So this is a pointer for later. All right, so that was the first part. So Silicon Valley Bank, now the story is gonna be very similar for other banks in the system. But how did we get here? Well, over the weekend of the 10th when Silicon Valley Bank was taken over and closed, we wanted to understand if Silicon Valley Bank was an outlier.

And by we, I mean me and my collaborators who were working on a study related to how much risk is there in the banking system. And over the weekend, we thought that, well, it would be a great idea to use the weekend to stress test the whole banking system because that's what professors do.

So we decided that we are gonna do something that probably Fed should have done, but why not? So we are gonna do what I showed you in the example with Silicon Valley Bank for all the banks, because we have all the data. So 4,800 banks will do the same thing.

So what I want to now do is to show you the same analysis that I showed you before, but instead of Silicon Valley Bank, for the entire banking system. So let's start. Here is, instead of showing you just Silicon Valley Bank, this is the entire US banking system.

On the asset side, you see similar kind of things that I showed you before, cash, securities, loans and so on. The entire US banking system is $24 trillion, okay? So not 200 billion now, 24 trillion. The liabilities, you can see again, there are deposits. The deposits are broken into insured and uninsured, which we'll get back to.

But there is about $9 trillion of insured and $9 trillion of uninsured, 50 50. And then there is that equity, which is important. It's $2 trillion, okay? So out of 24 trillion, 2 trillion is equity, something that I'll come back to again. So this is the starting point.

And then what we did was to ask the question that, hey, look, monetary tightening happened because of inflationary concerns. 2022 quarter 1, can we just do the same mark-to-market exercise for all the assets, all the banks, and then see how many banks are like Silicon Valley Bank? What made Silicon Valley Bank special?

So when we did that, the starting point was, okay, let's try to figure out what the unrealized mark-to-market losses there are in the banking system. So what I'm gonna do is first tell you in words, and then show you pictures, because my hope is one or both will stick.

So we'll see which one. All right, so here is the snapshot of what we find when you look at the entire banking system. When you do the same exercise, you see that there is a $2 trillion mark-to-market unrealized losses for the entire banking system, okay? And we can talk about how it's spread out between loans and securities and so on, okay?

So we did this. So one thing that is important here is that this number is much larger than what Fed and FDIC were doing because they were doing different kind of calculations. They were not looking at the entire loan portfolio, they were just looking at securities. We did this for everything that we had information on.

The other thing that's pretty important to remember is remember I showed you the equity. The size of equity in the US banking system was 2 trillion. This number looks very, very close to that. So that already tells you well, wow. Okay, so this number is pretty big relative to how much equity there is, because you are hearing chairman Powell and others argue that, there is enough capital in the system.

And it was like, okay, someone is not right here. Let's figure out what's going on. We also found that about 500 banks had losses which were worse than what Silicon Valley Bank suffered. Remember I showed you $20 billion of unrealized losses? There were 500 other banks which had worse losses.

So clearly, Silicon Valley Bank didn't just emerge as an outlier because it had a lot of losses. There were many others who had losses too. If you are interested in understanding what this 500 is, I'll show you. But this is what we call as there is turbulence in the system, okay?

It's not just one bank. It seems like there is turbulence in the system. All right, so what do I have in mind in terms of picture? Here is what I have, okay? So for those who are picture-minded, here is what you see on the x-axis is how much will your assets decline by in percent, and on the y-axis is the distribution.

Now, what's the vertical line? That's Silicon Valley Bank, if you remember, 20 billion on 200 billion. So that's about 10%. You can see that line, right, somewhere around that, approximately. But what I was trying to tell you was that there is this 500 banks who have more losses than Silicon Valley Bank, but they did not go under before Silicon Valley Bank.

So that was not it, that made Silicon Valley Bank special. Maybe there was a lot of hedging going on. Yeah, there is this interest rate risk. When there is threat of fire, an earthquake, we take insurance. Maybe a lot of banks must be taking insurance. Well, it turns out not so much, okay?

A simple way of capturing whether or not there is insurance going on is to think about, what are the duration? Remember I showed you the duration? Three years, four years, that's the extent to which my assets, my money is tied over. How many years is it tied over?

So if I am insured, that number goes down because I have insured it away. But if that number remains, that means I don't have hedging. So what is being plotted here is duration of banks that reported what was the duration of their investments net of hedging, okay? And there are two distributions here.

One is in orange, that's in 2021, and one is dark. That is in end of 2022, when the monetary tightening had happened. The key thing to see here is that this is not anywhere close to 0 for any bank that has reported it. Silicon Valley Bank, by the way, are those vertical lines.

And you can see that even before tightening, they had about three years of duration of assets. And in fact, during tightening, when you might think that interest rates risk is really, really high, they actually unhedged. And I'll come back to it again. So, four years, four and a half years, five years of duration of assets that we are talking about.

So, this turbulence, it's not hedged, so what sets Silicon Valley Bank apart? Maybe it's that they didn't have enough equity, remember this picture that I showed you? I mean, I had everything in 200 billion of dollars, but let's make it simple. This is a typical bank in the US, $100 of loans or assets, and then a bunch of liabilities from deposits and equity.

So, equity is pretty low, maybe that was the problem, that there were losses and equity was not there to absorb, that's why the depositors got spooked. Well, it turns out that if you look at how many banks had worse equity than Silicon Valley Bank, we're talking about 400 banks before the tightening, and 500 banks after tightening.

So, it's not like that Silicon Valley Bank really had low equity relative to any other bank in the system. In fact, equity doesn't set SVB apart, all the banks are pretty highly leveraged. If you're interested in a picture, again, here is a picture of what I was telling you.

Equity over assets, plotted here right now, in the orange is before tightening, in the black is after tightening. Why is the black shifted to the left? Because the position of the bank became worse, so that's what it is showing you. What are the vertical lines in each of these distributions telling you, that's where Silicon Valley Bank is.

And if you look to the left of Silicon Valley Bank in both distributions, there is a bunch of banks there. So, Silicon Valley Bank was not the worst when it came to how little equity it had. Yeah, it had less equity, but so were 400, 500 banks. So, what was it, well, this is where we came in with this idea which we were studying for some time.

This was actually what we were interested in before the crisis happened, the turbulence happened, which is, let's look at this notion of uninsured leverage. So, what's the idea, here is the picture again, if you have uninsured deposits, they behave in a very interesting manner, which is they're not insured.

So, they have incentives to monitor all the time, they're spooked all the time, and they have incentives to run whenever there is turbulence in the bank. Because if FDIC comes in, because FDIC is only protecting the insured depositors, they go back in the queue. Before FDIC comes in, they can run as soon as they want, and so they are ahead in the queue.

But once FDIC comes in and closes the bank, uninsured depositors go back in the queue, so their incentives are always to run, so it's an important idea. So, uninsured leverage the way you want to think about it is if you look at that uninsured deposits to total liability, you're talking about 90% in this simple figure here.

So, Silicon Valley Bank, if you looked at uninsured leverage, had more uninsured leverage than pretty much every bank in the US, okay? That did make it an outlier, and uninsured deposits is what we call the flight risk. All right, so if you want to look at it in the picture, here is again that distribution of uninsured leverage on the x-axis and the vertical line of Silicon Valley Bank.

And now you can see there is hardly any bank on the right hand side, that's what made it an outlier. So, then we wanted to understand, okay, we know that Silicon Valley Bank on uninsured leverage is an outlier. But how do we want to think about when does a run happen?

Runs didn't happen elsewhere, so how do we want to think about it? And what we did then in the study was to combine the notion of flight risk, which is uninsured leverage, how much spooky depositors you have, with turbulence, okay? Why because, why not, right, so here is the entire banking system, each dot here is a bank.

Size of the dot is telling you what the asset side of the bank is before monetary tightening. So, if you see Silicon Valley Bank, that black dot, that's $200 billion. So, when you see a pretty big orange dot in the middle, that's a trillion dollar bank that we're talking about, and so on, okay?

So, what are the two axes here, on the y-axis is turbulence. When you have mark to market losses, are they low or high? And on the x-axis is flight risk, which is how much uninsured leverage do you have, low or high? Why is this important, because when you cross the two, that's when you get to Silicon Valley Bank.

And we thought that well, we need to cross the two to understand how many other banks are at the risk of a run. And when we did this, we basically had to come up with a new notion of run because all of you have heard about bank runs, maybe.

Some of you might have heard about the Nobel Prize that happened last year. It was given to someone, Josh, and my former colleague from University of Chicago, Doug Diamond, and other co-authors, and they sort of laid out how to think about a run, and the bank run, and what policies emerged came from that line of work.

And the runs there are all about illiquid assets, that you have banks, banks invest in productive opportunities which sometimes become illiquid. And when that happens, if depositors get spooked, we might have runs. Had we held on to the long time period and somehow taken care of the spooky depositors, there were productive opportunities that bank had invested in and there wouldn't be a welfare loss, and hence came in FDIC insurance and so on and so forth.

But this is a little bit of different idea, assets that the banks have invested in are solvent, are not illiquid assets. These are the most liquid assets, these are treasuries, these are mortgage backed securities. You can't call them as illiquid assets, their value is known in the market immediately.

So, this is not the same idea of a run that was sort of worked on for many, many years and policies that have come from this. So, what's the idea here, well, when you combine turbulence and flight risk, what you get is a solvency run. What's the idea that if the interest rates are high and that leads to the asset value going down, if enough uninsured depositors get spooked, they could lead to the bank having to sell their assets and a different kind of run might emerge, which is self fulfilling.

What is important for this run, how much equity capital do you have? If you have less equity capital, more people get spooked, what else is important? How many spooky depositors you have, what else is important? What do these spooky depositors believe about other spooky depositors? If I believe a lot of spooky depositors are gonna line up, then I can end up in a situation where insolvent banks actually can go under.

And that's what we are sort of interested in, to figure out how many banks are out there. When we do this exercise, we come up with 200 banks, that dotted square that you see, 200 banks are potentially additionally insolvent. Till very recently, the Fed and the administration has still been calling these crises idiosyncratic.

Idiosyncratic means SVB, that's it, then it became first republic, that's it. Then it became Pacquest, that's it, but all of those, we don't have names here, because we didn't want to accentuate a run, but all of those names are in there, okay, so this is not a mystery.

If you take the notion of solvency run, another way to understand the whole picture is the following, that how much insolvency is in the system? How really can we tell, how many self fulfilling runs can happen? The simplest transparent way of thinking about it is on the x-axis here is, how many uninsured spooky depositors are going to run, because that's what other spooky depositors are thinking about.

When this number is high enough, you're gonna see a lot of failures. So, what's plotted is number of banks that will fail if a scenario is realized. What scenario, well, 10% depositors are withdrawing, 20%, 30% and so on. Now, you might think, what are these numbers, Silicon Valley Bank had 40%.

So, this is not crazy to think that uninsured depositors will withdraw this much amount of money. So, when I told you that about 200 banks are at risk, that's the yellow bar there. If half the uninsured depositors decided to leave some banks, we would have a run in those banks, and those banks would lead to another $300 billion of assets of banking disappearing very, very quickly.