For some things Harvard suffices; this blog is for the rest.

Silicon Valley Bank, Dead of Maturity Mismatch

Unless you live under a rock – or, even more unlikely, have a job that has nothing to do with finance or startups – you already know that Silicon Valley Bank (SVB) imploded last week. In this post, I want to discuss in some detail why that happened. As always, there is a proximate and an ultimate cause. As usual, everyone seems to be focused on the proximate cause even though the ultimate cause is much more important and interesting.

In good Austrian economics tradition, I believe that the ultimate cause of SVB’s collapse was maturity transformation (MT), or more specifically maturity mismatch. It turns out, that Mises was correct when he wrote in 1912:

For the activity of the banks as negotiators of credit the golden rule holds, that an organic connection must be created between the credit transactions and the debit transactions. The credit that the bank grants must correspond quantitatively and qualitatively to the credit that it takes up. More exactly expressed, “The date on which the bank’s obligations fall due must not precede the date on which its corresponding claims can be realized.” Only thus can the danger of insolvency be avoided.

Fiat Banking

When you deposit money in a bank account, you are loaning the bank money. It’s a weird loan because it has zero maturity – that is, you can withdraw your money at any time. Usually, you don’t withdraw your money, in which case the loan is automatically rolled over.

Banks take these deposits and lent them out or invest them. Since long-term loans and bonds generate the most yield, banks borrow short (zero maturity deposit from you) and lend long (30-year loan to your neighbor). This magic trick is called maturity transformation – the financial equivalent of time travel.

Although less relevant to our discussion here, it’s important to note that in our fiat banking system, banks do not intermediate savings as much as they just create money. Fiat money itself originates as an ex-nihilo loan. As a result, there can be no sound accounting and banking because the money is not sound. It exists only as a loan. The bank does not take your money and lend it out to someone else, it simply creates new money and gives it to someone. Worse, the money you loaned the bank was created in exactly the same way. In The Creature From Jekyll Island, G. Edward Griffin tells a story about this admittedly confusing point (emphasizes mine):

Marriner Eccles was the Governor of the Federal Reserve System in 1941. On September 30 of that year, Eccles was asked to give testimony before the House Committee on Banking and Currency. The purpose of the hearing was to obtain information regarding the role of the Federal Reserve in creating conditions that led to the depression of the 1930s. Congressman Wright Patman, who was Chairman of that committee, asked how the Fed got the money to purchase two billion dollars worth of government bonds in 1933. This is the exchange that followed.

ECCLES: We created it.

PATMAN: Out of what?

ECCLES: Out of the right to issue credit money.

PATMAN: And there is nothing behind it, is there, except our government's credit?

ECCLES: That is what our money system is. If there were no debts in our money system, there wouldn't be any money.

Later, Griffin quotes Robert Hemphill, credit manager of the Federal Reserve in Atlanta, who wrote in 1936:

If all the bank loans were paid, no one could have a bank deposit, and there would not be a dollar of coin or currency in circulation. This is a staggering thought. We are completely dependent on the commercial banks. Someone has to borrow every dollar we have in circulation, cash, or credit. If the banks create ample synthetic money we are prosperous; if not, we starve. We are absolutely without a permanent money system. When one gets a complete grasp of the picture, the tragic absurdity of our hopeless situation is almost incredible — but there it is.

(I’m sorry if you thought you were going to learn some ordinary economics here at Unfashionable. We only sell the weird stuff.)

All that aside, the way a bank makes money in practice is that it pays, say, 0% interest on all deposits and invests or loans out that money long-term for, say, 2% interest. In this example, the bank makes a 2% profit a year.

This practice creates a big problem called maturity mismatch: when you want your money back now, it’s not in the bank because the bank lent it out for, say, 30 years.

Luckily the bank can sell its claims on the market. Imagine you deposit $100 in your checking account. The bank loans out that money to Bob for 5 years at 2% interest. Let’s say that his promise to repay the bank is worth, say, $100 today (note that it could be worth more due to the interest). Therefore, the balance sheet of the bank looks as follows: $100 in liabilities (the zero-maturity loan you gave the bank, aka your deposit) and $100 in assets (the promise of bob that he will repay the loan). Great, our bank is deemed solvent. If you want your deposit back, the bank sells Bob’s mortgage at the market price of $100 and gives you your money back. The maturity mismatch doesn’t seem to be a big problem at all.

There are two problems with this story.

First, the Fed might raise interest rates to, say, 5%. This does not change anything for Bob; he will still pay you $100 plus 2% interest over 5 years. But it does change the market value of bobs mortgage because now anyone can get 5% interest. Naturally, something that pays only 2% (it doesn’t matter whether that’s a loan or a bond) is now worth less. Matt Levine provides some crude but intuitive math:

a new market-rate loan would pay $5 of interest per year for 5 years ($25 total), while your loan pays $2 of interest for 5 years ($10 total), so it is worth about $15 less, so people would pay you about $85 for it, though you’d get fired at a bank for doing bond math like that.

The bank is now in trouble. Its liabilities are still $100 (your deposit), but its assets are only worth $85 (the new market price for Bob’s mortgage). The bank is insolvent. If you want your money back, you either get 85$ instead of $100, or you need to come back in 5 years when Bob has paid back the loan.

Nonetheless, it still seems maturities are not important if we are careful to always have more assets than liabilities. In other words, it seems Mises was wrong: The credit that the bank grants must only correspond quantitatively, not qualitatively

But there exists a second problem. The practice of continually valuing our assets at the market price is called mark-to-market. Valuing your assets like that assumes that if you wanted to sell, you can always sell all of them at the market price. That assumption is wrong. In our example, the assumption holds because we’re talking about a single mortgage. But imagine a bank that needs to sell millions of loans (or, more realistically, bonds). There is no way to know at what price the bank can sell them before they actually sell them. On any ordinary day, banks are able to sell millions of bonds so it seems strange to worry about being able to sell them when necessary. The issue is that the finance system is highly connected. On good days nobody needs to sell bonds because customers don’t need their money back, but on bad days every bank might need to sell some bonds. In other words, many factors affect all banks at the same time: it’s either a good or bad day for all banks. On the day everyone wants to sell, assets might be worth significantly less than when you mark to market them.

In other words, maturity transformation is bad. Borrowing short and lending long is an inherently risky business that leads to insolvency. There is no way to teleport money from the future to the present without risking your bank blowing up. (Creating money, as all banks do daily, is arguably even worse.)

Intuitively this checks out. It just doesn’t seem sensible to have the solvency of your bank depend on the assumption that someone else will buy your assets if your customers ask for their money back. This might work most of the time, but in the cases where it doesn’t, the failure is catastrophic. “Most of the time” just isn’t good enough here.

When such an unpredictable event happens, experts often speak of a “liquidity problem.” That framing entirely misses the point. The problem is not that there are no buyers in the market, but that there are no buyers in the market at that price– that is, the price you mark-to-market your assets on the balance sheet. But your assets were never “worth” that much in the first place, you just hoped they were. So when bankers and regulators speak of “injecting liquidity” they simply mean propping up the price by finding (read, creating) more buyers at a higher price.

In Summary, we can’t escape Mises: “The credit that the bank grants must correspond quantitatively and qualitatively to the credit that it takes up.”

Silicon Valley Bank

Now that we have a general theory of modern banking or more specifically maturity transformation, we can take a look at what happened to Silicon Valley Bank. The problem they ran into is that they didn’t know what to do with all the money they received as deposits during the bull market – especially in 2021. The startups that banked with them kept depositing more and more money because investors kept giving them more and more money, as SVB explained in their last 10-K:

Much of the recent deposit growth was driven by our clients across all segments obtaining liquidity through liquidity events, such as IPOs, secondary offerings, SPAC fundraising, venture capital investments, acquisitions and other fundraising activities—which during 2021 and early 2022 were at notably high levels.

SVB decided to buy mostly long-term treasury bonds and agency mortgage-backed securities. In other words, they did what all banks do and borrowed short and lent (invested) long. As a consequence, the asset side of their balance sheet was extremely sensitive to interest rates as we have discussed above.

Bonds that pay 2% might have been worth $100 before the interest rate hike. But afterward, people were able to buy bonds for $100 that pay, say, 5%, so your 2% bonds fall in market value. Note however that only the market value declined. If you are able to hold your old 2% interest bonds until maturity, they are still worth the $100 you paid for them because the cash flow has remained the same.

SVB, instead of continually marking its assets to market, used a neat accounting trick to label their bonds “held-to-maturity” which allowed them to value them at cost – the $100 in the example above. There are some reasons this is allowed, which you can read about here, but the practice overall still strikes me as crazy in a world where maturity mismatches are tolerated. Due to this hold-to-maturity trick, SVB’s bonds could lose a lot of value on the market without losing value on the asset side of their balance sheet.

However, the asset side of SVB’s business was not the only problem. They had another unique startup bank problem: namely, that their depositors were all startups. In a bull market, startups raise and deposit more and more cash while in a bear market they not only don’t deposit but due to their running costs continually have to withdraw. Robert Armstrong calls this:

a double sensitivity to higher interest rates. On the asset side of the balance sheet, higher rates decrease the value of those long-term debt securities. On the liability side, higher rates mean less money shoved at tech, and as such, a lower supply of cheap deposit funding.

In summary, what happened is that Silicon Valley Bank invested its deposits in long-term treasury bills and mortgage-backed securities that lost a lot of value when the Fed increased interest rates. This loss was not obvious on the balance sheet because they valued some of their assets at cost instead of at the market price. At the same time, due to the interest rate increase, startups were depositing less money while continuing to withdraw. As a consequence, SVB had to sell some of their hold-to-maturity assets, which meant they had to mark them down (once you sell one bond you have to mark-to-market all your assets). At this point, it became obvious that they were in trouble and everyone started to withdraw their money. Then they collapsed.

There are many parts of this story one could propose as a proximate cause. Some are suggesting that it’s the Fed’s fault because they hiked rates so aggressively even though they claimed in their 2020 guidance that they wouldn’t.

Others think it’s mainly the fault of the weird regulation that allowed SVB to value some of their assets at the hold-to-maturity value instead of continually having to mark-to-market them.

There are also those who argue that SVB taking on venture dept was the main reason it collapsed. And there are many other proposed causes, such as the apparently incompetent risk management at SVB that didn’t “hedge their huge interest rate risk.”

Unfortunately, these explanations are diagnoses of symptoms. As we discussed above, the ultimate cause of these collapses is maturity mismatch which can’t be fixed by more regulations. If we were, for example, to change the strange hold-to-maturity regulation and force banks to mark-to-market their assets daily, the situation would probably improve. Nonetheless, it won’t prevent banks from blowing up. The only way to create a stable banking system is to listen to Mises and fix the underlying problem: maturity mismatch. I wouldn’t hold my breath on that one, but luckily, Bitcoin fixes this.

Thanks to Allen Farrington for edits and contributions.