If you paid any attention to collapse of Silicon Valley Bank, you likely heard about how risky the bank’s investments were and regulatory changes in 2018 that may have played a role in the bank’s collapse. Neither explanation quite made sense to me and I’ve been trying to piece it together ever since.

I’ve always understood government backed securities, i.e. bonds and treasury notes, to be considered an extremely safe/boring investment. They pay a fixed interest rate over a fixed period of time - you know exactly how much you put in up front, how much you end up with, and unless the US government allows itself to default there’s effectively zero risk. So when a bank fails and the explanation is that it held toxic government securities it catches my attention. When it happens to be a bank that is deeply ingrained in the very industry I’m a part of, I start to ask questions.

What we know so far

Silicon Valley Bank (SVB) got caught up in what can only be described as a modern day bank run. It was a bank run because they went insolvent after too many depositors withdrew cash all at once and modern because we’ve never seen a bank run fueled by a combination of social media, an extremely well capitalized and tight knit industry, and the speed with which digital bank transfers can be completed. Yes people did in fact line up at bank branches to try to withdraw their money, but almost all of it went out over the series of interconnected tubes we popularized in the 90s.

It’s been widely reported that SVB held a collection of risky government securities that, through a technicality, could be reported on corporate financials at face value even though they were millions of dollars under water. Basically, if those bonds were held to maturity they would get all their money back plus a small profit and finance laws allow the banks to mark the value of the securities at face value. Unfortunately for SVB they were running low on cash and needed to sell the securities. Because the Fed has increased interest rates dramatically over the last year, no one actually wants to buy old bonds that pay well under market rate and they weren’t worth anything close to face value.

The other issue often raised is a regulatory change in 2018. The Dodd-Frank Act was passed after the 2008 financial crisis in an attempt to better regulate banks that are, for better or worse, too big to fail. The important bit for SVB was a series of stress tests that banks with over $100 billion were required to pass a series of stress tests that were designed to see how a bank would weather different storms. One such test was a sharp increase in interest rates. This may have helped raise alarm bells earlier if the limit wasn’t raised in 2018 to only include banks with $250 billion in assets - SVB had around $200 billion shortly before it collapsed.

This can’t be the whole story though, right? Why to would a bank really hold onto assets that they know are losing more and more value as interest rates go up? Even if they took the Fed at it’s word a year ago when they didn’t expect to raise rates, a few rate hikes and you’d surely shore up your books. Well, it gets interesting when you come across a seemingly tiny rule change in 2013 that I have found only a few references to so far.

2013 - the Fed quietly changes the game

This one gets really in the weeds of how banking and finance works - I’m going to do my best to gloss over the details and hit the high points. This isn’t a full accounting of the change and I may miss a few things, bear with me! The Economist had a great article (archived link) that touches on the ideas below with more thorough explanation in case you’re interested.

Historically, banks weren’t able to go straight to the Federal Reserve to buy securities on a whim. The Fed may occasionally make a lot of securities available, but standard practice for banks involved what is called the repo market. One bank who has cash on hand and needs to invest it in something that is considered safe and liquid would buy government-backed securities from other banks - this is a repo transaction.

In 2013, the Fed was already a few years into near-zero interest rates and started to model what might happen when they begin to raise rates. The Fed can only changes the interest rate they charge to banks, what the Fed can’t directly change is the rates banks charge each other. The Fed began to worry that if they raised rates there would be a delay in the rates banks charged each other, removing some of the Fed’s control over the market recovery. They came up with a simple fix - if banks can buy securities directly from the Fed whenever they want then bank-to-bank interest rates will have to more closely follow the Fed’s rate. Why buy from a bank if the Fed pays you better rates?

This rule change created what was called a reverse-repo market, and may very well have been what really set SVB up for failure.

So what does that mean?

Well on the surface, that little rule change doesn’t seem to matter - all the Fed did was ensure they had a tighter control over rate increases. What could go wrong?

When one bank buys a bond from another bank on the repo market the money does in fact leave one bank but it doesn’t actually leave the banking system - commercial banks still collectively have the same level of liquidity (i.e. cash on hand).

When one bank buys a bond directly from the Fed in the reverse-repo market, that cash leaves the bank and is parked on the Fed’s books. That money has left the banking system - the total liquidity of commercial banks decreased.

Again, so what? Well, this may very well mean that SVB did exactly what the Fed wanted them to do. When the Fed started to raise interest rates, SVB considered buying securities directly from the Fed to get the best rates.

It just so happens in 2020 and 2021 the Fed was also busy printing money to the tune of more than $4 trillion dollars. All of that money ends up in banks, and because banks are bound by regulations that limit how much cash they can keep on hand they have to invest it somewhere.

These two factors combined, the Fed (at the request of Congress) gave banks trillions of dollars. The banks needed a safe place to put the money, government bonds. The Fed already changed the rules to ensure that in times of interest rate hikes, the Fed would likely certainly pay the best rates.

Can we blame SVB?

Sure the bank has to be on the hook at some level, but they were playing the game designed by the Fed. The Fed wanted banks to turn to them for securities in times of increasing interest rates. Federal bonds are supposed to be secure, safe investments. SVB may have even taken the Fed at its word when they continued to announce that printed $4 trillion dollars couldn’t cause inflation and that no interest rates were expected. Is that really their mistake though, trusting the Fed’s official announcements and predictions?

None of this is to say there was some grand conspiracy where the Federal Reserve was purposely trying to set banks up to fail. Any of the decisions and rules changes along the way seem reasonable in isolation. The question here is whether the bank really was as negligent as is often claimed, or was the Fed negligent in recognizing the potential risk of the changes they made along the way?

Should SVB have diversified its portfolio to cover the bad investments? Probably, but only if they had good reason to believe they would be running low on liquidity and be forced to sell those investments early.

Wrapping it up

I could go on and on about how all roads lead back to the Federal Reserve here. Suffice it to say there’s more to such a complex situation than the two big narratives that have gotten air time so far.

In the meantime, I’m still trying to answer the simple question of how all SVB deposits were made whole over a single weekend when the bank wasn’t sold or liquidated. As of the beginning of this year the Fed’s FDIC Insurance Fund only had $128.2 billion dollars in capital. The best estimates I’ve seen peg uninsured SVB deposits in the ballpark of $150 billion.

How the heck was that money put back in place given that the only public explanation was that tax payers wouldn’t be on the hook and banks would eventually pay out of pocket? Did they honestly just click a few buttons and re-enable account balances, or did they fund it from some other part of the Fed’s books? The answer is probably horribly mundane and boring but I’m really curious to know how such a massive depositor bailout was pulled off that quickly.