THE ORIGINAL SIN OF SILICON VALLEY BANK
How one decision triggered a bank run.
You can trace the mayhem in banking right now back to something akin to a modeling error.
The ground for the error was laid in early 2020, when Silicon Valley Bank was overtaken by a deluge of deposits flooding in from the government’s response to the coronavirus pandemic. The bank’s deposits tripled over the next two years, climbing from $61 billion at the beginning of 2020 up to $189 billion by the end of 2022.
Silicon Valley’s executives faced three options. They could have left the windfall in cash. They could have invested it in securities. Or they could have used it to buy or make more loans.
The first and third options weren’t viable. The first offered liquidity but would require the bank to absorb the cost of servicing over $100 billion worth of deposits without any offsetting income, leaving Silicon Valley’s executives vulnerable to criticism from analysts. The third option would produce income, by contrast, though it would come at the expense of intolerable credit risk. This left the second option, which promised a modest amount of income and liquidity yet no added credit risk.
The only outstanding issue was which securities to buy. This is less of a subjective analysis than mechanistic. "There is one financial commandment that cannot be violated,” preaches Jamie Dimon. “Do not borrow short to invest long — particularly against illiquid, long-term assets." A bank’s goal, to Dimon’s point, is to match the duration of its assets to the duration of its liabilities. If a bank has $10 million in five-year certificates of deposit, it’s safe to invest $10 million in five-year bonds. But if a bank’s liabilities are all short-term, then its assets should mirror that.
In this case, Silicon Valley’s asset-liability models seem to have convinced the bank to assume more duration risk than appropriate. Two-thirds of its deposits, or $125 billion worth, were held in non-interest-bearing accounts. By assuming these deposits behaved like ordinary transactional deposits, then they could be paired up against assets that reached a decade or more into the future.
Therein lies the original sin of Silicon Valley, for these deposits weren’t ordinary. From 2019 through 2021, the average account size at Silicon Valley grew from $2.9 million to $4.9 million, stuffed to the gils with PPP payments and stimulus checks. No one knew how sticky the additional $2 million in each account would be, but there was no reason to assume it’d be as sticky as any other non-interest-bearing deposit under ordinary circumstances.
Under the assumption that this deluge of deposits would behave like ordinary deposits — with a duration of seven or more years — Silicon Valley amassed an enormous pile of mortgage-backed securities of like duration, thereby compounding the problem. Like other fixed-income securities, mortgage-backed securities decline in value as interest rates rise. But unlike other securities, the incremental adjustment gets worse the further rates climb, as the duration of the security extends further into the future as the underlying mortgages become less likely to be refinanced by the borrowers. In essence, Silicon Valley was betting the proverbial bank that interest rates would stay at or near zero percent, as they had since the beginning of the pandemic, and that they would stay there for another decade.
There aren’t many examples of banks that have failed by placing a directional bet on interest rates. The Birmingham-Bloomfield Bank, in Birmingham, Michigan, is one of the few that comes to mind. It placed a bet in the late 1960s that interest rates would stop climbing and reverse course by building a large position in long-term municipal bonds funded by purchased deposits. When interest rates continued climbing, the $110 million bank was locked into low-yielding securities and seized by regulators in February 1971, becoming the first $100+ million failure administered by the FDIC.
A more prominent example occurred a decade later with the downfall of First Pennsylvania Bank, a predecessor of the first private bank founded in the United States. First Penn had grown rapidly in the late 1960s and early 1970s, with assets climbing from $2.1 billion in 1967 to more than $6 billion by 1976. Beginning that year, the bank used short-term deposit liabilities to make large purchases of long-term, fixed-rate government securities that yielded seven to eight percent. It held more than $1 billion in such bonds within three years. But interest rates kept climbing. By the first quarter of 1980, the market value of the bonds had declined by $300 million, equating to virtually all of First Penn’s equity, and the bank was stuck paying short-term rates of 15.5 percent to fund $1.2 billion of fixed-rate securities that earned just 8.7 percent.
A crisis of confidence ensued. Regional banks and deposit brokers that were customers of First Penn began withdrawing their deposits en masse, forcing it to the discount window of the Federal Reserve. It was stuck. It needed to increase liquidity by selling securities, but if it sold securities it would have to recognize a loss and thereby impair its capital. Ultimately, the government made the decision for the bank on May 29, 1980, arranging a bailout of First Penn, which, years later, would be rolled into Wells Fargo by way of Wachovia.
These examples if heeded would have discouraged Silicon Valley from placing such a large bet on interest rates. Yet, what’s particularly perplexing in this case is that Silicon Valley needn’t have consulted the annals of history for counsel against such risk.
Here’s Jamie Dimon on October 13, 2020, responding on JPMorgan Chase’s quarterly conference call to a question from Ken Usdin, an analyst at Jefferies Securities:
We're not going to do anything to protect our net interest income. We have $300 billion of cash we can invest today, and that becomes $400 billion. We're not going to invest in stuff making 50, 60 or 70 basis points so we get a teeny little bit more net interest income. We're going to make long-term decisions for the company. And if our net interest income in the end gets squeezed a little bit, so be it. But we don't want to be in a position where we lose a lot of money because we made investments in five- or ten-year securities which lose a lot if rates go up.
Rene Jones, the chairman and chief executive officer of M&T Bank, was even more prescient than Dimon, writing in his 2021 message to shareholders about the folly of undue duration risk:
A core operating tenet at M&T has always been to avoid reaching beyond our purpose and taking on too much risk, which can be in the form of credit risk from aggressive growth in loans or from fluctuations in interest rates. With a lack of loan demand during the year, many peers chose to invest a greater proportion of their excess cash into investment securities. It is notable that during the year, we chose to avoid following suit given the historically low rates of interest that did not seem to compensate us for the risk that rates might rise in the future. In essence, we decided it was better to hold our fire.
A hypothetical $10 billion invested in a three-year U.S. Treasury bond at the start of 2021, Jones went on to explain, would have earned $3 million in incremental interest income during the year but then suffered a $234 million decline in market value as rates came off their lows.
Jones doubled down on this a few months later at a conference in his answer to an analyst’s question about protecting M&T’s net interest margin:
You should do this. Take a twenty-year chart of the two-year, the three-year, the five-year, the fifteen-year and the thirty-year and look where we are now. Look how easy it would be for any of those rates to go up by 300 basis points and destroy tangible capital. And as we see rates move up, which they've been moving up, right? It reduces that risk, right? That's essentially what we're looking at. But there are certain places on the curve you just don't want to be because the pain could be too high.
None of these arguments resonated with Silicon Valley. Nine days after Dimon shot down the idea of protecting JPMorgan’s net interest income by exchanging cash for securities, here’s how Silicon Valley’s chief financial officer, Daniel Beck, answered the same question:
Our treasurer and his team go home tired every day. They've been working hard to put that money to work, and you saw in the quarter that we put close to $10 billion worth of money to work in investment securities. The bucket continues to be refilled with additional liquidity and we plan on continuing to deploy it.
And at about the same time that Jones’ letter came out, Beck had this to say on a conference call with analysts:
We've seen incredible liquidity growth coming through the end of last year and continuing through the quarter. Based on our view of that liquidity and the overall market conditions and how much incremental liquidity remains in the market, we are comfortable being able to put some of that money to work in the longer-duration, held-to-maturity category — think in the three to five-year range . . . Based on the trends that we continue to see and the stability of the liquidity in the market, we're comfortable putting that type of money to work . . . in a mix of agency mortgage backed securities as well as agency CMOs in that three- to five-year maturity range.
If Jones and Dimon, and others, could see the risk so clearly, why couldn’t Beck? It’s not that Silicon Valley performed poorly relative to peers; its performance was downright dastardly. A recent analysis by the Financial Times showed that if banks had to mark to market their securities portfolios today, Silicon Valley would be the only one to wipe out all of its tier one common capital.
Certainly greed and hubris play a role in this, buoyed by the phenomenal ascent of Silicon Valley’s stock in the years leading up to its failure. But more than anything it reminds us of a central premonition about banking. “You can afford to run much less risk in banking than in commerce and you must take greater precautions,” wrote Walter Bagehot in Lombard Street, the proverbial bible of central banking, “because a banker, dealing with the money of others, and money payable on demand, must be always, as it were, looking behind him and seeing if payment should be asked for.”
There is no margin for error in banking. And like Adam eating the forbidden fruit, the punishment can be eternal.
Maxfield on Banks is a reader-supported publication. To receive new posts and support my work and research, consider becoming a free or paid subscriber.