Forty Eight Hours: How Silicon Valley Bank Died

The Fastest Failure

On Wednesday, March 8, 2023, Silicon Valley Bank was a 209 billion dollar institution, banker to half the venture capital world, with a stock price that had made long term shareholders rich. By Friday, March 10, regulators had seized it. Forty eight hours, start to finish, the second largest bank failure in American history at the time and by far the fastest. This entry of the series is about how that happened, because the mechanics are a permanent lesson in how banks actually break.

A Balance Sheet Built for One Interest Rate

SVB\'s problem was not crypto, fraud, or bad loans. It was duration. During the 2020 and 2021 boom, covered earlier in this series, venture backed startups raised historic amounts of cash and parked it at SVB, whose deposits roughly tripled in two years. The bank had to put that money somewhere, and it bought long dated Treasury and mortgage backed securities yielding very little, because in 2021 everything yielded very little.

Then came 2022 and the fastest rate hiking cycle since Volcker. When interest rates rise, existing bonds fall in price, and long dated bonds fall hardest. That is duration risk. By late 2022 SVB was sitting on more than 15 billion dollars of unrealized losses on securities it had classified as held to maturity, an accounting bucket that lets a bank carry bonds at cost rather than market value on the theory it will never need to sell them. The losses were disclosed in footnotes, roughly equal to the bank\'s entire equity capital, visible to anyone who read the filings. Almost nobody did until it mattered.

The Two Day Run

The trigger came on March 8, when SVB announced it had sold 21 billion dollars of securities at a 1.8 billion dollar loss and would raise fresh capital to plug the hole. The announcement was meant to be reassuring. Instead it converted a footnote into a headline. Venture capital investors, a tightly networked group chat of an industry, spent March 9 advising their portfolio companies to move their money out immediately.

Here is the detail that makes SVB historic. Roughly 94 percent of its deposits were above the 250,000 dollar FDIC insurance cap, because its customers were companies holding payroll, not households holding savings. Deposit insurance exists precisely to make running on a bank irrational, and it did not apply to the people who banked at SVB. On Thursday, March 9, customers attempted to withdraw 42 billion dollars, about a quarter of the bank, in a single day, mostly through apps and wires rather than any line on a sidewalk. Regulators expected over 100 billion more to leave on Friday morning. There was nothing left to run on, and the FDIC took the bank into receivership before lunch.

The Weekend the System Blinked

The failure instantly raised a system wide question, if uninsured depositors at SVB take losses, which company keeps money at any midsize bank on Monday. Signature Bank was closed that Sunday. That evening, regulators invoked the systemic risk exception and guaranteed all deposits at both banks, insured or not, while the Federal Reserve created a new facility letting banks borrow against their underwater bonds at face value rather than market value. Shareholders and bondholders of the failed banks were wiped out, depositors were made whole.

The stress did not end there. First Republic Bank, facing the same uninsured deposit flight, limped along for weeks before being seized and sold to JPMorgan in May 2023, an even larger failure than SVB. Three of the four biggest bank failures in US history happened within eight weeks, and yet, remarkably, the panic stopped there. The backstop worked.

What It Should Teach You

Three lessons worth carrying. First, banks die of liabilities, not assets. SVB\'s bonds were money good if held to maturity, it was the deposits that could not wait. Watch funding concentration before loan quality. Second, accounting categories do not change economics. Held to maturity treatment hid the losses from the income statement, not from reality. Third, tail risks cluster around whoever grew fastest in the previous regime. SVB tripled in the era of free money and was therefore the single institution most exposed to its end. The fastest grower in any boom is the first name to check when the regime turns.

The Bottom Line

Silicon Valley Bank failed because it funded long term, low yielding bonds with the flightiest deposit base in banking, and because 2022\'s rate hikes turned that mismatch fatal. The run took two days, the rescue took a weekend, and the lessons, about duration, uninsured deposits, and smartphone speed panic, got written straight into how regulators and analysts look at every bank since. Of all the episodes in this series, this is the one where reading the footnotes literally would have let you see it coming.