Why Silicon Valley Bank Failed

What Startups Should do to Reduce their Risk

For many of us who worked on Wall Street in the 2000s, the memory of the 2008 financial crisis remains vivid. Since last week’s collapse of Silicon Valley Bank (as well as those of Silvergate and Signature Bank) I have been asked numerous times how it compares to the events of ‘08, why SVB failed, and what actions company leaders should take to manage the risk of bank failure.

The events of ‘08 were caused by a combination of poor underwriting standards and excessive leverage. When the housing bubble burst, the value of certain asset-backed and mortgage-backed securities (MBS) plummeted, pummeling those who were long in those investments. Although SVB could be thought of as a type of proxy for the early-stage tech sector, the ramifications of its failure should have far less of a domino effect than did Lehman Brothers’ failure in ‘08.

The failure of SVB was the result of risk management malpractice by the bank compounded by herd mentality in the tech sector. SVB’s customer deposits more than tripled from around $50 billion at the beginning of 2020 to over $180 billion by the end of 2022, the result of growth in the tech sector. The bank put those deposits to work largely in 10-year MBS that paid a premium to short-term Treasuries.

Despite clear signaling from the Fed that interest rates would increase, SVB’s management failed to hedge its interest rate risk and added to its asset-liability duration mismatch as it kept borrowing short-term from deposits while investing long-term in MBS. As interest rates rose, the value of SVB’s unhedged and interest-rate dependent investments fell. Moreover, SVB classified more than 75% of its bond portfolio as held-to-maturity (HTM) rather than as available-for-sale (AFS), which meant that capital requirements did not increase as the market value of the MBS fell.

Simultaneously, SVB’s stress testing was poor and did not accurately project the need its depositors would have for their cash. To meet withdrawal requests, SVB had to sell more than $20 billion in underwater assets and lost nearly $2 billion on the sale. The loss was a flag to the market, and attempts to raise equity and debt capital failed, in part because of botched public statements by SVB’s CEO. In the meantime, prominent venture capitalists began advising their portfolio companies to withdraw their deposits. Fears of SVB’s collapse spread like wildfire throughout its highly-concentrated customer base, and the bank was unable to meet withdrawal requests. By the next morning, regulators placed SVB under the control of the FDIC.

In the aftermath, and although depositors were bailed out, many startup founders are left wondering how they can protect their companies against a low-probability but high-consequence potential bank failure. The first step is to be an informed consumer and understand your bank’s risk profile. In retrospect, the reasons for the failures of SVB are obvious, but the bank’s customer concentration, high percentage of uninsured deposits, lack of interest rate hedging, and reach for yield could have been gleaned with some rudimentary digging and analysis. Second, consider diversifying banking relationships. For many early-stage companies with a limited amount of cash, utilizing a handful of banks and spreading deposits would be sufficient to have all of their deposits protected by FDIC insurance.

For larger companies with more cash it is not practical to keep $250,000 or less in each of tens or hundreds of banks. Such companies can consider utilizing a bank that participates in the Certificate of Deposit Account Registry Service (CDARS), which allows depositors to interface with a single bank but avoid having funds above the FDIC limits in any one bank. A less diversified option is to utilize a too-big-to-fail bank, whose depositors would likely be protected by a bailout in the event of failure. 

(to arrange interviews of Jason Searfoss, Chief Financial Officer of Boomtown, contact Tripp Baltz

Jason Searfoss
CFO, Co-founder