Laurence Olivier: “Is it safe?”
Dustin Hoffmann: “Yes, it’s safe: so safe you wouldn’t believe it.”
Laurence Olivier “Is it safe?”
Dustin Hoffmann “No it’s not safe. It’s very dangerous, be careful.”
Marathon Man (1976)
Less than seven weeks ago, amid a general mood of optimism that the worst was over for the technology industry, with expectations that the Federal Reserve was nearly done raising rates, Silicon Valley Bank shares rose 17% as it reported its full-year 2022 results. One bulge bracket bank announced that the company was “on the mend”; another suggested its clients start “aggressively accumulating SIVB shares.”
As a short-selling sceptic, I felt cheated. Amid the back-slapping bonhomie on the analyst conference call, no one asked about the elephant in the room: how big were the unrealized losses on their bond portfolio and how stable was their deposit base, which might allow the bank to hold the assets to maturity and so not be forced to make fire-sales at losses, which would put a big hole in its balance sheet?
Silicon Valley Bank helped fast growing private companies in California raise capital, then insisted those companies hold their cash deposits on its balance sheet, often paying no interest. In the blow-off technology top between 2018 and 2011, the bank grew exponentially as its deposits increased from just $52bn in 2018 to $190bn by 2021. With all this free money, instead of making customer loans, the bank entered what it believed to be a risk-free carry trade, investing in a securities portfolio of collateralized mortgages and US government bonds, with little default risk but which had in fact locked in a measly average yield of just 1.6%, at the top of the bond market. Whilst these assets were generally long-term in maturity, they were funded by deposits which could leave at any moment. This duration mismatch – which it never sought to hedge – would directly lead to the banks’ demise.
Management throughout were totally oblivious to interest rate – as opposed to credit – risk. In its results presentation in January, the bank claimed theirs was an example of a “high-quality balance sheet” with only 43% of its assets in customer loans (the lowest in its peer group) (See: Fig 1: SIVB Balance Sheet).1 But on March 1st, the company released its 10-K disclosure to the SEC, an important document for analysts, since it gives far more detailed disclosure than an Annual Report. This revealed $15.2bn of unrealized losses, on its securities portfolio. Although the bank was the 16th biggest by assets in the United States, this mark-to-market loss compared to tangible equity risk capital of just $11.8bn. In other words, Silicon Valley Bank was already insolvent.
Banks are allowed to account for security portfolio assets as “hold to maturity” and not mark them to market, so none of these losses were yet reflected in its reported results. If Silicon Valley Bank could continue to fund the assets, it could hold them to maturity without taking a hit on its P&L or shareholders’ capital. But their clients were burning cash and the bank couldn’t afford to be competitive in bidding for new deposits because these were likely at a higher cost than the yield on their securities portfolio. They were also already maxed out on their borrowing from their lender of last resort, the San Francisco Federal Home Loan Bank, and had already pledged much of their high-quality collateral to secure this funding. The bank was slowly running out of cash.
Then suddenly the dam burst. On the evening of Wednesday 8th March, Silicon Valley Bank announced that it had sold $21bn of held-to-maturity Treasury and Agency securities to raise liquidity. This crystallized some $1.8bn of their mark-to-market loss on its securities portfolio but also meant they needed to raise new capital (proposed $2.25bn) to plug the realized hole in their balance sheet. However, since investors now knew there was also a potential $13bn of further unrealized losses, enthusiasm for committing to new capital was limited. The capital raise would fail.
The next day – as its share price fell 60% - there was a run on the bank by its previously loyal depositors, most of whom were still earning no interest. Because the Federal deposit insurance scheme covered just the first $250,000 of only domestic depositors, 97% of its deposit base were highly incentivized to cash out as soon as possible, leading to a reported $42bn of withdrawals. Whilst replacing $2bn of tangible equity would have been tricky, no bank could survive losing a quarter of its deposit base in a single day. Unable to meet further withdrawals, the bank officially failed the next day, brought down not by short sellers but by panic amongst its Silicon Valley venture capital customer base.
There are a wide range of outcomes in terms of what happens next. Our estimate of Silicon Valley Bank’s pro-forma balance sheet suggest that not only will equity, preferred and debt holders be wiped out, but that uninsured depositors being made whole is dependent on the banks remaining assets being sold near par rather than at fire-sale discounts, though it has now been reported that the Federal government has overcome qualms about moral hazard to bail-out all depositors, returning cash to uninsured depositors immediately. This in our opinion together with easing of repo terms may ringfence the issue but there is still likely to be some contagion: banks with similar business models, mark-to-market losses, and unstable funding, such as Signature Bank, where Argonaut also has a significant short position, which also failed overnight.
However, we believe that the big banks and global financial institutions will emerge largely unscathed, since their securities portfolios are a smaller proportion of their balance sheet, they do not share the same funding issues, nor is this “silicon bust” yet a reflection of a deterioration in the credit cycle (though this might now happen). Nevertheless, the demise of Silicon Bank is a stark reminder that in this inflationary cycle investors should - like Lawrence Olivier in Marathon Man - be asking are “safe” assets really that “safe” after all?
1 Source: Company Presentation, Jan 2023