If you’ve been living under a rock, or head down in the weeds of corporate credit, you might not have seen the trouble that SVB Financial — Silicon Valley Bank, to its friends and the legions of venture capitalists and startups founders which have used it over the past 40 years — is in.

But it looks to us like it’s mid-way through a death spiral.

The shares were down 30% on Wednesday, down 60% on Thursday, and a further 46% in the pre-market before Friday’s open. Anecdotally, depositors are running in size, a classic bank run the like of which the US hasn’t really seen since the global financial crisis.

The immediate trigger for the slide was the bank’s announcement on Wednesday that it had sold $21bn of high quality securities, booking a $1.8bn loss in the process, and that it was launching a capital raise for an additional $2.25bn of equity, with Goldman Sachs underwriting and General Atlantic agreeing to inject $500m at the market price.

At the time, the bank positioned this as a turnaround story which would be accretive to P&L over time, and said it would be reinvesting the proceeds of the asset sale (mostly long-dated bonds and guaranteed mortgage-backed securities) into short term government debt.

But two days is a long time in the middle of a bank run, and Twitter is awash with tales of venture capitalists pulling funds out of the bank and telling their portfolio companies to do the same.

A cursory glance at the balance sheet (provided as a helpful “mid-quarter update” by the bank on Wednesday) seems reassuring — it has among the lowest loan-to-deposit ratios of any large US bank, at 43%. That’s reassuring, because when depositors take out their money en masse, banks have to sell assets to raise the money. Selling loans is hard; they’re illiquid, harder to value, specialist exposures. Much better to sell government debt.

SVB has a great deal of government debt (some $91bn) locked away in the “held to maturity” book. This is all going to be money-good; it’s mostly US Treasuries or “agency MBS” (that is, mortgage-backed securities issued by Fannie Mae and Freddie Mac, which trade as a liquid rates product in the US).

The trouble is, it was all bought when interest rates were low and bond prices were high. Now interest rates have gone up, bond prices have gone down, and selling these assets will crystallise a large loss of around $15bn. That’s enough to wipe out nearly all of SVB’s accounting equity. That’s no good for any company, but particularly bad news for a bank; financial institutions need to maintain minimum capital ratios. Blow through these and the regulators get involved, and it won’t end well.

So there’s a real constraint on what SVB can actually sell to meet any deposit outflows. It can’t sell the loan book, and selling the “high quality” assets, which are extremely liquid and readily available, will cause a huge loss.

Stresses started to emerge on SVB because of the specialist client base the bank serves. It’s highly geared to the tech ecosystem, with much of its deposit base driven by tech firm capital raising. Startups will do a funding round, deposit the cash in SVB, rinse and repeat. Typically, though, these startups are not standalone profitable, so the cash balance tends to draw down over time, unless replenish with new capital raising activity. 2022 was a terrible year for all sorts of financial markets, and particularly for VC fundraising, so the deposit base wasn’t being topped up.

Fast forward to now, and the risks of this highly concentrated deposit base are also becoming clear. Traditional banks raise deposits from a mix of individuals and institutions, with individual deposits often covered by some kind of government insurance. In the US, the Federal Deposit Insurance Corporation will cover $250k of deposits; beyond that, most individuals will tend to invest their wealth in financial assets, so in practice pretty much every regular person’s bank account is protected.

Institutional deposits can come from many different sources, including corporate treasury cash management desks, other financial institutions, funds or indeed governments, but they are uninsured (so effectively unsecured creditors of banks) and generally move much much faster than retail accounts.

SVB’s deposit base is overwhelming uninsured institutional deposits, and furthermore, it’s highly concentrated in VC and start-up ecosystem accounts. Concerns about its financial strength have ripped through these communities over the past couple of days, and prompted mass withdrawals of cash (at least, as far as we can tell from Twitter and other reporting; there’s no real time data available to the casual observer). 

There’s a herd instinct to any bank run, which is particularly acute when the deposit base comes from a specific community; there are also important behavioural issues about the startup community in particular. A large, profitable corporation might spread its cash management over multiple financial institutions; if one institution is in trouble, or even freezes withdrawals, it’s painful but it’s not existential. Startups, though, need this cash to survive. Even a deposit freeze which eventually makes deposits whole could drastically shorten runways, or requiring slashing staff, growth expectations or similar.

It’s worth quickly winding back to the global financial crisis, and spending a bit of time on the regulatory aspects. In 2008, to simplify grossly, the problem for banks was that they were funding assets of uncertainty quality (subprime mortgage-backed securities) with short-term wholesale funding through repo and money markets. Worries about the quality of assets meant the short term funding wasn’t rolled, and the banks were in big trouble.

So the post-crisis regulatory environment focused on making sure the banks had good, transparent and liquid assets, and funded them with nice wholesome deposits, instead of flighty bond market or money market liquidity.

That results in a bank that looks, uh, pretty much like SVB — what could be safer than a bank with a big pile of Treasuries, funded mostly by deposits? 

(SVB does have $13bn or so of short term funding as well; we’d expect that the counterparties are not going to be rolling this over).

Bank regulators also launched vigorous stress-testing programmes, in which banks were asked to simulate the impact of various macroeconomic bad news on their portfolios. But these mostly focused on credit risk; will these loans go bad, to what extent will companies draw down their liquidity. Duration risk (price impact of interest rate moves on zero credit risk assets) was probably somewhat understudied!

SVB also reveals another weakness in the regulatory system. The main lever for regulators to cut down on short term funding at banks was to require banks to hold enough cash, easily saleable assets etc to cover 30 days of outflows — that is, banks generally have enough money to cover the withdrawal of short term funding and the removal of all “sight” deposits (which can be pulled out without notice).

That means banks had a strong incentive to e.g. issue 31 day or callable commercial paper, or, in the case of SVB, offer a lot of accounts with 30 day+ notice periods for withdrawal. But if a bank run starts, 30 days notice doesn’t help very much. The market will incorporate the information and rumour about withdrawals, and the bank will still be collapsing. 

But it does, crucially, buy a little time for management and/or regulators to come up with a fix. It’s not really 9fin’s speciality to try to call the market one way or another, but the obvious fix for SVB is a sale to one of the US megabanks.

The stock prices of pretty much all US financial institutions slid sharply on Wednesday, but the likes of Goldman Sachs, Morgan Stanley, JP Morgan and Bank of America generate substantial excess capital, and, despite relatively poor Q4 results, have signalled plans to return capital to investors this year. 

Recapitalising SVB would be expensive, but it doesn’t really need a full recap. The assets are of reasonable credit quality, it’s the liquidity that’s the problem, and a merger with a large institution would remove the threat of having to sell the overvalued Treasury portfolio — a big bank could simply sit on the portfolio and amortise the high price through to repayment. 

For a money center bank wanting to make a big move into the tech ecosystem (wanting to be first in line for capital raisings, and IPOs when they come back) SVB’s client list should be valuable, and such an announcement should stop the run in its tracks.