This week we saw two interesting bank collapses: Silvergate Capital Corporation, and SVB Financial Group.
This is a reminder that diversification is important.
The most basic problem in both cases is that they got money from a rather undiverse set of depositors, who experienced unusually large fluctuations in their deposits and withdrawals. They also made overly large bets on the safety of government bonds.
As Ilmanen pointed out in 2021, low real interest rates created a need to choose between returns on safe investments of zero (i.e. becoming a bank that pays no interest on deposits), versus choosing higher than normal risks to pay nonzero rates to their depositors.
These two banks had a surge in withdrawals over the past year because their depositors had a bad year, much more so than would be the case for depositors in a more diverse set of industries. That meant that the long-term bonds that the banks had hoped to hold to maturity needed to be sold. Since long-term US government bond prices dropped over 30% in 2022, that meant the banks realized big losses on bonds they’d bought during the bond market bubble. Both banks had big inflows in 2021, so an unusually large fraction of their assets were bonds bought at bubble prices.
SVB in particular appears to have had a third of its assets in “held-to-maturity securities” (mostly bonds?) that it bought in 2021.
The book Fragile by Design says that US banks have been more fragile than banks in more advanced countries such as Canada because US banks used to be restricted to a single location. US banks have been allowed to grow to become national banks over the past few decades. That means many big banks are as diversified as Canadian banks. But the US still has vestiges of the old rules, in the form of banks that are localized enough to be dominated by one industry.0
Depositors have little reason to worry about this kind of bank failure if they diversify their deposits across multiple financial institutions, or have less than the FDIC-covered $250k in a bank. But some people probably depended too much on bank accounts that weren’t covered by FDIC insurance. Those people withdrew money in a panic last week, mainly due to the risk that their wealth would be in limbo for months. That’s the part that that made the collapses sudden.
And obviously investors who buy stock in banks should diversify across many banks. I paid some attention to SVB, and didn’t notice the extent of the risk. I found it much easier to spread my investments across 10 US banks and a few banks in other countries, than it would have been to analyze SVB carefully enough to detect the risks.
Wasn’t regulation supposed to prevent this kind of problem?
Part of the answer seems to be that the most widely used measures of whether a bank is well capitalized say that there’s no risk to buying US government bonds. That seems mostly due to the government’s desire to sell those bonds to banks, rather than any delusion that there was actually no risk of the kind of bond market crash we saw in 2022.
I think there are additional rules that say something like banks need to keep equity above something like 5% of total assets, regardless of how safe those assets are. But I don’t see a clear explanation of those rules. I don’t understand how they handle unrealized bond losses.
The liquidity coverage ratio requires banks to be able to handle 30 days worth of outflows. That sort of worked. SVB withstood many months of outflows before becoming desperate.
SVB’s balance sheet for September showed clearly that if it marked its investments to market, it would have had negative common stockholder equity (but still a cushion of more than 10% of deposits from preferred stock and borrowing). Its condition improved slightly by the end of the year, and bond prices have been stable since then. So regulators were able to see for about 5 months that SVB was in fragile shape. But apparently nobody figured out what to do about that.
Scott Sumner and John Cochrane have decent comments on the quality of the relevant regulators.
<Rant>
The financial system has been talking a lot about diversity recently. Yet the word diversity has been abused in ways that distract attention from the kinds of diversity the financial system needs.
Racial diversity has no relevance to banking crises.
Too much testosterone might have contributed to banking problems. But low testosterone levels for bank executives seems like a better goal than a diverse mix of testosterone levels.
</Rant>
Added 2023-03-19:
Arnold Kling knows a good deal more about banking rules than most economists. He clarifies what triggered SVB’s desperate attempt to get more investment was that the alternative was to sell some Held to Maturity (HTM) assets to pay depositors. SVB likely had over $76 billion in such assets, but selling any would require them to admit to around $15 billion in losses on those assets. Rules for banking capital adequacy would then recognize what well-designed rules would have said 6+ months earlier: SVB was unreasonably close to being insolvent.
I object to Kling calling HTM an accounting scam. There are likely situations where HTM is appropriate. Accounting rules make it quite clear that banks shouldn’t use those assets to meet capital requirements. Accounting rules should not be blamed for the mistakes of banks and bank regulators.
Kling also pointed me to Alf’s comparison to Europe. Bank regulations there require interest rate risk stress testing that SVB would likely have failed in late 2021. US regulations have no such test. Some Democrats have been blaming Trump’s deregulation of SVB-sized banks. Those Democrats are clueless about what pre-Trump regulation did. Our current mess stems more from a bipartisan refusal to imitate the practices of countries with wiser banking rules.