Thread by @paulgp

2023 was a crazy year -- remember how we had a bank run at Silicon Valley Bank that caused a banking sector collapse (over 20% decline in bank equities in a week!) and prompted a Federal Reserve facility intervention?

No? Well then I have the thread and working paper for you! 🧵

Following the collapse of SVB, there an immediate response in the stock market. Many banks other than SVB plunged (most notably First Republic Bank, who finally failed many weeks later). Overall, the bank sector corrected sharply downward.

There was, however, significant heterogeneity in this downward correction. In the first week, the decline was quite skewed, with smaller banks experiencing less of a decline. By May, more banks had followed, leading to a dispersed, symmetric (and negative distribution).

Contrast these declines to the cumulative returns from early 2022 to pre-run in 2023 -- a period when the Fed hiked rates significantly.

It turns out that the declines in 2022 are a significant predictor of declines in 2023 -- the market had priced risk into those banks that did the worst following the SVB run!

So what explains the heterogeneity in the cross-section of the returns? What are the bank char. that drove contagion? In 2022, what was the market focused on that predicted 2023 returns?

In a new paper with Choi and Yorulmazer, we document some answers.

https://t.co/PDjidAhXt6arxiv.org/pdf/2308.06642…

Our contribution is to use a large (200+) cross-section of publicly traded commercial banks and bank holding companies in the US to systematically examine what balance sheet characteristics explain the drops.

Data and code here: github.com/paulgp/bank_re…

Folks have pointed to some bank characteristics already in previous discussions (@JosephPolitano has had some very nice write-ups about this: ): uninsured deposit reliance and unrealized interest rate losses.

https://twitter.com/JosephPolitano/status/1655714652095913985

We find clear evidence that uninsured deposit reliance was priced very negatively following the SVB. This ties exactly in line with classic models of bank runs, as the uninsured depositors behave like a flighty source of funding for banks (unlike insured depositors).

Moreover, banks with unrealized interest rate losses (like First Republic Bank!) got slammed. Doubly so if they had large uninsured deposit bases. This is likely due to banks with deposit outflows needing to sell assets to deal with the outflow, triggering losses.

Notably, banks with cash (not securities!) were much better off, suggesting robustness to outflows was the necessary constraint.

A striking feature of this run was how concentrated the initial decline was in "super-regional" banks (between 50 billion to 1 trillion dollars in assets) in the first week following SVB.

This is in line with significant outflows happening at these banks: https://t.co/mOaSSFrtKLlibertystreeteconomics.newyorkfed.org/2023/05/bank-f…

However, by May this size criteria had disappeared:

Finally, given that 2022 returns predicted some of the decline, what features of banks' balance sheets (if any) were being priced?

It appears to be uninsured deposits and cash, but *not* unrealized losses, predict 2022 returns, and loadings shift when included in 2023 regs.

What's the takeaway?

For one, the bank characteristics that drove contagion last banking crisis (bad loans, wholesale funding and interbank connectivity) were not drivers this time (for wholesale funding spillovers in GFC, see my paper with Tanju):
https://t.co/AGW2GiMN69link.springer.com/article/10.100…

As Gary Gorton has said "Financial crises are bank runs. At root the problem is short-term debt (private money), which while an essential feature of market economies, is inherently vulnerable to runs in all its forms (not just demand deposits)."
nber.org/papers/w25891

In this case, it was the most boring possible version of private money that ran, rather than wholesale funding. But combined with a year's worth of interest rate hikes make bank capital much more tenuous.

We're hopeful that this paper can be a useful documentation of the SVB bank run that can inform research and policy work going forward!

Working Paper here:
Data and code here: https://t.co/OSnjaMj7hL

Comments and suggestions very welcome! https://t.co/6KUyzP6Cnnarxiv.org/abs/2308.06642
github.com/paulgp/bank_re…