SVB was not a typical bank run

 

We know that SVB faced a run on the bank. But there are bank runs and there are bank runs. Was it typical bank run ala Diamond and Dybvig (DD), with illiquidity and sunspot equilibria? Or was it different?

While there were some aspects of the typical bank run – reliance on demand deposits, sudden shift in depositor sentiment – there are important aspects in which it was not a typical bank run. To see that, notice that typical bank run of DD style features one key aspect: inability to liquidate assets without a penalty, and hence mismatch between liquidity of assets and liabilities.

(Sure, it might seem that liquidating assets by the SVB would be costly to the bank. But the reality is that it would not carry a penalty cost, but rather that it would be associated with realizing the losses which have already occurred in the real world.)

The key point is this: In contrast to DD scenario, the asset side of the SVB was extremely liquid, mostly in form of government bonds and government-backed securities. While selling those in bulk would carry some penalty, this would be small – and if the market reaction after the collapse is anything to go buy, plausibly the prices would have even moved in the favorable direction. So the SVB case did not feature the key aspect of DD bank run, the liquidity mis-match.

It was also not similar to the bank failures in 2008. Back then, the problem was still liquidity mismatch: while the assets were securities and not loans, and hence were partially liquid, they were risky assets and selling them in bulk was not an option: due to classical finance problems (information asymmetries, limited arbitrage capacity) the large-scale selling would lead to large penalty. This is also why a regulatory  environment requiring large amount of liquid assets would be irrelevant: SVB did hold large amount of liquid assets.

Therefore, the problem of SVB was more about insolvency rather than illiquidity. True, the insolvency was of unusual type – it was not credit or loan losses, but losses related to interest rate risk, something we typically do not think as major source of solvency risk; and if the bank would not be run on, it would probably turn around and become solvent. But irrespective, it was the current value of its assets, not its inability to quickly liquidate them, which was the problem. And at the same time the current market value of the assets was in now way related to liquidity problems, but simply reflecting different risk-free rates.

So no, this was not a bank run from the 1983 paper, and neither it was a bank run from the more modern fire sale literature.