U.S. bank run data challenge case for panic-driven failures

Traders Union
All News
Financial News
U.S. bank run data case

Highlights

Database of 3,984 U.S. bank runs (1863-1934) shows depositors usually target weak banks, with average receivership recovery at 50 cents on the dollar.
Out of 3,984 cases, only 53 were non-fundamental runs driven by misinformation, and strong banks in the top fundamentals tercile saw zero failures from such runs.
Silicon Valley Bank’s $20 billion FDIC loss and 9% asset hit suggest insolvency predated its run, challenging self-fulfilling panic theories.

Database findings from historical bank runs

As reported by Financial Times, Professor Emil Verner, together with Sergio Correia of the Richmond Fed and Stephan Luck of the New York Fed, has built an interactive database tracking 3,984 runs on individual U.S. banks from 1863 to 1934, with additional mapping of bank runs and suspensions stretching back to 1800.

The project draws on 380 million newspaper reports scanned through the Library of Congress's Chronicling America archive. Researchers narrowed those records to about 800,000 articles through keyword searches and then used LLMs to identify reports describing actual bank runs and match them to individual banks.

The research suggests the long-standing view that runs destroy otherwise solvent banks is overstated. Depositors appear, in most cases, to identify weak institutions accurately, and recovery values for failed banks indicate that lenders unable to survive a run are often already deeply impaired, with receivers collecting about 50 cents on the dollar of asset book value on average.

Implications for banking theory and modern failures

The study also finds most bank runs do not end in bank failure. When researchers compare bank strength with local economic outcomes, they conclude that runs on strong banks do not meaningfully damage local economies, while runs on banks with poor fundamentals are followed by sharp lending contractions and broader spillover effects.

The authors identify only 53 apparently non-fundamental runs, driven by misinformation or depositor confusion, out of the 3,984 cases. Six of those banks failed, but the researchers argue that strong banks in the upper tercile of fundamentals faced zero probability of failure even when hit by such runs, and at least one failed case involved fraud uncovered during the run itself.

Verner also argues the collapse of Silicon Valley Bank does not fit neatly into the idea of a purely self-fulfilling panic. He says the roughly $20 billion loss absorbed by the FDIC, despite representing 9% of pre-run assets and only a few billion dollars more than the bank's equity capital, supports the view that the lender was effectively insolvent when the run occurred.

The findings sit awkwardly alongside the influential theory of self-fulfilling bank runs developed by Douglas Diamond and Philip Dybvig, whose work underpins much of the logic for deposit insurance. Instead, the historical evidence presented here suggests that, in most cases, bank runs act more as a market signal of underlying weakness than as the primary cause of collapse.

Our earlier article on Fitch rating actions on SoFi Bank National Association explained how the review linked the bank’s credit profile to operating performance and a changing macro environment for lenders. We noted that the reassessment highlighted heightened scrutiny of balance-sheet strength, borrower credit trends, and broader credit conditions as investors gauge bank resilience.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.