What Cracked a Bank in March 2023? Uninsured Deposits versus Hidden Bond Losses in the SVB Window
DOI:
https://doi.org/10.13021/jssr2026.5632Abstract
Banks become vulnerable when asset values fall or depositors rapidly withdraw funds. Identifying which source of vulnerability matters most is important for helping banks and regulators reduce the risk of future bank runs and failures. The collapse of Silicon Valley Bank in March 2023 raised an important question: did investors primarily react to banks’ unrealized losses on long-duration securities, or to their exposure to runs by uninsured depositors? We study 165 publicly traded U.S. banks and measure stock-price changes beyond overall market movements from March 8-13, 2023. Using 2022Q4 regulatory filings, we compare these returns with uninsured deposits as a share of assets and unrealized securities losses relative to equity. Because these results depend on accurately matching bank tickers, holding companies, regulatory identifiers, and subsidiaries, I manually validate the dataset using FFIEC records, investor-relations pages, and regulatory filings. Preliminary evidence suggests that some automated matches contain name or identifier errors, although their effects on the results are still being investigated. The average bank experienced a 14.1% abnormal stock-price decline. A one-standard-deviation increase in uninsured-deposit exposure was associated with an additional 4.2-percentage-point loss. In contrast, unrealized securities losses were not statistically significant and had an estimated effect about half as large. These findings suggest that, during the SVB crisis, investors viewed the risk of uninsured depositor runs as a more important source of bank vulnerability than unrealized securities losses.


