Bank runs: a historical perspective
Rustam Jamilov, Tobias König, Karsten Müller, and Farzad Saidi examine the historical occurrence and impact of bank runs from 1800 to 2023 using a comprehensive dataset covering 184 countries. Their dataset combines narrative chronologies with statistical indicators of bank deposit withdrawals to identify bank runs. In addition, bank runs are classified as systemic if they are accompanied by a measurable contraction in the banking sector’s total deposit base.
Key Findings
- Likelihood of Bank Runs: The unconditional likelihood of a bank run occurring is found to be 1.2%, while significant deposit withdrawals occur in 12.7% of cases. Systemic bank runs, which involve substantial deposit withdrawals, lead to more severe economic consequences compared to non-systemic runs or deposit contractions alone.
- Economic Impact: Systemic bank runs are associated with significant output losses, with real GDP being, on average, 9.5% lower five years after such an event. This highlights the severe macroeconomic costs of systemic bank runs compared to non-systemic runs or mere deposit contractions.
- Triggers and Effects: Bank runs can occur even when banks are well-capitalized and not facing fundamental crises. Highly leveraged banks are more prone to runs, leading to credit crunches and reallocation of deposits across banks. Liability guarantees can mitigate output losses post-systemic runs, and the presence of a lender of last resort or deposit insurance reduces the likelihood of a run becoming systemic.
- Historical Patterns: The authors find that systemic bank runs have historically caused substantial economic downturns, even when not accompanied by broader banking crises or failures. The presence of a central bank and deposit insurance schemes are shown to lower the probability of a bank run turning systemic, especially before the 2007-08 Global Financial Crisis.
Two centuries of systemic bank runs
Authors: Rustam Jamilov, Tobias Konig, Karsten Müller, Farzad Saidi
From: University of Oxford, University of Bonn, National University of Singapore
Real-time analysis of a bank run
Marco Cipriani, Thomas M. Eisenbach, and Anna Kovner provide an in-depth analysis of bank runs that occurred in March 2023 in the U.S., using high-frequency interbank payments data. The authors use confidential data on wholesale and retail payments, which are available intraday at the transaction level for all U.S. banks. The primary data sources are interbank transfers through the Fedwire Funds Service and the Automated Clearing House (ACH), both operated by the Federal Reserve. A bank is considered to have experienced a run if it has significant net liquidity outflows exceeding five standard deviations of its historical net outflows during the period from March 9 to March 14, 2023. The methodology includes an analysis of the role of public signals, such as stock price movements and public announcements, in coordinating depositor behavior.
Key Findings
- Identification of Bank Runs: The study identified 22 banks that experienced significant deposit outflows during the March 2023 bank runs. This number is much higher than the two banks that failed (Silicon Valley Bank and Signature Bank) but lower than the number of banks that saw large stock price declines.
- Nature of the Runs: The runs were primarily driven by large institutional depositors rather than retail depositors. This was evidenced by the high value of payments rather than the number of transactions, indicating large withdrawals by a few depositors.
- Role of Public Signals: Publicly traded banks were more likely to experience runs, suggesting that public signals such as stock prices played a role in coordinating depositor behavior. However, the relationship between stock returns and liquidity outflows was not perfect, as some banks with significant stock declines did not experience runs.
- Bank Responses to Runs: Banks that survived the runs did so by borrowing new funds, primarily from Federal Home Loan Banks (FHLBs), rather than selling liquid securities. This borrowing strategy allowed them to maintain liquidity without realizing losses on securities.
- Impact of Public Announcements: The timing of runs was influenced by public announcements, such as the closure of SVB and the subsequent government announcement guaranteeing all deposits at SVB and Signature Bank. The runs were concentrated in the afternoon following these announcements, indicating the impact of public information on depositor behavior.
- Fundamentals vs. Coordination: While banks with weaker fundamentals (e.g., higher uninsured deposits and unrealized losses) were more likely to experience runs, many banks with similar or worse fundamentals did not. This suggests that coordination among depositors, possibly influenced by public signals, played a significant role in determining which banks were run.
Tracing Bank Runs in Real Time
Authors: Marco Cipriani, Thomas M. Eisenbach, Anna Kovner
From: Federal Reserve Bank of New York
