Bank Runs: Meaning, Causes and Historical Examples

Direct answer. A bank run occurs when many depositors or other short-term funders demand payment faster than a bank can turn its assets into settlement money. Runs often target weak institutions, but sounder banks can also face them; whether a run ends in failure depends heavily on asset quality, capital, liquidity, and access to support. Deposit insurance changed retail incentives, while central-bank liquidity and resolution changed how stress is contained and losses are allocated. S001S064S172S173

The textbook image is a queue outside a branch. The underlying mechanism is broader: uninsured firms transfer balances online, money-market lenders refuse to renew funding, counterparties demand more collateral, or other banks stop accepting exposures. Technology changes the speed and visibility of the run, not the balance-sheet mismatch that makes one possible.

1 · ConcernLoss · rumor · rate shock · outage · fraud

↓

2ARetail or uninsured deposits leave

2BWholesale funding or collateral tightens

↓

3 · Liquidity pressureUse reserves · borrow · pledge or sell assets

↓

4ASupport + credible assetsOutflows stabilize

4BForced losses + weak capitalFailure or contagion

Possible responsesLiquidity · insurance · resolution · recapitalization · closure

What is a bank run?#

A bank run is a rush to withdraw deposits or other short-term funding because funders fear nonpayment, delayed access or being left behind by others. Ordinary withdrawals or a busy branch are not enough to establish a run. It can happen through electronic transfers without a queue, and not every customer needs to participate. S199 S193

The Diamond–Dybvig model explains how depositors' individually protective withdrawals can become mutually reinforcing under particular liquidity and contract assumptions. Historical evidence also links outcomes to bank fundamentals. Coordination problems and genuine asset losses can interact; the model is not proof that all runs are baseless panic. S199 S200 S172

“Run in banks,” “run of banks” and “banks runs” are commonly intended as variants of “run on a bank” or “bank runs,” not separate technical mechanisms. “Run to bank” is ambiguous: it may simply mean a quick banking errand. In a financial-crisis context, the likely meaning is defensive withdrawal. A run at one institution does not by itself establish a system-wide panic. S200 S192

What does a run on deposits mean?#

A run on deposits is a rapid defensive outflow of bank-account funding. Transferring a deposit to another bank removes funding from the first bank even if the customer never withdraws cash. Interbank settlement and customer deposit entries are different parts of that payment. S004S005S006

For example, suppose worried customers transfer $20 million from Bank A to Bank B. A loses deposits and must settle the outgoing payments; B receives deposits and settlement funds. In this simplified transfer, deposits move within the banking system rather than disappear. Converting deposits into currency is a different transaction. This is an accounting illustration, not historical data. S004S005S006

What happens during a bank run?#

The bank must meet withdrawals with available payment funds, new borrowing, incoming funds or asset sales. Selling long-term assets quickly can cause losses. Some banks survive; others suspend payments, obtain support or fail. A temporary interruption of access is different from ultimate loss. Deposit insurance applies to eligible products and institutions under the relevant rules, not to every investment. S200 S172 S190

Consider an illustrative bank with $20 million in immediately available funds, $90 million in longer-term assets, $100 million in deposits and $10 million in equity. Assets total $110 million and balance against deposits plus equity. A demand for $40 million now creates a $20 million liquidity gap. The bank must fund that gap or realize assets; the arithmetic alone does not establish insolvency or inevitable failure. This example assumes the stated asset values and ignores other liabilities and transaction costs.

What causes a bank run?#

A trigger changes creditors' willingness to wait. It may be information about losses, doubts about access or expectations that other funders will leave. The bank's underlying vulnerability is a separate question: concentrated uninsured deposits, weak assets and inadequate liquidity can make the same shock more dangerous. S199 S194

The Federal Reserve's SVB review connects its concentrated business model, uninsured funding and risk-management weaknesses with the March 2023 run. A depositor-level study by Iyer and Puri finds that insurance status, bank relationships and social networks affected withdrawals in the bank they studied; that result is not a universal rule for every depositor. S194 S196

Technology can accelerate outflows, but electronic withdrawals existed decades before smartphones. Rose's historical comparison stresses the funding base and connections between depositors as well as speed. A complete explanation separates the bank's condition, the information shock, withdrawal incentives and its response. S193

Why a bank can be vulnerable to a run#

Banks commonly fund long-dated or illiquid assets with liabilities that can leave sooner. A mortgage or business loan may be valuable if held and repaid over years; a transaction deposit promises payment now. That maturity and liquidity transformation is economically useful, but it creates a timing problem.

The underlying accounting is easier to see in the worked explanation of deposits, loans, and interbank settlement.

If withdrawals are modest, the bank can use cash, reserve balances, incoming payments, liquid securities, market borrowing, or central-bank facilities. If outflows overwhelm those sources, it may have to sell assets rapidly. Forced sales can realize losses, depress prices, weaken capital, and convince more creditors to leave.

Runs therefore sit at the boundary between liquidity and solvency:

• A solvent bank can fail if it cannot obtain cash or settlement balances in time.
• An insolvent bank may appear liquid temporarily while losses remain hidden or unrecognized.
• A run can reveal genuine weakness and then amplify it through forced action.

The clean story of a baseless panic attacking an otherwise perfect bank is not a reliable description of every historical run. A 2026 New York Fed Staff Report assembling 3,984 U.S. bank runs from 1863 to 1934 finds that runs were more likely at weaker banks; runs also occurred at strong banks, but generally led to failure only when fundamentals were poor. That evidence supports a joint fundamentals-and-coordination view rather than one universal cause. S172

The main run channels#

Retail deposit runs#

Households or small firms withdraw cash or transfer deposits because they fear delay, loss, or loss of access. Deposit insurance can reduce this incentive for covered balances if the protection is credible, understood, and payable quickly.

Uninsured and concentrated deposit runs#

Large balances above an insurance limit can move rapidly. Concentration matters: a relatively small connected group of firms or investors may coordinate through shared information and transfer a large share of funding at once. The 2023 U.S. regional-bank failures renewed attention to uninsured deposits, interest-rate risk, and the speed of digital outflows. S105S122

Wholesale funding runs#

Banks also borrow from other banks, money-market funds, secured-funding markets, and institutional investors. These creditors may refuse to renew short-term funding, shorten maturities, raise prices, or demand more collateral. No branch queue is required.

Collateral and margin spirals#

Falling asset values can trigger collateral calls. Selling assets to meet those calls can push prices lower, generating further demands. This channel was particularly important in modern market-based and non-bank intermediation crises, but it can transmit stress to banks through exposures and funding.

Payment and operational runs#

A cyber incident, outage, fraud report, or communication failure can make customers doubt access even before asset quality is known. If payment delays are interpreted as insolvency, an operational event can become a funding event.

From private banknotes and clearing houses to central banks#

Nineteenth-century U.S. banking panics unfolded in a fragmented system with many note issuers, seasonal liquidity pressure, and no modern national lender of last resort. Clearing houses sometimes pooled information, issued loan certificates, or coordinated restrictions to conserve cash. Those private arrangements could contain pressure but were incomplete and uneven. S060S103

The Panic of 1907 spread through trust companies and financial markets. Private rescue coordination highlighted the absence of an elastic central reserve mechanism and helped build political support for the Federal Reserve System, established in 1913. S062S063S102

In Britain, the Bank of England's emergency role evolved through repeated crises rather than arriving fully specified in 1694. The historical lender-of-last-resort idea joined a trusted settlement institution to the ability—and expectation—to supply liquidity against assets during panic, while preserving incentives and managing losses remained contested. S051S053

Bank-run examples in the United States and beyond#

U.S. runs therefore include pre-Federal-Reserve panics, Depression-era crises and modern electronic outflows. Public backstops and funding arrangements differed. The 2026 Correia–Luck–Verner study identifies 3,984 U.S. runs during 1863–1934 from historical newspapers and separates runs with and without failure. That is a historical research sample, not a count of banks currently in trouble. These examples do not establish that any bank is experiencing a run today. S172

Bank runs during the Great Depression#

In Great Depression history, a bank run means depositors rushing to withdraw because they feared loss or restricted access. Banks held much of their assets in loans and investments, so simultaneous demands could force asset sales and disrupt payments. A run is a withdrawal event within the broader crisis, not another name for the stock-market crash of 1929. The panics of 1930–31 also show how correspondent relationships transmitted distress. S192 S200

These measures addressed different parts of the crisis. The bank holiday, emergency reopening law and creation of deposit insurance should not be treated as one event.

The Great Depression and the arrival of federal deposit insurance#

U.S. banking panics from 1931 to 1933 involved waves of withdrawals, suspensions, failures, and monetary contraction. Asset losses and deflation weakened banks; the banking crisis in turn damaged credit and economic activity. S074

The Banking Act of 1933 created federal deposit insurance and changed the institutional response to retail runs. Credible protection gives a covered depositor less reason to race others to the door. It also shifts part of the trust framework from private reputation and liquidity toward a public insurance system funded and governed under law. S064S138

Insurance did not abolish runs. Coverage is bounded by eligible institutions, products, owners, and limits. Uninsured funds can still leave. Institutions can fail from losses even without a retail queue. Poor design can encourage excessive risk if creditors expect protection without adequate pricing, supervision, resolution, and loss allocation.

The International Association of Deposit Insurers revised its Core Principles in 2025 following lessons that included the 2023 turmoil. The principles treat deposit insurance as part of a wider financial-safety net and emphasize mandate, governance, coverage, funding, public awareness, reimbursement, failure resolution, and coordination. National systems differ; this article intentionally does not publish a global table of current limits. S173

The trade-off: confidence versus moral hazard#

Credible insurance can stop a destructive race among covered depositors and protect people who cannot continuously monitor a bank. It can also weaken scrutiny if bank owners, managers, or creditors expect losses to be shifted without consequence. The design problem is therefore not simply whether insurance exists. Coverage, risk-based funding, supervision, corrective action, resolution, recoveries, and which claims remain exposed all shape incentives. S104S173

That trade-off is why deposit insurance belongs to a safety net rather than standing alone. Protection for covered depositors can coexist with the failure of an institution and losses for owners or other creditors under the applicable legal order.

Selected runs and the institutional changes they exposed#

The evidence table selects episodes for institutional change, not drama or a ranking of severity. Some are classic deposit runs; others show wholesale funding, asset-sale, currency, sovereign, or payment channels that changed the rules around banking stress.

Selected evidence

Runs matter when they change institutions

Cases are selected for mechanism and response—not ranked by drama or size.

Selected run-related banking crises

Episode	Underlying vulnerability	Transmission	Institutional response or lesson
1825: Panic of 1825United Kingdom and Latin American finance	Weak country banks and limited crisis liquidity	Runs, bank failures and commercial contraction	Often treated as a milestone in lender-of-last-resort development S051S053S073
1866: Overend Gurney crisisUnited Kingdom	Wholesale funding and concentrated credit risk	Money-market panic and threat to payments	Influential episode in development of lender-of-last-resort practice S051S053
1907: Panic of 1907United States	No central lender of last resort; uneven reserve distribution	Runs, call-money stress and payment disruption	Catalyst for National Monetary Commission and Federal Reserve Act S062S102
1929–1933: Great Depression banking panicsUnited States and global	Undercapitalized banks, correspondent exposures and no federal deposit insurance	Runs, failures, monetary contraction and debt deflation	Expanded federal safety net and bank supervision S074S120S138
2007–2009: Global financial crisisUnited States and global	Leverage, securitization opacity, wholesale funding and weak underwriting	Runs in wholesale and shadow-banking markets, failures and payment/credit stress	Led to Basel III, stress testing, resolution regimes and macroprudential policy S076S077S120S080
2023: Regional bank failures and rapid digital runsUnited States	High uninsured deposits, weak interest-rate risk management and fast information channels	Very rapid deposit outflows and confidence spillovers	Reopened questions about deposit insurance, supervision and speed of runs S105S122
2023: Failure and emergency acquisition of a globally systemic Swiss bankSwitzerland / global	Weak profitability and repeated control failures	Wholesale and depositor confidence deterioration	Raised debate over resolution credibility and too-big-to-fail frameworks S121S073

Inspect the 27-record research tableDownload CSV

How central-bank liquidity changes a run#

An eligible bank facing temporary outflows may borrow against acceptable collateral. That can replace fleeing short-term funding and avoid forced asset sales. System-wide facilities can also assure markets that settlement will continue.

Liquidity support is not the same as recapitalization. If assets are worth less than liabilities, more borrowing can delay rather than repair the loss. Valuation is difficult in a crisis, and authorities must decide under uncertainty. Facility design therefore involves collateral, haircuts, pricing, eligibility, disclosure, and legal authority.

The comparison page separates the different crisis roles of commercial and central banks from the roles of insurers, governments, and resolution authorities.

The policy tension is durable: support can stop destructive coordination, but an expectation of unconditional support can weaken market discipline. The answer is not “never lend” or “always rescue.” It is an institutional design that distinguishes liquidity from loss as well as the evidence allows and has a credible way to resolve failures.

What deposit insurance solves—and what it does not#

Deposit insurance can:

• reduce the incentive for covered depositors to run;
• protect smaller depositors who cannot monitor bank balance sheets;
• preserve access to transaction money during a failure;
• give authorities a defined reimbursement or transfer mechanism.

It cannot by itself:

• make bad assets good;
• protect every creditor or product;
• prevent uninsured or wholesale runs;
• remove operational disruption;
• replace supervision, capital, liquidity, and resolution;
• eliminate the public cost or distributional choices in a systemic crisis.

Coverage rules and limits are volatile legal facts. Readers should check the current official insurer or regulator for their jurisdiction rather than rely on an historical article.

Resolution: keep critical functions, allocate losses#

When a bank is no longer viable, resolution aims to preserve critical services while allocating losses under law. Tools can include a purchase and assumption, transfer to a bridge institution, bail-in of eligible claims, sale, recapitalization, or liquidation. The mix varies by jurisdiction and bank type.

A clear resolution regime can reduce panic by explaining who acts and how insured deposits remain accessible. But uncertainty about valuation, legal priority, operational readiness, and political willingness can remain acute. Deposit insurance and resolution should therefore be understood as connected pieces, not interchangeable labels.

Digital speed changes the clock#

Online and mobile banking let customers transfer large sums without visiting a branch. Social networks and private group chats can spread interpretation before an official statement is prepared. Corporate treasury tools can automate concentration and movement. That makes the time between doubt and a material liquidity outflow much shorter.

The underlying economics are old: short claims fund slower assets, information is incomplete, and creditors care what others will do. The operational response must be faster: real-time liquidity monitoring, tested collateral, clear communication, functioning payment rails, and resolution plans that work over a weekend—or less.

A practical mechanism map#

Read a reported bank run in this order:

1. Initial concern: asset loss, fraud, rate shock, rumor, outage, or policy event.
2. Funding exposure: which deposits or wholesale liabilities can leave, and how concentrated are they?
3. Liquidity buffer: what cash, reserves, saleable assets, collateral, and facilities are available?
4. Loss position: would asset sales or revaluation consume capital?
5. Transmission: are payments, counterparties, markets, or similar banks affected?
6. Public response: insurance, liquidity, guarantee, resolution, recapitalization, or closure.
7. Aftermath: who bears loss, and which rule or practice changes?

That sequence prevents two common mistakes: assuming every run is irrational, and assuming every run merely reveals an already-failed bank. History supports both channels and, often, their interaction.

The durable lesson#

Bank runs are not relics of branch banking. They are a consequence of issuing money-like or short-term promises against assets that cannot all be converted immediately at known value. Deposit insurance, central-bank liquidity, supervision, capital, and resolution changed who runs, how quickly authorities can respond, and who bears loss. They did not remove the need to understand the bank's assets and funding.

See the crises that changed banking institutions for the wider chronology behind those responses.