What it means
Banks take in deposits and lend most of that money out, keeping only a modest reserve in cash or in assets they can sell within hours. The arrangement works because on any normal day only a tiny fraction of depositors ask for their balance back.
It stops working the moment that assumption breaks. A run begins when enough depositors doubt the bank can pay everyone.
For any individual depositor, waiting politely is the losing move, so perfectly rational people join the queue, and the queue itself becomes the evidence that something is wrong. The prediction of failure is what makes failure happen.
For a business, the exposure is rarely shareholding in a bank; it is operating cash sitting above the deposit insurance limit. If a payroll account is frozen inside a failed institution, the eventual accounting loss may be modest but the operational damage lands in the same week.
Modern runs move at the speed of a banking app rather than a queue on the pavement, and balances can leave in hours instead of days. That is why regulators now watch liquidity coverage and the concentration of large uninsured deposits so closely.
The same structural weakness sits outside traditional banking as well: money market funds, stablecoin issuers and lenders that fund long-dated assets with overnight borrowing all share it. The common thread is a mismatch between assets that take time to sell and liabilities that can be demanded instantly.
In practice
Real-world examples
Example 1
A software company keeps $4,200,000 of operating cash in one mid-sized bank. When rumours about that bank spread on social media, the finance director spends the morning moving payroll funds to a second institution, joining the run purely to protect December salaries rather than out of any real view on solvency.
Example 2
A building materials supplier notices its bank refusing to renew a credit facility at short notice and quietly tightening limits. Reading that as a liquidity signal, the supplier splits its balances across three banks so that no single failure can freeze more than a third of its cash.
Example 3
A crypto exchange promises instant customer withdrawals while lending customer balances out on 90-day terms. When withdrawals surge over a weekend, the exchange suspends them, and the suspension itself triggers exactly the panic it was meant to contain.
Think of it
Imagine a musical chairs game where ten people are playing, but there are only two chairs. As long as the music plays, everyone is calm. But the moment someone yells that chairs are running out, a mad dash begins, and most people are left standing.
Formula
Calculation
Withdrawal coverage = liquid reserves / withdrawals demanded. A regional bank holds $800,000,000 of customer deposits and keeps $80,000,000 in cash and instantly sellable securities, a reserve of 10%. Bad news spreads and 15% of deposits are demanded in a single day: $800,000,000 x 0.15 = $120,000,000. Coverage is therefore $80,000,000 / $120,000,000 = 66.7%, leaving a shortfall of $40,000,000. To raise that cash the bank has to sell loans carried at $50,000,000 on its books, but in a forced sale they fetch only 80 cents on the dollar: $50,000,000 x 0.80 = $40,000,000. The bank meets the withdrawals but crystallises a $10,000,000 loss, which makes the next day's headlines worse and the following day's withdrawals larger.Case study
Seen in the real world
The following is an illustrative and entirely fictional example. Meridian Coast Savings was a mid-sized regional bank with $800,000,000 of deposits, most of them belonging to local construction firms and property developers. It was profitable, well capitalised on paper and had never missed a regulatory test.
In the illustrative scenario, a widely read newsletter noted that Meridian Coast had a large book of long-dated fixed-rate loans that had fallen in value as interest rates rose. Within two days, 15% of deposits had been withdrawn through the mobile app, mostly by a few dozen business customers holding balances well above the insurance limit. Meridian Coast was not insolvent when the week began, but selling loans at distressed prices to fund the outflow made it so by Friday.
The lesson the fictional bank's survivors drew was about concentration rather than credit quality. A deposit base made up of a small number of large, well-informed business accounts can leave far faster than a base of thousands of small insured savers.
Watch out
Common mistakes
- Assuming bank runs only happen to badly managed banks, when in fact any institution funding long-dated assets with instantly withdrawable deposits is structurally vulnerable.
- Treating deposit insurance as covering the whole company balance, when the guarantee applies only up to a per-depositor limit and anything above it ranks as an ordinary claim.
- Confusing a liquidity problem with insolvency; a bank can have more assets than liabilities and still fail simply because it cannot turn those assets into cash fast enough.
Questions
People also ask
Can a profitable bank fail in a run?
Yes, because profitability measures earnings over time while a run is a demand for cash today, and the forced sale of good assets at bad prices can wipe out capital in days.
How should a business protect its cash from a bank failure?
Spread balances across two or three institutions, keep amounts within insurance limits where practical, and keep a backup account already opened so funds can be moved within one working day.
Do central banks stop runs?
Often they do, because a central bank acting as lender of last resort can lend against sound collateral so the bank pays withdrawals without fire-selling assets, but that support comes with conditions and rarely protects shareholders.