Common mistakes about bank runs and banking crises
Most beliefs about bank runs are anchored on two outdated images: the 1930s queue at the teller and the 2008 credit crisis. The 2023 regional bank failures showed a different mechanism — a solvent bank felled in days by interest-rate risk and a deposit flight coordinated at digital speed. This guide corrects seven recurring misconceptions and links each to its full explanation.
In this guide
- “A bank run only happens when a bank is already insolvent”
- “2023 was just a smaller version of 2008”
- “Bank runs are slow, queue-at-the-branch events”
- “Too big to fail means only the largest banks matter”
- “Deposit insurance protects all my money”
- “SVB failed because of risky loans”
- “One bank’s failure drags down the whole system”
- The pattern behind these mistakes
- Practical observation
- Frequently asked questions
Why these mistakes persist
Public intuition about banking crises is shaped by two vivid but partial pictures: depositors lining up outside a branch in the 1930s, and the slow unravelling of mortgage credit in 2008. Both describe real episodes, yet both mislead about how a modern run actually unfolds. The 2023 failures were faster, were driven by interest-rate risk rather than bad loans, and spread through screens rather than sidewalks. The corrections below replace folklore with the documented mechanics.
→ New to financial stability? Macro-financial regimes
“A bank run only happens when a bank is already insolvent”
The common belief: Depositors only withdraw once a bank has genuinely lost money, so a run is just a rational reaction to an insolvency that is already a fact.
What the data shows: The framework that won the 2022 Nobel Prize — Diamond and Dybvig (1983), alongside Bernanke — describes a run as a self-fulfilling equilibrium. A solvent intermediary holding liquid liabilities against illiquid assets can fail purely because each depositor rationally rushes to be first, anticipating that others will do the same. Silicon Valley Bank was in sound condition days before it collapsed, yet customers requested roughly 42 billion dollars in withdrawals on a single day, 9 March 2023. The trigger was coordination, not a balance sheet that was already underwater.
→ Full explanation: What causes a bank run and why are they self-fulfilling?
“2023 was just a smaller version of 2008”
The common belief: The 2023 episode was a mini-2008 — the same kind of crisis, simply contained to a few regional banks.
What the data shows: The mechanisms were opposites. 2008 was an asset-side, credit crisis: bad mortgages, opaque securitization, and leverage eroded solvency over months. 2023 was a liability-side, duration crisis: banks held safe US Treasuries and agency bonds that lost market value when the Fed raised rates by 525 basis points between March 2022 and July 2023. One crisis came from credit losses, the other from interest-rate risk meeting a concentrated, uninsured deposit base. Reading 2023 through the 2008 template misidentifies both the cause and the speed.
→ Full explanation: How did the 2023 regional bank crisis differ from 2008?
“Bank runs are slow, queue-at-the-branch events”
The common belief: A run looks like the 1930s — anxious savers physically lining up at the teller over several days.
What the data shows: Silicon Valley Bank collapsed in roughly two days in March 2023, with the bulk of withdrawal requests arriving in a single day, coordinated through messaging apps, venture-capital group chats, and social media. A concentrated, digitally connected depositor base can act far faster than any branch queue. The Diamond-Dybvig coordination problem still applies, but it now resolves at smartphone speed rather than over a week of physical queues.
→ Full explanation: How does social media accelerate bank runs?
“Too big to fail means only the largest banks matter”
The common belief: Systemic risk is about absolute size — only the giant banks are too big to fail, and mid-sized lenders are not a system-wide concern.
What the data shows: The phrase was popularized in the 1984 rescue of Continental Illinois, then the seventh-largest US bank with about 40 billion dollars in assets. Yet in 2023 regulators invoked a systemic-risk exception for Silicon Valley Bank, only the 16th-largest. Size is one axis; interconnectedness, depositor concentration, and the credibility of the backstop matter alongside it. The implicit promise of rescue also creates moral hazard, since institutions expecting support face weaker incentives to limit risk.
→ Full explanation: What are too-big-to-fail banks and the moral hazard?
“Deposit insurance protects all my money”
The common belief: If a bank is FDIC-insured, every dollar held there is protected if it fails.
What the data shows: FDIC coverage is capped at 250,000 dollars per depositor, per bank, per ownership category — a limit raised from 100,000 dollars in October 2008 and made permanent by Dodd-Frank in 2010. At Silicon Valley Bank, about 89% of deposits exceeded that ceiling at the end of 2022, according to the FDIC. That concentration of uninsured money is precisely what made the run so fast: large depositors had every incentive to flee, because insurance caps the panic only for smaller accounts.
→ Full explanation: How does FDIC insurance actually work?
“SVB failed because of risky loans”
The common belief: Like most failed banks, Silicon Valley Bank must have made reckless or speculative loans.
What the data shows: SVB mostly held US Treasuries and agency mortgage bonds — among the safest credit assets available. The problem was duration: those long-dated bonds lost market value as rates rose. A 1.8 billion dollar loss on a securities sale, announced 8 March 2023, made the gap visible, and a concentrated, uninsured, digitally connected depositor base then pulled tens of billions within days. The failure came from interest-rate risk and funding concentration, not from credit risk on the loan book.
→ Full explanation: Why did SVB fail so quickly in March 2023?
“One bank’s failure drags down the whole system”
The common belief: Bank failures spread like dominoes, so one collapse automatically threatens the entire financial system.
What the data shows: Contagion travels through specific channels — direct exposures, shared asset holdings marked to the same prices, and confidence or information spillovers. In March 2023, Signature Bank closed on 12 March and First Republic failed on 1 May, not because they had lent to SVB, but because depositors generalized the same template of unrealized bond losses and uninsured deposits to comparable banks. Yet most regional banks did not fail. Contagion is channel-dependent and confidence-driven, not an automatic cascade.
→ Full explanation: What is contagion in financial crises?
The pattern behind these mistakes
The common confusion is treating a bank failure as a verdict on asset quality, when it is more often a verdict on the speed and concentration of funding. By regime, the contrast is sharp: a credit-driven crisis like 2008 originates on the asset side, builds over months from loan losses and leverage, and reflects genuine insolvency; a rate-driven crisis like 2023 originates on the liability side, builds in days, and can strike an institution that is solvent on paper but holds long-duration bonds marked down by a 525-basis-point hiking cycle. Both reduce to the same Diamond-Dybvig core — any intermediary with liquid liabilities and illiquid assets is exposed to self-fulfilling flight. The transition parameter is the deposit base itself: the share of uninsured money and the speed at which it can move determine how fast a confidence shock becomes a run, and in 2023 the pivot was the combination of marked-down duration with roughly 89% uninsured deposits at the epicentre. The same maturity mismatch sits well outside deposit-taking — in money funds, securitisation vehicles and open-ended funds alike — which is what puts the shadow banking system and what its scale implies on the stability agenda.
A bank does not need to be insolvent to fail — it only needs its depositors to believe, together and fast enough, that it might be.
→ Framework in view: Systemic fragilities and financial stability
Practical observation
What the data suggests for framing your own analysis:
- Question to ask yourself: for a given bank, what share of deposits sits above the insurance limit, and how concentrated is the depositor base around a single industry or network?
- Data to monitor: the share of uninsured deposits and the size of unrealized losses on securities portfolios, both reported in FDIC quarterly banking data.
- Historical parallel: Continental Illinois, 1984 — then the largest US bank failure, with about 40 billion dollars in assets, whose rescue first popularized the phrase “too big to fail.”
- What the literature documents: Bernanke (1983) showed that bank failures deepened the Great Depression by disrupting the supply of credit, beyond their direct monetary effects.
This is descriptive information to help you frame your own analysis. Eco3min does not provide investment advice.
Go deeper
📊 Complete analysis: Restrictive monetary policy: delayed effects and credit transmission
📁 Datasets: US bank lending standards · Financial conditions index
Related guides
Frequently asked questions
Is a bank run still possible when deposits are insured?
Yes, but mainly for the uninsured portion. FDIC insurance caps coverage at 250,000 dollars per depositor, per bank, which removes the incentive for small savers to rush. The risk concentrates in deposits above that limit, where there is no backstop. At Silicon Valley Bank, roughly 89% of deposits exceeded the cap at the end of 2022, so the run was driven almost entirely by large uninsured depositors moving together. Insurance dampens panic among small accounts, but a bank funded mostly by uninsured deposits remains structurally run-prone, regardless of the headline coverage figure. Alongside that funding fragility runs a second post-2008 concern, the risk a bank takes on its own account, which is the object of the Volcker rule as it stands today.
Why can a solvent bank fail in days?
Because a run is a liquidity-and-speed event, not a slow credit event. A bank can be solvent on a hold-to-maturity basis yet unable to meet sudden withdrawals if it must sell long-duration bonds at a loss to raise cash. When the depositor base is concentrated and digitally connected, withdrawal requests can arrive within hours rather than over weeks, as the roughly 42 billion dollars requested at SVB in a single day on 9 March 2023 illustrates. The combination of marked-down assets, uninsured deposits, and instant coordination turns a confidence shock into a funding shortfall faster than the institution can liquidate, which is the core distinction from the slow credit failures of 2008.
Does one bank’s failure mean the whole system is at risk?
Not automatically. Systemic contagion requires a transmission channel: a direct exposure to the failed bank, a common holding marked down across many institutions, or a confidence spillover where depositors generalize a pattern. In 2023, the failures of Signature and First Republic followed from depositors applying the SVB template — unrealized bond losses plus heavy uninsured funding — to similar banks, rather than from direct links to SVB. Crucially, the great majority of regional banks did not fail, which is why the episode is better described as a confidence-driven, channel-dependent scare than as a system-wide cascade. Channel-dependence cuts the other way too: in March 2021 several prime brokers discovered simultaneously that they had margined the same concentrated book, which is the March 2021 unwind that cost several banks ten billion dollars.
Last updated — 28 July 2026
Disclaimer – Financial Information: The analyses, commentary, and content published on eco3min.fr are provided for informational and educational purposes only. They do not constitute investment advice or a solicitation to buy or sell financial instruments. Past performance is not indicative of future results. All investment decisions involve risk and are the sole responsibility of the reader.
Read next
Full pillar →Common mistakes about credit and spreads
Most credit misconceptions share one root: treating the credit market as a system that reacts to the economy…
Common mistakes about inflation
Most inflation mistakes share one root: treating inflation as a single, mechanical number rather than a regime. This…
Common mistakes about retirement planning
Most retirement errors share one move: turning a context-dependent variable into a fixed constant. The nest egg, the…