Diamond-Dybvig model · Interbank contagion · Deposit insurance
This simulator models a bank run in a fractional-reserve banking system. Each bank keeps only a fraction of deposits as liquid reserves, lending the rest. When depositors lose confidence and rush to withdraw, the bank may run out of reserves and fail. Failed banks trigger interbank contagion — linked banks lose their interbank deposits, potentially cascading into a systemic crisis. Activating deposit insurance slows or halts the panic by restoring depositor confidence (the Diamond-Dybvig insight).
The classic bank run model by Diamond & Dybvig (1983) showed that bank runs are a rational self-fulfilling prophecy: if you believe others will withdraw, it's rational for you to withdraw too — even if the bank is fundamentally solvent. Deposit insurance (like FDIC in the US or ФГВФО in Ukraine) breaks this cycle by guaranteeing deposits up to a certain amount.
This simulator models a bank run within a fractional-reserve banking system, based on the Diamond-Dybvig (1983) model of bank fragility. Each bank holds only a fraction of its deposits as liquid reserves and lends the rest; when depositors lose confidence and rush to withdraw simultaneously, reserves are exhausted and the bank collapses. Users can observe how a single failure propagates through interbank lending links, triggering a cascading systemic crisis — and how deposit insurance breaks the self-fulfilling panic cycle.
Bank runs have shaped modern financial regulation: the US bank panics of 1930-1933 led to the creation of the FDIC in 1933, while the 2007-2008 runs on Northern Rock (UK) and Washington Mutual demonstrated that the phenomenon remains relevant in the digital age.
A bank run occurs when a large number of depositors simultaneously withdraw their funds from a bank out of fear it may become insolvent. Because banks operate on fractional reserves — keeping only a fraction of deposits on hand — even a solvent bank can fail if enough depositors withdraw at once. The run becomes self-fulfilling: the fear of failure causes the very failure it anticipates.
Click "Start Panic" to trigger a depositor panic at a randomly selected bank and watch confidence collapse as withdrawals drain reserves. Adjust the Reserve Ratio slider to control how much liquidity each bank holds, and use the Contagion Speed slider to speed up or slow down how quickly panic spreads through interbank links. Toggle Deposit Insurance on at any point to see how guaranteed coverage restores confidence and halts the cascade.
When a bank's reserves fall to zero and it cannot honor a withdrawal request, it is marked as failed. Its interbank counterparties — the banks it lent to or borrowed from — immediately suffer a confidence shock of 0.3 and a direct reserve loss equal to half the interbank exposure. This models the real-world mechanism by which one institution's failure impairs the balance sheets of connected institutions, potentially triggering further runs.
Published by Douglas Diamond and Philip Dybvig in 1983 (for which they shared the 2022 Nobel Prize in Economics), the model formally proved that bank runs are rational equilibria — not mere panics. In the model, banks create value by transforming illiquid long-term loans into liquid demand deposits. However, this maturity transformation creates two possible equilibria: a "good" one where no one runs, and a "bad" one where everyone runs because they expect others to. The key insight is that deposit insurance uniquely eliminates the bad equilibrium by making early withdrawal unnecessary for self-protection.
The US Great Depression saw over 9,000 bank failures between 1930 and 1933, wiping out roughly a third of the money supply. More recently, the 2007 run on Northern Rock — the first UK bank run in 150 years — saw queues around the block before the government guaranteed all deposits. In 2023, Silicon Valley Bank experienced a digital-age bank run: $42 billion in deposits were withdrawn in a single day via mobile apps after news of losses spread on social media, illustrating how contagion speed has accelerated dramatically.
A higher reserve ratio (set with the Reserve Ratio slider) means the bank can satisfy more withdrawal requests before exhausting liquidity, so it takes longer to fail once a panic begins. However, it does not eliminate the risk entirely — given enough simultaneous withdrawals, even a bank with a 50% reserve ratio can fail. Full-reserve banking (100% reserves) would prevent runs but would eliminate the credit-creation function of banks, which is why most economists consider deposit insurance a more practical solution.
Walter Bagehot described the mechanics of banking panics in "Lombard Street" (1873) and proposed the classic lender-of-last-resort doctrine: central banks should lend freely, at a penalty rate, against good collateral, to solvent but illiquid banks. Diamond and Dybvig (1983) gave the first rigorous game-theoretic proof of run equilibria. Ben Bernanke's 1983 work showed that bank failures during the Great Depression caused lasting economic damage beyond the immediate loss of deposits — influencing his decisions as Fed Chair during the 2008 crisis.
Bank runs are closely related to currency crises (speculative attacks on exchange rate pegs), sovereign debt crises (sudden stops in rollover of government debt), and crypto exchange collapses such as the 2022 FTX failure, which exhibited identical self-fulfilling withdrawal dynamics. The simulation's interbank contagion mechanism also mirrors fire-sale externalities studied in network contagion models used by central banks to stress-test financial systems.
Deposit insurance guarantees depositors up to a specified limit — $250,000 per account in the US (FDIC), $85,000 in the UK (FSCS) — eliminating the rational incentive to run, since insured depositors lose nothing even if the bank fails. In this simulation, activating insurance gradually restores confidence at 1% per tick, halting panic once confidence exceeds 70%. The key limitation in practice is coverage caps: uninsured depositors (such as SVB's venture-backed companies holding millions) still face incentives to run, as the 2023 crisis demonstrated.
Modern research focuses on digital and social-media-accelerated runs (the "Twitter bank run" phenomenon documented in 2023), the stability of stablecoins and crypto exchanges which replicate fractional-reserve dynamics without deposit insurance, and network topology effects — how the structure of interbank lending graphs determines systemic fragility. Central banks are also studying the implications of central bank digital currencies (CBDCs), which could enable instant mass withdrawals at scale, potentially amplifying run dynamics in ways the existing regulatory framework was not designed to handle.