A network of banks lends to one another overnight. Shock one bank's assets and watch losses propagate along interbank claims — a bank that cannot absorb its losses defaults, wiping out part of what it owes its own lenders, who may then default in turn.
Systemic risk in a financial network: an isolated shock to one institution can cascade into a chain of defaults if capital buffers are thin and banks are heavily interconnected. It illustrates why "too interconnected to fail" can be as dangerous as "too big to fail."
Click "Shock Random Bank" to hit one bank with a sudden asset loss. Watch orange pulses travel along interbank loan edges as losses transmit to counterparties. Raise the capital buffer slider to make banks more resilient, or raise interconnectedness to see how a denser lending web changes contagion risk.
Research after the 2008 crisis showed that interconnectedness has a double-edged effect: at low levels it barely matters, but past a threshold it can turn a manageable shock into a full systemic collapse — a pattern regulators now stress-test for annually.
This simulation models a network of banks connected by interbank loans, the short-term overnight lending that keeps the banking system liquid. Each bank holds a capital buffer — equity that can absorb losses before the bank becomes insolvent. When a shock hits one bank's assets (a bad loan, a market crash, a run by depositors), it first eats into that bank's capital. If losses exceed the buffer, the bank defaults, and every bank that lent to it loses its claim, transmitting the shock further through the network — a process financial economists call contagion.
This mechanism was central to the 2008 global financial crisis, when the collapse of Lehman Brothers triggered losses across counterparties that had previously seemed unrelated. Academic models such as those by Allen and Gale (2000) and Eisenberg and Noe (2001) formalized how network structure and capital adequacy jointly determine systemic fragility, work that directly informs the capital requirements and stress tests used by regulators such as the Federal Reserve and the Bank of England today.
Financial contagion is the spread of financial distress from one institution or market to others that are directly or indirectly connected to it. In an interbank network, contagion occurs when a bank's default causes losses for the banks that lent to it, potentially pushing them below their own solvency threshold and causing further defaults — a chain reaction similar to how a disease spreads through a contact network.
Eighteen banks start solvent and connected by random interbank loans. Click "Shock Random Bank" to inflict a sudden asset loss on one bank. If its capital buffer cannot absorb the loss, it fails, and orange pulses show losses propagating to its lenders. Adjust the capital buffer slider to see how larger equity cushions prevent cascades, and the interconnectedness slider to see how a denser lending network changes systemic risk.
A more connected network diversifies risk in normal times, spreading small shocks thinly across many banks so no single lender is badly hurt. But past a critical density, the same connections become channels for contagion: a large shock can reach far more banks, and each one only needs a modest exposure to a failed counterparty to be pushed toward default itself. This "robust-yet-fragile" property is a defining feature of financial networks.
A capital buffer is the portion of a bank's balance sheet funded by shareholders' equity rather than debt, acting as a first-loss cushion. If a bank's assets fall in value, losses are absorbed by equity first, protecting depositors and other creditors. The simulation shows that even a small increase in the capital buffer slider can prevent an entire cascade, because it raises the loss a bank can absorb before defaulting, which in turn reduces the loss transmitted to its own lenders. This nonlinear effect is why post-2008 regulation (Basel III) significantly raised minimum capital requirements for large banks.
The clearest example is the collapse of Lehman Brothers in September 2008. Lehman's failure was not contained to its own shareholders; it caused immediate losses to money-market funds, insurers such as AIG, and hundreds of counterparties in derivatives and repo markets, freezing global credit markets within days. The interconnectedness of modern finance meant a single firm's failure produced systemic effects far beyond its own balance sheet, motivating regulators to map and stress-test the interbank network as this simulation does in miniature.
Central banks and regulators run "stress tests" that are conceptually similar to this simulation: they apply a hypothetical shock (a market crash, a major counterparty default) to real bank balance sheet data and simulate how losses would propagate through actual interbank exposures. The Federal Reserve's annual CCAR/DFAST tests and the Bank of England's stress tests use this logic to determine whether banks hold enough capital to survive severe but plausible scenarios without triggering contagion.
A bank run is depositors withdrawing funds en masse from a single bank out of fear it may fail, which can itself force a solvent bank into a liquidity crisis. Financial contagion is the broader network effect where the failure of one institution transmits losses to other institutions through direct financial claims like interbank loans. The two often interact: news of one bank's distress can trigger runs on other banks perceived as similarly exposed, amplifying the network effect modeled here.
Yes. Size matters, but the 2008 crisis revealed that "too interconnected to fail" can matter just as much. A mid-sized institution with many counterparties (as this simulation's interconnectedness slider explores) can trigger contagion just as effectively as a giant bank if its failure hits enough lenders simultaneously. This insight led regulators to designate certain firms as "systemically important financial institutions" (SIFIs) based on interconnectedness and complexity, not just balance-sheet size.
The default cascade modeled here is mathematically similar to epidemic spreading (SIR models) on contact networks, cascading failures in electrical power grids, and information cascades in social networks. In all these systems, a local disturbance can either fizzle out or trigger a large-scale cascade depending on the network's connectivity and the "resilience" of individual nodes — a general pattern studied in the science of complex networks.
Active research includes modeling "fire sale" contagion, where a distressed bank sells assets at depressed prices, hurting other banks holding similar assets even without a direct lending relationship; incorporating central clearing counterparties that concentrate and redistribute risk; and using machine learning on real transaction-level data to reconstruct hidden interbank exposure networks, since actual lending relationships between banks are often confidential and only partially observable to regulators.