A bank run is not a failure of nerve; it is a triumph of game theory. Every depositor who joins the queue is making a perfectly rational calculation: if enough others withdraw, the bank will fail, and those who waited will lose everything. The tragedy is that this individual rationality produces collective catastrophe. A solvent institution can be destroyed in hours by nothing more than a shift in belief.

This dynamic has fascinated economists since Walter Bagehot codified central banking principles in the nineteenth century, but it remains poorly understood by the public. Most people imagine bank runs as products of rumor and hysteria — panicked crowds responding to whispers. The reality is more unsettling. A run can ignite even when depositors know their bank is fundamentally sound, because the relevant question is not "Is this bank healthy?" but "Do other depositors believe it is healthy, and will they act on that belief before I do?"

The coordination problem at the heart of banking

Banks operate on a structural mismatch that makes them inherently vulnerable. They borrow short — accepting deposits that can be withdrawn on demand — and lend long, extending mortgages and business loans that cannot be called back overnight. This maturity transformation is the engine of credit creation, allowing economies to fund long-term projects with short-term savings. It is also a loaded weapon pointed at the banking system's own head.

Under normal conditions, the mismatch is manageable because withdrawals are staggered and predictable. Banks need only keep a fraction of deposits as reserves. But when confidence cracks, the mismatch becomes fatal. No bank, however prudent, holds enough cash to satisfy every depositor simultaneously. The first to withdraw get paid in full; the last get nothing. This creates what economists call a "sequential service constraint" — a polite term for a brutal first-come-first-served queue.

Why deposit insurance doesn't fully solve the problem

Governments learned long ago that explicit guarantees can short-circuit the coordination problem. If depositors believe their funds are protected regardless of the bank's fate, they have no incentive to run. Deposit insurance schemes, pioneered in the United States during the Great Depression, have largely eliminated classic retail bank runs in developed economies.

Yet the problem has migrated rather than disappeared. Modern financial systems feature vast pools of uninsured money — corporate treasury accounts, money market funds, wholesale funding markets — that remain vulnerable to the same coordination dynamics. When Silicon Valley Bank collapsed in 2023, the proximate cause was a run by uninsured depositors, many of them technology startups with balances far exceeding insurance limits. The speed was breathtaking: social media and instant transfers compressed what once took days into hours.

The contagion mechanism

A single bank's failure need not remain isolated. Runs spread through two channels: informational and financial. The informational channel operates when depositors at Bank B interpret Bank A's collapse as evidence that similar institutions may be unsound. They need not know anything specific about Bank B's balance sheet; the mere fact of a failure elsewhere updates their beliefs about the sector's fragility. The financial channel operates when Bank B actually holds claims on Bank A, or when both banks depend on the same funding markets. A liquidity crunch at one institution can drain liquidity from others.

Central banks exist, in part, to interrupt this contagion. By lending freely to solvent institutions during a panic — Bagehot's famous prescription — they provide the liquidity that private markets withdraw. But the lender-of-last-resort function requires delicate judgment. Lend too freely and you encourage recklessness; lend too stingily and you let the fire spread.

Our take

The bank run is a reminder that finance is, at bottom, a confidence game — not in the pejorative sense, but in the literal one. The entire edifice of credit rests on shared beliefs about the future behavior of strangers. When those beliefs waver, even briefly, the system's fragility is exposed. Understanding this does not prevent crises, but it does clarify why they recur despite every regulatory innovation. The coordination problem is baked into the architecture of banking itself, and no amount of capital buffers or stress tests can fully neutralize a sufficiently widespread loss of faith.