Bank runs used to take days and involve queues. Recent ones have taken hours and involved nobody leaving their desk. The mechanism is the same and the speed is not.
The underlying fragility
Banks borrow short and lend long. Deposits can be withdrawn immediately; loans and securities cannot be converted to cash immediately without loss.
This maturity transformation is what banks are for. It is also what makes them fragile.
A bank can be entirely solvent — its assets worth more than its liabilities — and still fail if enough depositors demand cash at once, because the assets cannot be sold fast enough at full value.
The self-fulfilling part
The decision to withdraw is rational if you expect others to withdraw, regardless of what you think about the bank.
Which means a run can begin without any deterioration in the bank's actual position, purely on the expectation of a run.
Deposit insurance was designed specifically to break this, by removing the incentive for insured depositors to move first.
It works, for insured deposits. It does nothing for balances above the limit.
Why uninsured deposits matter disproportionately
A bank whose deposits are largely insured has a stable base.
A bank whose deposits are largely above the insurance limit — typically business accounts and wealthy individuals — has a base with every incentive to leave at the first sign of trouble.
Which is exactly the concentration that appeared in several recent failures, where a narrow customer base of similar businesses held large uninsured balances.
Those customers also talked to each other, which compressed the timeline further.
The interest rate connection
The recent failures had a specific cause worth understanding.
Banks holding long-dated bonds bought during the low-rate period faced large unrealised losses when rates rose, because bond prices fall as yields rise.
Accounting rules allowed some of those holdings to be carried at cost rather than market value, on the basis that they would be held to maturity.
Which is fine until deposits leave and the bonds must actually be sold, at which point the losses become real and the capital position deteriorates immediately.
Depositors who understood this moved first, which forced the sales, which confirmed the problem.
Why it is faster now
Two changes.
Transfers are instant and can be initiated from a phone at any hour, so there is no queue and no closing time.
And information spreads through networks in minutes rather than through news cycles in days.
The combination means the window in which authorities can respond has compressed dramatically, and supervisory frameworks designed around slower dynamics have been reassessed accordingly.
The lender of last resort
Central banks provide liquidity against collateral to solvent institutions facing withdrawals, which is the classical response to a run.
The constraint is collateral — a bank must have eligible assets, prepositioned and valued, to borrow against.
Banks that had not prepositioned collateral could not access facilities quickly enough, which has become a supervisory focus since.
The stigma problem is also real. Using an emergency facility signals distress, so institutions avoid it until too late, which is a well-recognised design difficulty with no clean solution.
Resolution
When a bank fails, resolution authorities can transfer deposits and assets to another institution, create a bridge institution, or impose losses on creditors according to a defined hierarchy.
The frameworks built after the financial crisis were intended to allow failure without either taxpayer support or systemic disruption.
Their application in recent cases involved decisions that departed from expectations in some respects, which generated substantial legal and political argument.
What an individual can reasonably do
Know the insurance limit and know which licensed institution each brand sits under, since two brands may share one limit.
For balances above the limit, spreading across institutions restores coverage.
Beyond that, monitoring bank health is not realistically something a retail customer can do, and the insurance system exists precisely so that they do not have to.
Stress testing
The supervisory response to all of this, and its limits.
Regulators run scenarios against bank balance sheets to assess whether capital would survive a severe downturn.
Which has genuinely strengthened the system and is constrained by the scenarios chosen.
The recent failures involved a rate shock that several stress frameworks had not emphasised, having concentrated on credit losses in a downturn rather than on interest rate risk in the banking book.
Scenarios have been broadened since, which is the pattern — each episode reveals the gap in the previous framework.
The size threshold problem
Regulatory intensity generally scales with institution size, on the reasonable basis that small failures are containable.
Thresholds were raised in some jurisdictions in the years before the recent failures, which removed several institutions from the strictest requirements.
Some of those institutions subsequently failed, and their failure proved less containable than the size-based logic assumed, because they were concentrated in a single sector whose participants all reacted together.
Which suggests that concentration matters alongside size, and frameworks have been adjusted to reflect that.