A bank can report healthy capital and still face an uncomfortable question: how quickly could its customers move their money? Deposits have long been a central source of bank funding, but their apparent stability is not uniform. A savings balance held for years by one household behaves differently from a large operating account whose owner monitors rates and risk every day. Fast digital transfers make the difference more consequential when confidence weakens.
The failures of several US banks in 2023 gave that question urgency. The Federal Reserve’s review of Silicon Valley Bank identified a concentrated business model, high reliance on uninsured deposits and weaknesses in managing interest rate and liquidity risks. The lesson for the wider industry is not that all deposits will run. It is that treating balances as interchangeable can hide important differences in who owns them, why they are held and how they might move.
A deposit balance is not a funding promise
When a customer deposits money, the bank owes the customer that amount under the account terms. It can use funding to support assets while retaining liquidity to meet withdrawals. A bank therefore needs to estimate not just how much money it holds today but how much could leave under different conditions. Deposits tied to payroll, payments or long-standing household relationships may prove steadier than balances attracted by a short-lived rate promotion, though none are guaranteed to remain.
Insurance coverage matters because it changes incentives. A fully protected retail depositor has less reason to make an immediate judgement about a bank’s ability to repay. An uninsured corporate treasurer has a duty to manage exposure and may spread balances across institutions. Yet insured status is only one variable. A payment outage, a rapid change in rates or uncertainty over access can also affect customer behaviour.
The FDIC’s staff study of deposit flows at three failed banks offers direct evidence about how withdrawal patterns developed during the 2023 stress. That experience should inform scenario design without being copied mechanically into every bank’s forecast. A diversified retail lender and a bank serving one tightly connected sector do not have the same depositor network.
Speed is changing the shape of stress
Customers can check balances and move funds without visiting a branch. A treasurer can arrange large transfers within an ordinary working day. Digital convenience is generally valuable; it also reduces the friction that once slowed decisions during a confidence shock. The pace of communications can amplify the pace of transactions, especially when many clients respond to the same information at once.
An IMF analysis of technology-driven bank runs argues that technology can accelerate outflows and complicate conventional assumptions about how long a bank has to respond. That does not mean social media alone causes bank failures. Losses, business concentration, risk management and supervisory issues remain material. The more precise point is that a bank with fragile funding may have less time to address concerns once they become widely shared.
Banks are therefore paying closer attention to intraday information. End-of-day reports can obscure a morning of large withdrawals or a rapid shift from one group of customers. Liquidity teams need reliable data about account ownership, deposit insurance, payment rails and which customers may act together. More frequent reporting has value only when executives know what action the information should trigger.
Concentration can matter more than a headline ratio
Suppose two banks have the same amount of total deposits. One serves millions of households across regions; the other depends heavily on a small number of corporate groups within one industry. Even if both show a similar balance sheet today, the second is exposed to a more correlated decision. A market event affecting that industry could cause many clients to withdraw or reduce balances at the same time.
Effective segmentation goes beyond counting accounts. Banks can examine the size of each balance, links among depositors, how the money is used, historical behaviour during changes in interest rates and whether balances depend on one intermediary or platform. Where a large proportion of funds arrive through brokers or digital marketplaces, the bank should understand how quickly those channels can redirect them.
The Basel Committee’s liquidity coverage framework requires qualifying banks to hold high-quality liquid assets against a defined 30-day stress scenario. It establishes an important baseline. Yet a standardised ratio cannot capture every concentration, reputational problem or timing mismatch. Internal stress tests should probe the specific sources of each bank’s funding, including unusually fast outflows.
The other side of the balance sheet still matters
Deposit stability cannot be assessed in isolation. Banks hold loans and securities with different cash flow and market characteristics. If withdrawals require a bank to sell assets quickly, unrealised losses may become realised losses. Rising rates can increase the cost of retaining deposits while reducing the market value of some fixed-rate assets. A funding plan that looks comfortable in normal conditions can become strained when those pressures occur together.
Management can hold liquid assets, secure diversified contingent funding and reduce mismatches between asset maturities and liabilities. Each choice has a cost. Cash and low-risk securities usually earn less than some loans; term funding can be more expensive than deposits. The objective is not to eliminate every risk at any price, but to ensure that the bank’s growth strategy is consistent with its ability to meet obligations in stress.
The US Office of the Comptroller of the Currency’s risk perspective is a reminder that liquidity risk sits alongside credit, interest rate and operational risks. For a bank with large commercial deposits, a review of funding should also ask how quickly assets could be pledged or sold and whether the related processes have actually been tested.
Relationship banking gains a measurable role
Banks often describe customer relationships as a source of loyalty. The relevant operational question is more specific: does a customer use the bank for recurring payments, payroll, collections and credit, or only to place a balance at a competitive rate? A deep relationship may make deposit flows easier to forecast, but it is not a guarantee against withdrawal when a client perceives elevated risk.
Rate competition adds another complication. When market rates rise, customers may move funds in search of better returns even if they remain confident in their bank. That is ordinary competition rather than a run. A liquidity model should distinguish price-sensitive migration from distress-driven flight because the bank’s responses differ. Higher deposit rates can retain funds but compress margins; a confidence problem calls for credible information and a robust funding plan.
Research on bank runs and public communication from the BIS examines how beliefs can influence collective behaviour. Clear communication can help, but it must be supported by facts and operational readiness. Vague reassurance will not replace access to liquid assets or sound supervision.
Better monitoring needs clear decisions
A useful funding dashboard can show the largest depositors, expected recurring flows, uninsured balances, rate sensitivity and current liquid resources. It should also show concentrations hidden behind related entities or a single customer acquisition channel. Banks can test what happens if several large clients withdraw together, if an online rumour raises concern or if an operational incident disrupts payments.
The difficult part is linking thresholds to action. Who can raise funding, draw on a facility, slow new asset growth or contact supervisors? Are securities eligible to be pledged, and have the bank’s teams completed the documentation? Has management considered what different actions might signal to customers? A stress test that assumes an emergency response is available without checking execution is incomplete.
Deposits remain a valuable and often resilient foundation for banking. The adjustment is to treat stability as a characteristic that must be evidenced rather than assumed. Banks that understand why customers keep money with them, and how those reasons could change, can make better lending and liquidity decisions. That is a stronger measure of funding quality than the size of deposits on a single reporting date.
Why payment data can improve the forecast
Customer balances are a snapshot; transactions reveal how the money behaves. Payroll and supplier payments, tax cycles, seasonal sales and investment maturities can all create large but predictable movements. Banks that understand these rhythms can distinguish an ordinary working-capital withdrawal from an unusual run of transfers. Doing so requires combining account data with relationship knowledge and protecting sensitive customer information.
Data can be misleading if it is stripped of context. A large outflow from a company during an acquisition may be expected, while a smaller but widespread series of transfers from otherwise steady clients could signal broader concern. Trends should be assessed against the relevant peer group and time period. Alert systems should help a treasury team ask the right question, rather than label every unexpected movement a crisis.
Models also need to account for customer choice when interest rates move. A balance held in a non-interest-bearing account may become less stable when attractive alternatives are readily available. Historical experience from a low-rate period can understate this sensitivity. Scenario analysis should make the assumption explicit and compare it with observed behaviour as conditions change.
Stress plans must work outside an exercise
A bank can identify sources of emergency funding on paper without being ready to use them quickly. Securities need to be eligible and correctly recorded as collateral. Legal agreements and access to central bank facilities, where available, must be current. Teams need authority to act and a way to coordinate across treasury, operations, communications and customer service. Testing these steps can expose gaps that a high-level liquidity ratio cannot.
An orderly response also depends on communication. Customers may ask whether payments will continue, how deposits are protected and whether account access is affected. Clear, accurate information can reduce confusion, but the bank should not improvise claims it cannot substantiate. Internal escalation and regulatory contact should be assigned in advance, with a clear record of decisions and changing assumptions.
Not every situation calls for the same response. Paying up to retain deposits may be reasonable in a competitive market; using a temporary funding source may bridge an ordinary liquidity need; slowing balance-sheet growth can reduce pressure when funding becomes less reliable. The aim of contingency planning is to make these choices earlier and on better information.
What boards should ask
Directors need to understand more than a single regulatory ratio. They can ask how much of the deposit base comes from the largest related clients, what share is uninsured, how balances changed during recent rate moves and whether the bank has actually tested the liquidity options assumed in a severe scenario. They should understand how quickly management would know that assumptions had stopped holding.
A useful board report should explain what would have to happen for the bank to change lending, pricing or funding plans. It should show the costs of maintaining resilience as well as the losses a failure to prepare could cause. That makes the discussion a strategic one: the bank’s appetite for particular customers and assets must be consistent with the funding it can keep through difficult periods.
Questions readers may ask
Are uninsured deposits inherently unsafe
Insurance limits concern the protection available to depositors, not a simple verdict on an individual bank. Large clients commonly manage uninsured balances through diversification, cash management policies and assessment of their banking partners.
Does a high liquidity ratio guarantee safety
No. It is an important measure under a defined scenario. Concentrated withdrawals, difficult-to-sell assets, operational delays and weaknesses in governance can still create problems.
What changed with digital banking
Transferring funds became easier and faster, while information spreads rapidly. Banks need more timely visibility into deposits and realistic plans for responding to a sharp change in behaviour.
Sources
Federal Reserve’s review of Silicon Valley Bank
FDIC’s staff study of deposit flows at three failed banks
IMF analysis of technology-driven bank runs
Basel Committee’s liquidity coverage framework
US Office of the Comptroller of the Currency’s risk perspective