What a Run Actually Does to a Bank
The subject here is the bank-level mechanics between a depositor’s decision to withdraw and the institution’s eventual outcome. The key question is not simply whether a bank has enough assets on paper, but whether it can meet claims that arrive now without destroying the value of assets that pay off later. Following that sequence makes it possible to see why a run can expose weakness, amplify it, or be contained before it becomes a failure.
The timing problem inside a bank
A bank typically holds a mixture of assets with different timing. It has cash or other readily available funds, but it also has loans and investments whose value is realized over time. On the liability side, depositors may have the right to demand payment before those longer-term assets mature. That arrangement is useful when withdrawals are ordinary and staggered. It becomes fragile when many depositors want payment at once.
The New York Fed study’s simple framework captures the problem with two ingredients: liquid cash and illiquid loans financed partly by demandable deposits. The bank does not need enough cash to pay every depositor immediately under normal conditions. It needs enough to meet the withdrawals it expects while continuing to hold loans until they repay. A run changes that expectation. The immediate question becomes how many claims will be presented before the bank can convert longer-term assets into cash without a large loss.
This is why “liquidity” and “solvency” should not be treated as synonyms. Liquidity concerns the ability to make payments when they fall due. Solvency concerns whether the value of the assets, if managed and realized over time, is sufficient to cover what the bank owes. A bank can be solvent but temporarily illiquid. It can also be liquid for a while even though its underlying assets are worth less than its obligations. The outside view may not reveal which situation is present, especially when the bank’s assets are difficult for depositors to evaluate quickly.
The distinction is not merely technical. It determines whether a withdrawal wave is a queue that the bank can manage or a race that changes the value of the bank itself. The sequence is easier to see in a simple map:
The diagram does not say that every shortfall ends in failure. It shows why the risk exists. A bank can move from a payment problem to a permanent institutional loss if it cannot buy time, obtain support, or convince depositors that remaining claims are safe. It also shows why failure can make the asset side worse: assets that might have produced a higher value under ordinary management may be sold or administered under less favorable conditions.
How a run can destroy value even when loans might repay
Suppose a bank owns a portfolio of loans that would repay in full if borrowers continued to perform and the bank kept managing those relationships. That does not mean the bank can pay everyone today. If withdrawals exceed its cash, it has to find another source of liquidity, restrict convertibility, or sell assets before their natural repayment date.
A forced sale can turn a timing problem into a value problem. An asset sold quickly may fetch less than its expected value at maturity. A receiver may not manage the portfolio with the same information, incentives, or relationships as the original bank. The historical framework represents this as value destruction in failure: the loan portfolio is worth less when it is liquidated or handled through receivership than it would have been under continuing operation.
The first depositors served may receive full repayment. Later depositors, or those who remain after the bank fails, may receive only a share of what is left. That sequential feature creates an incentive to withdraw early when people believe others will run. The decision is not necessarily a judgment that every loan is worthless. It can be a judgment that waiting is dangerous because the order of payment matters.
This is the self-reinforcing part of a run. If everyone remains, the bank may continue to collect loan payments and serve ordinary withdrawals. If everyone expects everyone else to withdraw, each person has a reason to move first. The collective result can force the bank into the very failure state depositors feared. In that narrow sense, a run can be more than a symptom: it can change the value of the bank’s assets by changing how and when they must be realized.
Yet this possibility should not be mistaken for a description of every historical run. The New York Fed evidence shows that most recorded runs did not end in failure. A mechanism that can produce a bad equilibrium is not the same as a claim that every observed episode reached one.
Failure also matters because a bank is not only a warehouse of assets. Lending depends on information about borrowers, their projects, and their ability to respond to trouble. A failed bank can disrupt that credit relationship even if another institution eventually acquires some of the loans. The historical literature treats this loss of intermediation as one reason bank failures can have persistent effects on credit. The run is therefore connected to the wider economy through two routes: it can damage the bank’s funding immediately, and failure can interrupt the institution that was screening, monitoring, and renewing credit.
Three routes from the same visible withdrawal wave
The same queue or transfer pattern can arise from different mechanisms, and the mechanisms make different predictions about bank strength.
A self-fulfilling liquidity panic
In a non-fundamental run, the withdrawal decision is driven mainly by expectations about other withdrawals. A relatively healthy bank can be exposed because depositors do not need to discover an actual hole in its assets before they act. They need only believe that too many others will act first.
This mechanism explains why a small shock can have a large effect in theory. It also explains why the run itself can become a source of failure: once cash is exhausted, forced liquidation reduces the value of the asset portfolio. But the theory carries an important prediction. If a sound bank can suspend, obtain support, or otherwise avoid the value-destroying sequence, then the run need not have the same ending as the classic model assumes. The institutional response changes the available equilibrium.
A run responding to weak fundamentals
A fundamental-based run is not random pressure on an otherwise unchanged bank. It is a withdrawal response to a bank with less financial room. The historical study measures this condition through variables such as capitalization, leverage, liquid assets, and reliance on noncore funding. The exact measures are imperfect, but the pattern is consistent: runs are more likely at banks that are closer to insolvency or less able to finance withdrawals cheaply.
Here the run can still be an amplifier. A weak bank may have survived for a time because depositors were inattentive, information was incomplete, or losses had not yet become salient. Once withdrawals begin, the bank loses funding and time. A failure may arrive sooner, and the economic damage may be larger, even if the underlying weakness preceded the queue.
An information-based run
An information-based run starts with a signal that is not necessarily specific to the target bank. A fall in the stock market, a rise in local business failures, or a run at another bank may lead depositors to revise their beliefs about the safety of bank deposits generally. When bank assets are opaque, depositors may not be able to tell the strong bank from the weak one quickly enough.
That is why a bank can be run without having changed its own balance sheet. The new information may be about the economy, about another institution, or simply about what other depositors appear to know. The target bank becomes part of a changed information environment. If it is strong, the run may be contained. If it is weak, the same signal may uncover a problem that was already present.
The three routes can also overlap. A deteriorating economy can weaken borrowers, make a bank’s assets less valuable, and create alarming news at the same time. Depositors can respond to the news while the bank’s fundamentals are genuinely worsening. The evidence therefore cannot be read as a clean choice between “rational information” and “irrational panic.” The useful question is which mechanism is doing the most work in a particular episode and what it implies for the bank’s ability to survive.
Why another bank’s run can change this bank’s risk
A run at another bank is a particularly powerful public signal because it is visible and easy to interpret as evidence that something is wrong with banking. Depositors may not know whether the first bank failed because of a unique balance-sheet problem or because all banks face the same threat. Under uncertainty, they may protect themselves by moving funds before they have resolved the distinction.
The historical evidence supports the first part of this chain: runs were more likely after a run on another bank in the same city. But the signal did not carry the same implication as every other adverse event. A run that followed another local bank’s run was associated with a lower probability of failure than other runs—about 14 percentage points lower in the reported analysis. That pattern is consistent with a “jittery” response in which some healthy banks are pulled into the withdrawal wave.
By contrast, a rise in nonfinancial business failures points toward losses among borrowers. It is not only a signal that depositors are frightened; it may indicate that banks’ loan portfolios are deteriorating. In the study, a one-percentage-point increase in the state business-failure rate—roughly a one-standard-deviation rise—was associated with about a seven-percentage-point increase in the pass-through from a run to failure.
These estimates are not universal constants, and the underlying variables are proxies. A local business-failure rate can reflect a common shock that affects both depositors and banks, so it does not by itself prove a one-way causal path. Still, the contrast is analytically valuable. “Contagion” is not a single diagnosis. Some contagion transmits fear; some transmits information about real losses.
Why the theories disagree about failure
Theories disagree because they place the decisive moment in different parts of the sequence. In a canonical self-fulfilling model, the run equilibrium leads to liquidation, so a run and failure are closely linked by construction. If the liquidation can be avoided, the model may no longer predict a run in the first place.
Information-based theories allow a different outcome. Depositors may be wrong about a strong bank, but the bank can respond. It may accommodate withdrawals, persuade depositors, receive an injection of equity or a loan, borrow from another bank, or temporarily suspend payments while its condition is examined. These actions do not erase the run. They alter the bank’s time horizon and the information available to depositors.
This is the central mechanical reason that a run is not the same thing as failure. Failure is not determined solely by the number of people who withdraw. It depends on the bank’s starting condition, the value lost if assets must be liquidated, and the resources—financial, informational, and institutional—available before that point. A strong bank can face real pressure and still reopen. A weak bank can be forced into closure by a run that was only the final trigger.
The historical paper therefore treats the competing theories as evidence-guiding frameworks rather than as mutually exclusive verdicts. Its data show that fundamentals matter, public signals matter, and survival mechanisms matter. The next question is which observable conditions distinguish a bank that is merely exposed to a run from one whose run is likely to end in failure.