By George G. Kaufman
A run on a bank occurs when a large number of depositors, fearing that their bank will be unable to repay their deposits in full and on time, simultaneously try to withdraw their funds immediately. This may create a problem because banks keep only a small fraction of deposits on hand in cash; they lend out the majority of deposits to borrowers or use the funds to purchase other interest-bearing assets such as government securities. When a run comes, a bank must quickly increase its cash to meet depositors’ demands. It does so primarily by selling assets, often hastily and at fire-sale prices. As banks hold little capital and are highly leveraged, losses on these sales can drive a bank into insolvency.
The danger of bank runs has been frequently overstated. For one thing, a bank run is unlikely to cause insolvency. Suppose that depositors, worried about their bank’s solvency, start a run and switch their deposits to other banks. If their concerns about the bank’s solvency are unjustified, other banks in the same market area will generally gain from recycling funds they receive back to the bank experiencing the run. They would do this by making loans to the bank or by purchasing the bank’s assets at non-fire-sale prices. Thus, a run is highly unlikely to make a solvent bank insolvent.
Of course, if the depositors’ fears are justified and the bank is economically insolvent, other banks will be unlikely to throw good money after bad by recycling their funds to the insolvent bank. As a result, the bank cannot replenish its liquidity and will be forced into default. But the run would not have caused the insolvency; rather, the recognition of the existing insolvency caused the run.
A more serious potential problem is spillover to other banks. The likelihood of this happening depends on what the “running” depositors do with their funds. They have three choices:
They can redeposit the money in banks that they think are safe, known as direct redeposit.
If they perceive no bank to be safe, they can buy treasury securities in a “flight to quality.” But what do the sellers of the securities do? If they deposit the proceeds in banks they believe are safe, as is likely, this is an indirect redeposit.
If neither the depositors nor the sellers of the treasury securities believe that any bank is safe, they hold the funds as currency outside the banking system. A run on individual banks would then be transformed into a run on the banking system as a whole.
If the run is either type 1 or type 2, no great harm is done. The deposits and reserves are reshuffled among the banks, possibly including overseas banks, but do not leave the banking system. Temporary loan disruptions may occur because borrowers have to transfer from deposit-losing to deposit-gaining banks, and interest rates and exchange rates (see foreign exchange) may change. But these costs are not the calamities that people often associate with bank runs.
Higher costs could occur in a type 3 run, because currency (an important component of bank reserves) would be removed from the banking system. Banks operate on a fractional reserve basis, which means that they hold only a fraction of their deposits as reserves. When people try to convert their deposits into currency, the money supply shrinks, dampening economic activity in other sectors. In addition, almost all banks would sell assets to replenish their liquidity, but few banks would be buying. Losses would be large, and the number of bank failures would increase.
In practice, bank failures have been relatively infrequent. From the end of the Civil War through 1920 (after the Federal Reserve was established in 1913 but before the Federal Deposit Insurance Corporation was formed in 1933), the bank failure rate was lower, on average, than that of nonbanking firms. The failure rate increased sharply in the 1920s and again between 1929 and 1933, when nearly 40 percent of U.S. banks failed. Yet, from 1875 through 1933, losses from failures averaged only 0.2 percent of total deposits in the banking system annually. Losses to depositors at failed banks averaged only a fraction of the annual losses suffered by bondholders of failed nonbanking firms.
A survey of all failures of national banks from 1865 through 1936 by J. F. T. O’Connor, comptroller of the currency from 1933 through 1938, concluded that runs were a contributing cause in less than 15 percent of the three thousand failures. The fact that the number of runs on individual banks was far greater than this means that most runs did not lead to failures.
The evidence suggests that most bank runs were then and are today type 1 or 2, and few were of the contagious type 3. Because a type 3 run—a run on the banking system—causes an outflow of currency, such a run can be identified by an increase in the ratio of currency to the money supply (most of the various measures of the money supply consist of currency in the hands of the public plus different types of bank deposits). Increases in this ratio have occurred in only four periods since the Civil War, and in only two—1893 and 1929–1933—did an unusually large number of banks fail. Thus, market forces and the banking system on its own successfully insulated runs on individual banks in most periods. Moreover, even in the 1893 and 1929–1933 incidents, the evidence is unclear whether the increase in bank failures caused the economic downturn or the economic downturn caused the bank failures. As a result of the introduction of deposit insurance in 1933, runs into currency are even less likely today. The threat of runs from perceived troubled large banks, which have sizable uninsured deposits, to perceived safe banks serves as a form of market discipline that may reduce the likelihood of runs on all banks by giving them an incentive to strengthen their financial positions.