Understanding Bank Runs

Featured in print Digest

The longer customers have had their money in a bank, the less likely they are to join the stampede to withdraw their funds.

The image of long queues of agitated customers clamoring to withdraw their deposits from failing banks traditionally has been associated with the Great Depression - that is, until the very recent bank runs in the United States (Countrywide, IndyMac), Britain (Northern Rock), and India (ICICI Bank). In their timely study, Understanding Bank Runs: The Importance of Depositor-Bank Relationships and Networks (NBER Working Paper No.14280), Rajkamal Iyer and Manju Puri analyze a unique database of minute-by-minute withdrawal activity at a besieged bank in India. Their analysis not only determines factors that propelled the bank run, but also points to policies that may help to mitigate the predilection for a bank run.

Iyer and Puri accessed customer activity information at a local bank in the Indian state of Gujarat following revelations in 2001 of a massive loan fraud perpetrated by the largest cooperative bank in the region. When the major bank collapsed, smaller cooperative banks experienced runs by panicked depositors. The bank in Iyer and Puri's study had no connection to the collapsed bank and in fact was quite solvent, but as the authors note, "depositors can run even in anticipation of a run."

Iyer and Puri moreover found a phenomenon they call the "contagion effects of bank run behavior," something akin to the spread of a disease. The researchers in fact applied methodologies designed for the study of how epidemics spread to analyze the behavior of the depositors. Just as a person might contract a disease via contact with an infected neighbor, so too is a depositor likely to withdraw money from the bank because of such activity in his or her social network. Iyer and Puri's minute-by-minute withdrawal data show that community networks are important. They use various measures of such networks, such as one's neighborhood, similar ethnicity in the neighborhood, and networks based on links through the introducer who helped open the bank account. In all cases, they find, if people in your network run, you are more likely to run.

Equally significant are the factors the researchers find that can mitigate the propensity to make a run on the bank. Deposit insurance helps, but only partially. An intriguing finding is that the length and depth of bank-depositor relationships are highly significant. The longer customers have had their money in a bank, the less likely they are to join the stampede to withdraw their funds. Similarly, if the customers have taken loans from the bank in the past, they are less likely to run in response to rumors about the bank's solvency. Even given the important influence of one's social network, Iyer and Puri find that the length and depth of a customer's relationships with the bank act as a dampening factor on the depositor's inclination to run.

From the bank's point of view, these results highlight the importance of relationships with a bank in influencing depositors' incentives to run. This suggests that one rationale to encourage cross-selling of deposits and loans to depositors is not simply to enhance revenues, but also to help protect the bank by acting as a complementary insurance mechanism. Moreover, the results of this study imply that allowing banks to provide an umbrella of products could strengthen the relationship with the depositor, which in turn could help reduce fragility.

These findings on the importance of bank-depositor relationships are provocative. Previous studies indicated that small banks generally supply more credit to small borrowers and give better terms. Iyer and Puri say that a side benefit of such lending is a reduction in the vulnerability to runs. Similarly, banks tend to give better terms to depositors who borrow from them. Iyer and Puri say this makes sense because such lending acts as a complementary insurance mechanism.

One important question that has not been addressed in prior literature is whether bank runs have long-lasting effects, even if the banks remain solvent. Iyer and Puri's analysis finds that most depositors who run do not return to the bank, suggesting that the effects of bank runs are indeed long lasting.

-- Matt Nesvisky