How Noisy should a Noisy Signal be: A Model of Bank Runs

In the literature on bank runs where depositors decide whether to withdraw early from the bank or not based on the noisy signals they receive about the future returns, a unique equilibrium is established with a threshold level below which depositor would withdraw. However, these papers assume precise information. In reality noise levels need not be very small. The information that is available to the depositors can be endogenised. This paper finds that to either minimise the probability of a bank-run or maximise the expected utility of the depositors, there should be high transparency of the banks' long-term returns.