Herding, asymmetric information, and bank runs

In the first two chapters, I model the withdrawal decision of an individual depositor as dependent upon her private information about the bank's portfolio and her observations of the withdrawals by other depositors. Depositors' beliefs about bank fundamentals are based upon privately observed signals about bank portfolio productivity, recursively updated through public observation of withdrawals.; In the first chapter, I find that---given a demand-deposit contract---a bank run occurs as a result of a herd of withdrawals when beliefs of the depositors about the bank's portfolio are affected unfavorably from observing a large number of withdrawals. Computed examples indicate that the optimal demand-deposit contract can permit herding runs because a run-admitting contract not only provides more liquidity to the depositors to insure against liquidity shocks, but it also encourages depositors to reveal the signals they receive.; In the second chapter, the bank is allowed to pay the depositors contingent on the publicly observed withdrawal history. I show that in a two-depositor, two-informative-stage economy, there exists a perfect Bayesian equilibrium. Depositors withdraw if their beliefs are less favorable than some threshold, and wait otherwise. The contract that the bank offers ex-ante may encounter a creditability problem ex-post. Computed examples show that the optimal contract, either with or without commitment, can permit bank runs.; In the third chapter, I extend the analysis of panic-based rums to include an asymmetric-information, extrinsic randomizing device. Depositors observe different, but correlated, signals about the stability of the bank. I find that if the signals are highly correlated, there exists a proper correlated equilibrium for some demand-deposit contracts. In this equilibrium, depending upon the realization of the signals, either a full bank run, or a partial bank run, or no bank run will occur. Computed examples indicate that in some economies, if the probabilities of fall runs and partial rims are small, a demand-deposit contract that tolerates full runs and partial runs is better than the run-proof contract. This correlated equilibrium obtains in a broad class of banking mechanisms which includes partial suspension of convertibility.