Sequential decisions in a bank run model

Diamond and Dybvig (JPE, 1983) and the subsequent literature mod-elled bank runs as a simultaneous-move game, even though empirical evidence indicates that depositors have information about others ´ deci-sion. This paper introduces explicitly sequential moves into the Diamond-Dybvig model. Depositors decide consecutively whether to withdraw their funds or continue holding balances in the bank. If agents can observe the actions of all previous depositors, I show that, contrary to Diamond and Dybvig, there are no bank runs in equilibrium. However, when only with-drawals are observed (and depositors do not know their exact position in the sequence) bank runs re-emerge as possible outcomes. I also consider a third setup in which keeping the funds in the bank is unobservable, but depositors are allowed to make this decision observable, at a cost. If the cost is moderate, then there will be no bank runs. This result suggests that allowing for communication between the bank and the depositors can help prevent bank runs