Contagious bank runs : experimental evidence

We conduct a laboratory experiment to examine under which circumstances a depositor-run at one bank may lead to a depositor-run at another bank. We implement two-person coordination games which capture the essence of the Diamond-Dybvig (1983) bank-run model. Subjects in the roles of followers observe the deposit withdrawal decisions of leaders before they make their own deposit withdrawal decisions. In one treatment followers know that there are no economic linkages between the leaders' and the followers' banks. In a second treatment followers know that there are economic linkages between the leaders' and the followers' banks. Our results suggest that deposit withdrawals are strongly contagious across banks only when depositors know that there are economic linkages between banks. The contagion of withdrawals is by a change in beliefs about bank asset quality and in beliefs about the behavior of other depositors, with the latter channel being more pronounced. Our results reconcile panic-based and information-based explanations of bank runs