Bailouts and Bank Runs

This paper analyzes the effects of government bailouts in a modified Diamond-Dybvig model. Following Keister (2010), my model includes both a private good and a public good. Bailouts are assumed to crowd out pub- lic good provision and improve the ex-post allocation of resources during a bank run. I introduce two major innovations to this model. The first is the (realistic) assumption that the depositor and hence her bank takes the public good provision as given. Thus, the bank does not see the effect of its deposit contract on the level of public good provision. The wedge caused by the difference between the banks and the government's objectives results in excessive systemic risk. The second innovation is to determine the run prob- ability by using the global game approach in Goldstein and Pauzner (2005), making the run probability endogenous. Bailouts increase the ex-ante run probability through two channels. The first channel is moral hazard: Runs are less costly for banks when there are bailouts. Hence, banks take on more risk than is socially optimal. The second channel works through the change in the depositor's incentives to run: Bailouts increase the probability that a depositor will get her money if she participates in a run, thus increasing the likelihood of a run