Let the Worst One Fail:A Credible Solution to the Too-Big-To-FailConundrum

Abstract

We study time-consistent bank resolution mechanisms. When interventionsare ex post efficient, a government cannot commit not to inject capital into thebanking system. Contrary to common wisdom, however, we show that the gov-ernment may still be able to implement the first best allocation because it canuse the distribution of bailouts across multiple banks to provide ex ante incen-tives. We show that the efficient mechanism has the feature of a tournament. Ifeach bank’s net transfer from the government can only depend on its own per-formance, no credible mechanism can prevent maximal risk-taking by all banks.In stark contrast, using relative performance evaluation during the crisis canimplement the first-best risk level while remaining credible. In particular, weanalyze properties of credible tournament mechanisms that provide support tothe best performing banks and resolve the worst performing ones. We extend ourframework to allow for contagion and imperfect competition among banks. Ourmechanism continues to perform well if banks are partially substitutable and ifbanks are heterogeneous in their size, interconnections, and thus systemic risk,as long as bailout funds can be targeted to particular banks.