Too-many-to-fail and the design of bailout regimes

We analyze the design of bailout regimes when investment is distorted by a toomany-to-fail problem. The first-best allocation equalizes benefits from more banks investing in high-return projects with endogenously higher systemic risk due to more banks failing simultaneously. A standard bailout policy cannot implement the firstbest, as bailouts cause herding by banks. However, a targeted bailout policy that assigns banks to separate bailout regimes eliminates herding and achieves the firstbest. When such a policy is not feasible, targeted bailouts can be implemented by decentralizing bailout decisions to independent regulators. Our results have various implications for the optimal allocation of regulatory powers, both at the international level and domestically.