Recent reforms give regulators broad powers to “bail-in” bank creditors during financial crises.
We analyze efficient bail-ins and their implementation. To preserve liquidity, regulators must avoid
signaling negative private information to creditors. Therefore, optimal bail-ins in bad times depend
only on public information. As a result, the optimal policy cannot be implemented if regulators
have wide discretion, due to an informational time-inconsistency problem. Rules mandating tough
bail-ins after bad public signals, or contingent convertible (co-co) bonds, improve welfare. We
further show that bail-in and bailout policies are complementary: if bailouts are possible, then
discretionary bail-ins are more effective.