Research Discussion Paper – RDP 2019-10 Emergency Liquidity Injections

This paper compares the effectiveness of different forms of emergency liquidity injections, including secured lending (repo), unsecured lending and securities purchases. The model features an endogenous banking crisis, funding and market liquidity interactions, and fire sale externalities. Injection policies are compared by their effects on ex ante incentives and on ex post outcomes. The model demonstrates that lending to banks via repo can curb fire selling of relatively illiquid securities that are accepted as collateral, due to binding collateral constraints. The mitigated securities price depression, relative to an unsecured lending policy, counteracts the effects of fire sale externalities. This reduces banks' losses on illiquid securities without incentivising more liquidity risk-taking. Under an unsecured or secured lending policy, the authority can charge ‘penalty rates’ to deter liquidity risk-taking, but to be credible, lending should be long term so that repayments are due after liquidity conditions improve. Otherwise, the repayments can cause further liquidity distress, compromising the policy objectives. Liquidity injections via securities purchases cannot credibly be penalising, because the policy does not require banks to commit future income.