VoxEU Column Financial Markets

Liquidity insurance

Correlated liquidity risks caused subprime mortgage problems to spread widely and sow panic that led to the credit crisis. This column proposes a mandatory liquidity charge to insure against collective bank runs in the future. It argues for charges proportional to securities’ maturity mismatches so as to discourage practices that create systemic risk.

Capital requirements were aimed at absorbing asset losses, not coping with correlated liquidity risk. Securitisation runs around this Maginot line by shifting long-term assets in outside legal boxes funded by short-term markets. Yet banks kept essentially most risk through back-up credit lines, which forced all risk back when subprime mortgages were repriced. Yet the real damage came once panic spread, with losses well beyond what subprime prices would justify. Why?

Securitisation relied on extreme maturity mismatch, which led to rapid spreading of panic. Wholesale lenders pulled out fast, forcing fire sales across other markets, in turn causing margin calls and more panic in an accelerating spiral.

Charging for liquidity risk

Leading proposals aim at higher capital requirements indexed to asset growth, total leverage, and maturity mismatch. To deal with both asset and rollover risk, a second tool is needed. In our new CEPR Policy Insight No. 31, we propose a liquidity and capital insurance arrangement for systemic events, which is simple, offers better incentives, and would have stronger political support.

A mandatory liquidity charge should be paid continuously to a supervisor by all insured intermediaries. In exchange, an Emergency Liquidity Insurance Fund would provide immediate liquidity support, guarantees on uninsured funding, and capital injections upon collective (not individual bank) runs. Prudential restrictions on bank choices, e.g. on compensation and dividends, would be activated.

This applies the principle of Pigouvian taxation for pollution by discouraging bank strategies that create systemic risk for everyone. The charge should be increasing in the maturity mismatch between assets and liabilities, possibly equalising funding costs up to one-year maturity. Insured retail deposits would be exempted.

Systemic risk – the simultaneous realisation of correlated tail risk – is hard to estimate, as different asset classes may trigger it. In contrast, maturity mismatch is easy to compute. As liquidity runs are present in all systemic crises affecting banks and in fact reinforce high correlation, discouraging rollover risk reduces the speed of panic and systemic risk creation.

Liquidity charges would reduce reliance on short-term market funding and thus the speed of panic upon runs. Wholesale short-term funding has a low cost because it absorbs little risk, which is shifted to capital and taxpayers. Reducing this bias would make banks internalise the potential damage caused to others.

The charge for liquidity risk could be seen as an insurance premium, a pre-payment for contingent support. It would make emergency intervention politically more acceptable.

Why are higher bank capital ratios not a solution to liquidity risk? First, banks’ own capital would need to be huge during normal times. This has several disadvantages. Shareholders see bank capital as an asset to which they are fully entitled. Banks with plenty of capital on their books may “lever it up” through riskier investment strategies. Large accounting capital may hinder seizing a bank ahead of default, as a violation of private property. In contrast, insurance is cheaper as it arranges for a contingent injection of capital and liquidity in systemic crises only.

Liquidity insurance may encourage a shift to short-term funding in a shadow banking sector. This would not create systemic risk if unregulated intermediaries enjoy limited recourse to banks. The scheme should fully charge credit lines to intermediaries with mismatched funding.

Note: This column was prepared for testimony to the Select Committee for Banking Regulation and Supervision, House of Lords, February 24, 2009

210 Reads