It is easy to think of large banks as tall city buildings. Banks are highly leveraged intermediaries, building credit volume on an equity-capital foundation. Houses, like banks, are worth more when located in dynamic cities where land costs are high. Builders naturally seek to economise on lot size, depth of stone foundations, and construction material. In a city, buildings are tall and close.
Bank capital is usually less than 5% of assets. Most bank credit is built on less committed funding, which may be withdrawn at various speeds.
Just as runs are the main threat to banking systems, the historical threat to cities was rapid fire propagation. There are interesting analogies.
Historically, stone and bricks were expensive building material. The most common house construction was in wood and thatch, which are less resistant to storms and fire.
History can be instructive in this regard. London in 1666 was a rapidly growing city. As population density increased, houses were made taller, and dense slums surrounded the mediaeval core of the city. Here is an abridged extract from the Wikipedia entry on “The Great Fire of London”:
John Evelyn in 1659 called London a "wooden, northern, and inartificial congestion of Houses," and expressed alarm about the fire hazard posed by the wood and the congestion. By "inartificial", Evelyn meant unplanned and makeshift, the result of organic growth and unregulated urban sprawl.
The City was essentially medieval in its street plan, an overcrowded warren of narrow, winding, cobbled alleys. It had experienced several major fires before 1666, the most recent in 1632. Building with wood and roofing with thatch had been prohibited for centuries, but these cheap materials continued to be used.1
What increased further the risk of fire propagation was the over-leveraging of upper floors. Typical six- or seven-storey timbered London tenement houses had "jetties" (projecting upper floors) on a narrow footprint at ground level, "encroaching" on the street to maximise the use of land by a bursting population. The gradually increasing size of upper storeys meant the top jetties all but met across the narrow alleys. The fire hazard was well perceived—"as it does facilitate a conflagration, so does it also hinder the remedy", but "the covetousness of the citizens and connivancy of Magistrates" worked in favour of jetties.
In 1661, Charles II issued a proclamation forbidding overhanging windows and jetties, but was largely ignored. Charles' sharper edict in 1665 warned of the risk of fire, and authorised both imprisonment of recalcitrant builders and demolition of dangerous buildings.
The Great Fire came one year later, raging over three days. It originated with a small flame, which spread irresistibly across the cramped wooden buildings. The fire did not spread fast, but it was relentless as it leapt across wooden buildings. It burned the entire City, St Paul’s Cathedral, and over 80% of the housing stock.2 The loss was immense.
The experience taught that house density, flammable construction material, and excessive interconnectedness were key causes of fire propagation.
What did the city elders do to avoid a repetition? Force households to keep a bucket of water, or mandate city houses henceforth to be built from bricks and stone? The correct answer is B. It was more expensive, but essential. Probably many Londoners objected to the higher cost. But buckets, while fine for villages and small fires, are not for cities of closely connected buildings. Only bricks and stone stop or slow down fires.3
The lesson from history is that preventing propagation is more important than focusing on how to fight the fire once it had spread. Prevention of risk propagation should be the mission of macroprudential policy, not just ex post resolution and crisis management. Aggregate risk largely arises from collective choices, not by accident. Were it just about ex post intervention, such as liquidity support, resolution, and bail-in, it would fail to address the root causes, and might still shift large losses on citizens. Ex post support works largely by transferring value from risk-averse long-term monetary savers to risk-taking short-term borrowers. While it boosts up risky assets ex post, it is an irresponsible policy overall.
Are Basel rules preventive? Fire prevention, water buckets, and hydrants
Basel III rules on higher capital requirements and tougher liquidity rules will increase resilience. These tools are of two types. Buffers are classic instruments which enhance individual resilience after a shock. In contrast, standards on fire-resistant construction targets (such as the net stable funding ratios) are a major innovation, and can target propagation risk directly. They are essential to move banks away from a failed business model, which encouraged huge gambles with massive risk externalities.4
By now it is apparent that water buckets are not enough for serious fires. It is doubtful anyway whether banks can self-insure on liquidity. In the recent European banking distress, liquidity buffers were quickly run down without stopping runs. Only central-bank hydrants slow down the runs, but if used indefinitely they ultimately debase the currency, and thus run out.5 The truth is that there are not enough safe assets in the world, given the accumulated flammable wood. Making the buckets larger by defining more assets as liquid will help, but not much. Liquidity runs rarely last only 30 days, so most regulators admit the liquidity coverage requirement provides at most a buffer to grant time to the central bank to decide on liquidity support.
In contrast, construction standards target aggregate propagation risk, offering proper prevention. The net stable funding ratio is the main innovation in liquidity-risk control, limiting the use of unstable (flammable) construction material. The long delay till introduction is justified because it forces a major change for bank strategies. In the last decade of abundant liquidity, banks became used to decide on lending ahead of funding, assuming away all funding risk. The assumption of an infinitely elastic supply of bank funding was shared by regulators and academics. That view has now been shattered.
Relying on sufficient long-term funding (insured deposits, medium-term credit) is more expensive, but also more resilient. It takes more time to raise than wholesale funding, which in good times is just a phone call away, but it is more stable. Such funding also has ex ante advantages. As long-term investors bear some risk, they are more concerned with bank risk than overnight lenders. Overall, the net stable funding ratio can induce more gradual credit expansion, more local funding and lending, and fewer under-examined global bets and carry trades.
Naturally, banks fiercely resist the net stable funding ratios, calling them very expensive, so regulators are under immense political pressure to dilute and postpone. General public opinion does not understand liquidity risk, though it affects the overwhelming proportion of bank funding. Therefore, failing to introduce net stable funding ratios in the Capital Requirements Directives ratification round will ensure its silent death, as public attention may become less alert.
Introducing net stable funding ratio requires a credible implementation to prevent avoidance or collective inertia. Industry resistance ahead of a long-delayed introduction is to be expected. The further banks are from the standards set for 2018, the tougher it will be to accept relaxation to avoid a credit crunch. Announcing common standards without the means to nudge banks towards compliance would be irresponsible.
Strong transitional tools must ensure that norms be perceived as credible. We discuss in sequence microprudential and macroprudential tools. Some do exist, but have been too sparingly used. Others, especially on the long-neglected liquidity-risk front, do not exist yet, but absolutely should.
Microprudential tools: An intervention ladder
A simple scheme for preventive intervention has tools of increasing intensity, as the intermediary falls below some capital or liquidity standard.
We seek here to provide a ladder of escalating measures, to be activated sequentially. The first stage arises once general prudential ratios on capital or liquidity are reached. The next stage arises once compliance is unsatisfactory, but the intermediary is still above some well-defined ratios. This calls for some escalation. In these stages, microprudential regulators naturally lead the intervention.
Reaching a stage when the need for liquidity support arises will require engaging some macroprudential oversight. Going-concern contingent capital should be triggered at this stage.
Once the intermediary approaches minimum conservation ratios, fiscal authorities need to take over the leadership of the intervention, since public resources are likely to be needed to forcibly recapitalise or to contain the spillover effect of default.
We attempt next to detail these stages of intervention.
First step: Warnings
At an initial stage, authorities tend to prefer discreet pressure, as disclosure of a breach may trigger adverse market reactions. In case of non-compliance, this calls for a delicate balance. Even in the US, where the Federal Deposit Insurance Corporation has a clear doctrine of prompt corrective action, in practice it has rarely been applied to any bank of any significant size. Extreme sanctions, such as the threat of licence withdrawal, are always possible, yet such ‘nuclear’ interventions are extremely rare for obvious reasons. Market reaction concerns calls for moral suasion, which may lead to forbearance. Taken to its limit, this logic implies complete abdication of preventive policy in the case of significant institutions.
The problem with standards is that a bank is either compliant or not. Rigid norms potentially create trigger points and resultant precipices. Are there smoother tools available to the regulator when warnings are not heeded?
Second stage: Charges
We advance here the case for using prudential charges once ratios deteriorate. These would take the form of a flow of payments commensurate to the extent of the breach. The aim is to induce timely adjustment, removing the incentive to delay and forbearance.
Introducing prudential charges would be a major innovation, and would lower the threshold for active preventive policy while allowing graduated pressure.6 They draw inspiration from the recognition that risk incentives are best contained by limiting the upside in good times rather than by threatening harsh punishments in distress, which are not credible because of limited liability or because of stability concerns.
Charges may be used also for aggregate liquidity-risk management. They should be flexible, so that they may be escalated when conditions would enable faster compliance, and relaxed when the banking sector comes under stress.
Third stage: Escalating charges
Charges induce incentives to adjust while allowing a smoother process over time, and may be less distortionary across banks. If charges fail to obtain enough response, they need to be raised at some intermediate ratios, signalling determination without necessarily introducing triggers or deadline effects.
If the intermediary has issued going-concern contingent capital, the conversion of such debt into equity should be triggered at some point in this phase. Delaying action on the ground of avoiding adverse market responses creates an unsustainable situation. Suppressing bad news at this stage only increases the chance of mistrust and a sharper response later. Timely disclosure at a stage where the situation is not yet critical, associated with concrete actions, is critical for preventive policy, and allows one to defuse or attenuate future risks.
Fourth stage: Direct constraints on payments
At some stage, once intermediate prudential ratios are breached, risk incentives become very skewed. Raising charges further would not help, as it would weaken solvency while enabling the bank to keep operating at a stage of elevated risk-shifting. Charges may also fail to stop overconfident bankers. Once ratios breach some minimum conservation ratios, there should be less reliance on charges – or even suspension of them entirely – and a shift to a general prohibition of bonus and dividend payments.
Final stage: Seizure or resolution
Seizure or forced recapitalisation by public means becomes a clear necessity below some prudential ratios, where the intermediary has neither the capacity nor the incentives to take corrective action. At this stage the resolution mechanism should come under the oversight of representatives of the Treasury, and bail-in instruments should be converted or written off.
Ultimately, the threat of bankruptcy ought to play a role in containing excess risk-taking, so reforms removing the too-big-to-fail principle have a clear preventive role. Yet resolution cannot fully remove the risk-shifting incentives created by limited liability and high leverage. Even if microprudential measures could achieve this, they cannot address the subtle risk externality created by liquidity runs. Shareholders may internalise direct losses from unstable funding, but not the external effect of their fire sales. This raises the issue of macroprudential prevention, to which we turn.
Macro tools need to be flexible to adjust to overall market circumstances. Raising capital in hard times is difficult, so Basel norms on countercyclical buffers demand that capital be built up during expansion years, with some leeway to be run down in stress times.7 On liquidity, however, there is no countercyclical instrument, and in fact no clear aggregate policy framework.
National prudential regulators should be empowered to charge ‘prudential risk surcharges’ on the gap between a bank’s current liquidity position and the new (Basel III–based) norms.8 Such charges would ensure reliable progress in adjustment as well as compensate for individual banks’ contribution to propagation risk. They would therefore serve both as transitional tools for microprudential regulators as well as macroprudential tools to contain aggregate systemic liquidity risk. Revenues would flow to a fund for financial stability, and may be used by authorities for direct capital support of individual banks.
Surcharges may start low, and be raised when circumstances allow to nudge banks into compliance. Critically, charges would enable a timely countercyclical liquidity policy. They would be lowered in hard times, but will also enable to push for faster adjustment in good times. At present, there are no clear tools for this purpose.
Surcharges would induce early adoption of safer standards, while giving banks the flexibility to plot their own path toward convergence. Adjusting the charges would be smoother than adjusting or postponing the ratios.
Finally, surcharges can be targeted better than higher interest rates, which hit everyone and not just the gamblers.
Consequences of the charges
The primary goal of the surcharges is to induce a longer bank funding maturity and reduced contingent outflows upon external shocks. Surcharges would create a wedge between shorter- and longer-term rates. Past evidence suggests that this wedge would not much affect the volume of bank credit, as it induces savers to take more term deposits.
By making it expensive for banks to use short-term funding, charges will affect the yield earned by investors on demandable (nonretail) claims. This will correct a major loophole. Right now, wholesale short-term funding is de facto insured but evades deposit insurance charges. Once regulations will be properly extended to shadow banks, charges are likely to lead to a cost increase for money-market funds and conduits. This will reverse a disintermediation trend which was driven solely by regulatory arbitrage.
A key question is how much costlier will credit become, and how reliable. The cost of credit will be raised in good times, at most by the amount of the charges, if banks were to pass the entire cost to investors. However, the cost and volume of credit would be more stable in bad times, for two reasons. First, banks will suffer less rapid outflows and would need to deleverage less, or less rapidly. Second, countercyclical charges will be lowered in times of credit crunch. On average, it may have little effect. Critically, since the crisis there is a consensus that volatile credit access is very costly for businesses and taxpayers.
More affected will be money-centre banks with few retail deposits, though they could adjust by choosing medium-term funding. On the supply side, money-market funds will find it more costly to operate as quasi banks. They may reduce their reliance on demandable liabilities, becoming intermediaries for more medium-term savings.
Implementation of liquidity policy in the EU
The ambition in the EU to achieve a high degree of harmonisation (the ‘single rulebook’) currently stands in the way of macroprudential activism, not least because there is to date no common macroprudential authority (as the European Systemic Risk Board has purely an advisory role). A common set of EU standards for microprudential liquidity policy will be defined soon under the Fourth Capital Requirements Directive; but at the macro level, the reality is that perceived sovereign and banking risk vary markedly across countries at present. As bank solvency is naturally linked by markets to domestic public solvency, it is obvious that the transition process to stable funding cannot be uniform across countries. National setting of prudential charges can introduce sorely needed national flexibility in the implementation process, reducing the rigidity imposed by monetary union.
Implementation of prudential charges in the EU the exact technical specifications of liquidity coverage requirements and net stable funding ratios are received in EU legislation, charges may be applied to both. Of the two, it is much more important to target individual convergence to stable funding, as other instruments exist to compensate for liquidity buffers, which also act as a form of indirect penalty.9 At present, EU countries run different liquidity frameworks, and will do so at least until 2018 in the absence of a common policy.
While set at the national level, they could represent a common EU framework for the transition, and would be coordinated by the European Systemic Risk Board.
International coordination of rate-setting is desirable. Unless an EU systemic risk agency is empowered to set them, they may differ across countries in the transition to common norms. But a level playing field requires that riskier banks face higher charges, or else competition is distorted.
It would have been desirable in 2005–07 for Spain and Ireland to have higher charges than, say, Germany, where there was no foreign credit-fuelled real-estate bubble. This flexibility would actually reinforce the cohesion of the Eurozone, reducing the rigidity imposed by a single monetary area.
A concern often raised in continental Europe on net stable funding ratios is that they make it harder to operate the universal bank model, since retail deposits may not be sufficient to support their lending. While the concern is legitimate, it is essential to stress three aspects. First, charges would be flexible, and can be graduated to contain the risk of a credit crunch. Second, they would reduce the risk of rapid deleveraging at the date of the introduction of the liquidity norms. But most importantly, the flow of credit to the real economy is more closely related to other regulatory features. In particular, capital-risk and liquidity-coverage ratios still penalise corporate loans over securitised or sovereign credit, even though the crisis proved these assets to have high correlation risk. If there are concerns on containing the squeeze on bank lending, it is more useful to work on the relative capital risk weights for securitised credit and commercial lending. Undermining standards essential for long-term stability would be irresponsible, as there are better means to soften pressure on bank credit volumes.
Unlike transaction taxes, liquidity-risk charges target risk creation. By this, they contribute not just revenues, but also increase financial and ultimately fiscal stability. In addition, the establishment of a common framework for liquidity management would contribute to the process of EU convergence to a common market.
Microprudential standards target risk choices and resilience at the level of individual intermediaries. In contrast, macroprudential policy needs to target propagation risk, aiming to prevent at least as much as to contain crises. As Andrew Crockett (2000) had said early on: “The received wisdom is that risk increases in recessions and falls in booms. In contrast, it may be more helpful to think of risk as increasing during upswings, as financial imbalances build up, and materialising in recessions”.
To shift emphasis towards preventive policy, rather than ex post intervention, preventive tools need to be provided to prudential regulators. This would ensure a clear mandate to act in a timely fashion to avoid excessive forbearance and risk-shifting across countries. Nowhere is the need more urgent than for liquidity risk, the critical transmission mechanism for both the 2007–08 and 2011 bank crises.
The EU ratification of Basel III offers a unique opportunity. Coordinated rate-setting ensures a common convergence process, in place of each country setting its own incompatible transition rules. Rate flexibility at the national level in the transition would also reinforce the cohesion of the Eurozone, reducing the rigidity imposed by a single currency. It would enable the authorities to contain credit risk creation in individual countries. At present, bank solvency is a national responsibility, but a common currency within the Eurozone causes unstable bank funding to spill over to other countries. A specific financial stability tool, distinct from interest rates, also avoids having the common monetary policy carrying a dual responsibility, which puts ambiguous pressures on the ECB.
Acharya, Viral, Arvind Khrishamurthy and Enrico Perotti (2011), “A Consensus View of Liquidity Risk”, VoxEU.org, 14 September.
Brunnermeier, Markus, Gary Gorton and Arvind Khrishamurthy (2011), “Risk Topography”, NBER Macroeconomics Annual 2011, Volume 26.
Crockett, Andrew (2000), “Marrying the Micro- and Macro-Prudential Dimensions of Financial Stability”, Bank for International Settlements working paper.
Perotti, Enrico and Javier Suarez (2010), “The simple analytics of systemic liquidity risk regulation”, VoxEU.org, 16 March.
Perotti, Enrico (2012), “How to stop the fire spreading in Europe’s banks”, Financial Times, 4 January.
1 The only stone-built areas were parts of the City where the merchants and brokers lived. It was surrounded by an inner ring of overcrowded poorer parishes which contained many fire hazards –foundries, smithies, glaziers', theoretically illegal, but tolerated in practice.
2 The Tower of London was saved only by ring-fencing, as soldiers blew up an entire neighbourhood to stop the advance of the flames.
3 Other measures included ruling out jetties, and to a lesser extent discourage density. In the end, London was rebuilt on the old street structure.
4 This strategy has been termed ’collecting pennies in front of a steamroller’. Accordingly, the recent crisis has been called ’the revenge of the steamroller’.
5 The real value of monetary creation is finite, and decreasing beyond some amount. It forces high losses on savers, hits incentives for long term investment, and ultimately results in less credit (less water pressure for productive or household use).
6 Regulators at present have a limited scope to impose charges. User fees are not risk-based. Fines are admissible only after a breach, so preventive use is ruled out. They can be challenged legally, and in practice, are imposed only upon fraud.
7 Solvency II may soon introduce a countercyclical regime for capital ratios in insurance. Initial proposals focused on tolerating lower standards in times of illiquid markets, but failed to require higher buffers in boom times. As in banking, there is broad support for forbearance when prices are distressed, but less for tighter standards when risk premia are too low.
8 Since central banks are the natural and better-informed regulators of aggregate liquidity, they would seem to be the natural authority for such charges. In some countries the authority may be vested in the microprudential regulator, in coordination with the monetary authority.
9 LCRs imply a marginal charge against less stable funding when bank funding spreads are high. However, the effect is market-driven and procyclical, so it is definitely not preventive.