Andrew Hauser: Between feast and famine - transparency, accountability and the Lender of Last Resort

Speech by Mr Andrew Hauser, Executive Director for Banking, Payments and Financial Resilience of the Bank of England, at "The Lender of Last Resort: An International Perspective", a conference sponsored by the Committee on Capital Markets Regulation, Washington DC, 10 February 2016.

The views expressed in this speech are those of the speaker and not the view of the BIS.

I am grateful to Nicholas Butt, Ronnie Driver, Tarkus Frost, Nicholas Govier, Rachel Greener, Andrew Lavender, Tim Taylor and Caspar Worthington for their help in preparing these remarks.

Introduction

Of all the actions taken by central banks during the financial crisis, some of the most contentious involved the extension of huge sums in liquidity assistance to banks and other financial institutions. Those actions helped prevent global economic and financial meltdown. But they also triggered intense public debates around the globe about the appropriate scope and nature of central bank powers.

One strand of this debate is led by those who feel that central banks lent "too much", too often or too freely during the crisis, using public money - so the allegations go - to support institutions of dubious solvency, sometimes in secret, protecting bankers from their own mistakes, and thereby stoking moral hazard for the future. The main policy conclusion from such thoughts is that central banks' Lender of Last Resort (LOLR) activities should be constrained.

But another strand in the debate draws the opposite conclusion from the crisis - focusing on the risk that central banks might lend "too little", too rarely or too late, amplifying what might begin as short-term liquidity shortages into deeper or more persistent solvency concerns, and subsequently causing financial institutions to over-insure, reducing their capacity to lend. The main policy conclusion from this way of thinking is that central banks need stronger toolkits for detecting and responding to liquidity crises.

These (deliberately stylised) perspectives seem diametrically opposed. National debates have tended to lean one way or the other, depending on each country's specific crisis experiences. But, in truth, it is vital we pay heed to both.

Without an effective LOLR framework, economies cannot hope to foster strong and sustainable growth over the long run. History shows that, sooner or later that growth, and the employment it supports, will be threatened by an unexpected withdrawal of liquidity from key parts of the financial system, unconnected to the underlying quality of real assets in the economy. Historically that liquidity risk resided primarily in the banking sector. But as economies migrate towards more market-based forms of finance, liquidity risk is increasingly likely to originate (or be concentrated) elsewhere: including broker-dealers, central counterparties (CCPs), or traded instruments themselves. The herd mentality that can develop during crises can sometimes only be tamed by well-designed intervention - be it covert or overt - using a credible backstop. Countries lacking such backstops are likely to sustain large-scale, and preventable, economic damage. So the risk of central banks having "too little" scope to carry out LOLR is real.

At the same time, the public has a legitimate interest in avoiding "too much" LOLR. We cannot afford for financial institutions to view liquidity support as a substitute for prudent financial management: central banks must not be "lenders of first resort". And lending to institutions that are in need of capital rather than liquidity is the wrong solution. Given the uncertainties involved, neither outcome can be avoided with certainty. But it is right to seek robust safeguards against them.

A well-designed LOLR framework must therefore get the balance right between lending "too little" and "too much":

- To help avoid "too little" LOLR, central banks must be given clear responsibilities and the tools and powers needed to carry them out, including the ability to lend to individual institutions, covertly if necessary to maintain financial stability.

- To help avoid "too much" LOLR, central banks must have strong risk management, be ring-fenced from special interests (in public or private sectors), have a clear public mandate specifying their scope for action, and be subject to strong transparency and accountability requirements.

There are powerful parallels between these design challenges, and those for monetary policy. Like monetary policy, LOLR is vitally important to an economy's health, but involves intrinsically complex, technical judgments - including the solvency or otherwise of potential borrowers, the pricing, scale and maturity of loans, and the appropriate collateral. It is important that these judgments are made by experts. But, also like monetary policy, the consequences of these expert judgments have potentially wide-ranging effects on society. So they should only take place within a clear public framework of delegated authority. This combination of expert discretion within public parameters is sometimes termed "constrained discretion".

The large scale LOLR interventions undertaken by the UK authorities during the financial crisis played a significant role in preventing a much deeper economic and financial collapse. But they also exposed a number of shortcomings in the Bank's pre-crisis frameworks. Through the course of the crisis, and subsequently, that framework has been transformed through an extensive series of public reforms¹. In the rest of my remarks today I want to draw out from that work four particular lessons from the crisis that have helped the UK in its quest to plot a more effective course between "too little" and "too much".

Four lessons from the crisis in the United Kingdom

The first lesson drawn by the UK authorities was that a key part of reducing the risk of conducting "too much" LOLR is having effective and coordinated macro- and micro-prudential supervisory and resolution regimes.

Pre-crisis, the UK lacked a comprehensive micro-prudential liquidity regulation framework, and banking supervision did not focus enough on liquidity management. This meant that many banks took too much liquidity risk. There was no system for conducting system-wide stress tests for determining the appropriate level of bank capital, and micro-prudential capital requirements were inadequate, so banks tended to be undercapitalised. And there was no regime for resolving insolvent firms in an orderly way.

This placed the Bank of England in a difficult position. In particular, there was a heightened risk that LOLR for a particular bank would turn out to be solvency support, with no credible exit strategy. In response, the Bank adopted a conservative approach to providing liquidity assistance in the early stages of the financial crisis. Ironically, therefore, concerns about the risks of lending "too much" led to perceptions the Bank would lend "too little".

Post-crisis reforms have changed this situation significantly. Banks are now subject to heightened liquidity and capital regulation, and annual stress testing. And the UK has a statutory resolution regime. Taken together, these steps mean that LOLR no longer has to shoulder as much of the burden of preventing insolvency ex ante, or dealing with it ex post. As such, the Bank is now able to state, as a presumption, that all banks and building societies meeting the supervisory threshold conditions for authorisation may sign up for, and access, the Bank's published liquidity facilities². So changes designed to reduce the risks of lending "too much" have actually allowed the UK to offer a more effective LOLR scheme.

The second lesson drawn in the UK was that, to avoid the twin pitfalls of "too little" and "too much", the central bank should have access to the best possible information to judge solvency; and wherever possible be ready and willing to lend on pre-defined terms, against pre-defined collateral.

Prior to the crisis, the Bank said very little about the terms on which it would be willing to provide LOLR, or the collateral against which it would lend. Indeed the provision of liquidity insurance wasn't explicitly acknowledged in its public framework, in large part with a view to minimising moral hazard. In addition, the Bank had limited access to information on the financial position of firms to whom it might have to provide LOLR, because banking supervision was carried out in the Financial Services Authority, a separate organisation.

In practice, efforts to tackle moral hazard had little effect: in the absence of formal regulatory requirements, banks took minimal steps to self-insure against liquidity risk. Yet at the same time, the Bank was poorly placed in the early stages of the crisis to launch well-structured liquidity assistance rapidly, either to the market as a whole or to individual firms. It lacked the information to evaluate firms' solvency. And, where LOLR was required, it lacked well-defined pricing or maturity schedules, pre-arranged legal documentation or - most importantly of all - a clear ex ante understanding of the amount and quality of suitable collateral on banks' balance sheets against which it could lend.

Prudential supervision has now returned to the Bank, facilitating a more coordinated approach to financial crisis management, horizon scanning and proactive contingency planning, and improving the Bank's ability to evaluate solvency ahead of any LOLR. And the bulk of the Bank's liquidity insurance toolkit has progressively been made public since the height of the crisis. The Sterling Monetary Framework (SMF) now contains a range of facilities with documented price, maturity and quantity schedules, including a regular auction-based market-wide Indexed Long-Term Repo facility, a bilateral on-demand Discount Window Facility and a discretionary Contingent Term Repo Facility. All of these facilities can be used against a wide range of eligible collateral, including securities and raw loans. But this collateral must normally be "pre-positioned" with the Bank so that it can be credit assessed, priced and haircut well before any need to use it arises, rather than in the heat of a crisis. The most recent published data show that nearly half a trillion pounds of collateral have been delivered to the Bank in this way.

The existence of a wide range of facilities, coupled with the presumption of access, has helped reduce concerns about the Bank lending "too little". But at the same time, it has also strengthened safeguards against lending "too much". Counterparties can see clearly that the Bank's facilities are priced to incentivise the use of private markets in normal trading conditions. And the detailed pre-positioning process, underpinned by published collateral eligibility criteria, provides assurance against the possibility that the Bank is taking sub-standard or overpriced collateral.

The third lesson from the UK is that richer tools for LOLR require strong governance and accountability mechanisms, to reassure society that the central bank will operate according to clearly established procedures, and will report on its activities ex post.

Assurance that the Bank will not lend "too much" under its published facilities (backed by its own resources) is provided through a combination of ex ante safeguards and ex post transparency. The ex ante safeguards include: the new liquidity, capital and resolution regimes; the link between access to facilities and the threshold conditions for supervisory authorisation; the requirement to pre-position collateral; and the pricing schedule. Ex post transparency of aggregate amounts lent is provided immediately for market-wide liquidity provision, and with a lag for any usage of the bilateral Discount Window. The Bank also publishes an annual report on the operation of its published facilities³.

The Bank's published facilities now provide a rich toolkit for responding to a wide range of sterling liquidity needs. But not every eventuality can be predicted in advance - and UK institutions might on occasion require assistance in larger size, against alternative collateral, or in currencies other than sterling. The terms and conditions for any bilateral lending done outside of the Bank's published facilities - so-called "Emergency Liquidity Assistance (ELA)" - would use as a starting point those in the SMF, including pricing and collateral eligibility and haircuts. But, recognising the potential for discretionary deviation from these terms, and the large sums of money that such lending can involve, ELA is also subject to enhanced governance and accountability arrangements, published in a Memorandum of Understanding (MoU)⁴ between the Bank and HM Treasury.

These arrangements are somewhat shaped by history. In his autobiography⁵, the then Chancellor of the Exchequer Nigel Lawson records his deep annoyance at being given only half an hour's notice by the Bank in October 1984 to decide whether to provide a government guarantee to back a large loan to Johnson Matthey Bankers, "a bank of no great consequence", which the Bank had been working for some time to save.

The 2012 MoU sets out a panoply of safeguards against the repetition of such an incident, including a requirement for timely notification from the Bank to HM Treasury of any risk to public funds, and analysis of the systemic impact of alternative policy options. The MoU makes it clear that the Chancellor has sole responsibility for the use of public funds, approves all ELA (including the terms on which it is provided), and may if necessary direct the Bank to extend ELA - though this is a backstop power, whose use is not expected to be considered in the majority of crisis situations.

Where it is judged necessary to keep ELA covert for a period in order to minimise the risks to financial stability, key Parliamentary Chairs will be briefed orally by senior officials on any government indemnity attached to such lending - and full disclosure to Parliament would be made at the earliest subsequent opportunity once the risk of damage to the public interest had passed.⁶

Covert operations, particularly those involving a fiscal component, are often cited as the most troubling forms of LOLR by those concerned that central banks have a bias to lend "too much". In the United Kingdom, the disclosure of large covert loans to HBOS and RBS in late 2009 (more than a year after the loans were advanced) triggered a public debate about the appropriate boundaries for covert action. At the same time, the consequences of premature disclosure were vividly illustrated by the case of Northern Rock in September 2007, when a press leak led to long queues at branches, and required the government to issue a blanket guarantee to ease public concern. Where liquidity shortages are widespread, or central bank support would otherwise be seen as clearly confidence-enhancing, there are strong advantages in central banks conducting overt operations. But the capacity to operate covertly, subject to accountability safeguards, remains an essential part of the central bank toolkit.

The final UK lesson is that LOLR frameworks must be flexible enough to respond to changes in the structure of financial markets if societies are to guard against the twin perils of lending "too little" or "too much".

At the start of the crisis, the UK's LOLR toolkit was poorly equipped to deal with the needs that emerged, and had to be innovated rapidly. The changes introduced in recent years have reduced the risk that the Bank would lend "too little" if the same risks re-occurred.

In all probability, however, the next crisis will not be like the last one. It is important that policy steps taken to prevent central banks lending "too much" to today's financial institutions do not also impair their ability to maintain financial stability as the structure of financial markets adjust and policy needs change.

One important change is the shift from bank to market-based finance. Historically, central banks have relied on the assumption that, even if liquidity shortages emerge in the non-banking sector, banks are likely to remain an effective conduit for LOLR. The efficacy of that channel may be less reliable in the future, however. For that reason, the Bank of England has extended access to its published facilities to broker dealers and CCPs overseen by its supervisory arm the Prudential Regulation Authority, and has ensured it faces no technical obstacles to providing ELA if required. The Bank has also set out principles under which it might operate as market maker of last resort - consistent with the role it played in sterling corporate bond and commercial paper markets during the crisis.

Conclusions

Ensuring central banks lend neither "too little" nor "too much" in pursuit of financial stability is vital if countries are to retain LOLR toolkits that can both protect their economies against genuine liquidity risks whilst maintaining broad-based public support and legitimacy.

In recent years, reforms to the Bank of England have focused on minimising the risk that it might lend "too little", significantly increasing the Bank's ability to provide structured LOLR assistance, to a wider range of counterparties, against a wider set of collateral, at longer maturities and somewhat lower prices. But at the same time, safeguards have also been strengthened against the Bank lending "too much": better supervisory and resolution frameworks; transparent terms for access, pricing and collateral; and stronger accountability and governance.

The UK's arrangements are no panacea. The decision to lend or not to lend will always be a difficult one, conducted with imperfect information and in the midst of a crisis. Ex post judgments on whether the central bank lent too much or too little must recognise this constraint. But the UK's experience can I hope show one way to address the risks of lending "too much" through enhanced accountability whilst also still allowing central banks the "constrained discretion" to exercise expert judgment in the pursuit of financial stability.

I look forward to our discussion of these issues today.

Thank you.

---