The Subprime Crisis and the Lender of Last Resort (ARI)

The Subprime Crisis and the Lender of Last Resort (ARI)

Theme: This ARI reviews the traditional doctrine regarding the role of central banks as lenders of last resort, in view of the financial crisis which began in the summer of 2007.

Summary: This ARI reviews the traditional doctrine regarding the role of central banks as lenders of last resort, in view of the financial crisis which began in the summer of 2007. The systemic nature of the crisis and its duration have forced central banks to adapt some of the orthodox prescriptions, such as lending at a penalty interest rate. Broad and flexible mechanisms, like those of the ECB, have proved to be more adequate for this kind of crisis. However, one year after its inception, the tension persists, indicating that the problems are global and that no central bank has been able to resolve them satisfactorily, despite the diverse range of reactions.


The crisis unleashed in the subprime segment of the US mortgage market has spread unexpectedly and persistently to international interbank markets, whose liquidity had hitherto been taken for granted, and which perform a pivotal role in monetary policy transmission mechanisms. Central banks have had to intervene with substantial amounts and for a considerable period of time, as lenders of last resort, sometimes contravening orthodox prescriptions. The traditional doctrine did not envisage the possibility of a prolonged systemic liquidity crisis in the interbank markets.

Furthermore, the spread of the crisis to engulf solvency issues at some institutions (Northern Rock in the UK, Bear Stearns and, more recently, Fannie Mae and Freddie Mac and AIG in the US) have given rise to a controversy regarding the role of the central bank versus that of the Treasury, liquidity injections versus capital injections, conditions for using tax-payers’ money and moral hazard. In this connection, it is interesting to recall the premises of the traditional doctrine on the lender of last resort.

The Traditional Doctrine
The role of the lender of last resort (LOLR) is pivotal for central banks, and it lies at the very origin of their creation. In one of the classical treatises on central banking, Lombard Street (1873), Walter Bagehot established the bases of the traditional doctrine, according to which the mission of a central bank in a liquidity crisis is to lend without limit, at a penalty interest rate, with adequate collateral, valued at a price between the pre-banking panic price and the value (usually much lower) after the crisis has erupted.

The subprime crisis called into question some of the basic tenets of this traditional doctrine.

The Concept of Liquidity
The role of the LOLR has always been closely linked to the concept of liquidity. The central bank’s action is justified by a market failure which means that an institution with a liquidity problem, but not a solvency one, is unable to obtain liquidity from its usual private counterparties in the market. Obviously, the problem is compounded when it affects an entire segment of the financial system, or indeed the entire system. However, most recent literature on this issue has been based on the assumption that the interbank market would work efficiently, even in times of stress.

One of the most unexpected aspects of the subprime crisis has, in effect, been that the interbank market (the quintessentially liquid market) has been subjected to turmoil and insufficiencies for such a prolonged period (see Chart 1). This illustrates the difficulties in defining and delimiting the concept of liquidity. According to the traditional definition, an asset is liquid if it can be quickly transformed into cash, without its holder having to incur losses as a result of the swiftness of the transaction. A related concept is that of funding liquidity risk, defined as the incapacity of a financial intermediary to renew its debts as they mature. In this crisis, both concepts of liquidity have been affected (see IMF, 2008), which raises interesting questions regarding what determines liquidity in the various markets and how this evolves over time.

Chart 1. 12M LIBOR-OIS spreads in EUR, GBP and USD (bp)

One of the provisional lessons of this crisis is that financial regulation has perhaps focused too much on solvency and capital, and has neglected liquidity, which was taken for granted in the most developed markets (UK Treasury, 2008). Financial markets have an increasingly significant role versus intermediaries and the liquidity of the instruments traded therein is difficult to establish a priori, because it can endure sharp and unexpected alterations, which break with normal patterns, and it is difficult to protect against this. It is increasingly necessary, and increasingly complex, to conduct stress exercises and draft contingency plans vis-à-vis the risk of illiquidity.

Banking Panic versus Financial Market Panic
In a traditional banking run, customers rush to their bank branches to withdraw their funds amid suspicions that the institution may not be able to meet its commitments. It is interesting to note that, although we have seen this kind of phenomenon in this crisis (Northern Rock), the latter has rather been characterised by the evaporation of liquidity in entire segments of the financial markets (for example, some structured products), which, coupled with the need to value portfolios as market prices, has triggered copious losses at the financial institutions most exposed to these instruments. In some cases (Bear Stearns), exposure to these markets has led counterparty risk to soar, not only in unsecured loans, but even in secured loans. In other words, the uncertainty has spread to segments which it was not expected to reach, evidencing new modalities of systemic risk, that the authorities do not have instruments to tackle, among other things because institutions like Bear Stearns were not counterparties of the Fed until the reforms implemented by the latter in March 2008 (see Geithner, 2008).

Liquidity Injection versus Easing of Monetary Policy
In principle, the LOLR is designed to help specific institutions, but it cannot be ruled out that several institutions might be affected by liquidity problems at the same time, or that these might affect entire segments of the banking system, or indeed the system as a whole. In cases where a liquidity crisis affects the market as a whole, the necessary injection often involves a cut in interest rates, especially if the problems are persistent. The LOLR’s role is confused in these situations with monetary policy easing.

One of the most significant differences between the reaction of the Fed and that of the ECB to tensions in financial markets in recent months is that, while both central banks have pumped liquidity generously into the market, at various maturities and, in the case of the Fed, with a range of instruments, intermediaries and collateral that has broadened as the crisis worsened, the Fed has also cut interest rates significantly, while the ECB has actually raised them (Chart 2). These different reactions can be explained by various factors: (1) the problems in the real estate sector and the vulnerability of the financial sector in that respect, as well as the impact of the crisis on the real economy, were worse in the US than in Europe, at least initially; (2) the Fed is more sensitive than the ECB to financial stability targets; (3) among monetary policy targets, the Fed pays more attention than the ECB to growth and employment than to inflation; (4) the Fed follows a risk-management approach to monetary policy which means that, in times of stress, it does not necessarily aim at the core scenario.

Chart 2. Benchmark rates: ECB, Fed and Bank of England

The persistence of risk premiums in both the dollar and euro interbank markets appears to show: (1) that liquidity injections per se do not solve confidence problems or tackle higher risk aversion, although they might mitigate them; and (2) that it is by no means clear that interest rate cuts have helped improve the situation in the US money markets as compared with Europe (although they will certainly have an impact, with the usual delays, on the real economy and on the profitability of banks).

Who Should have Access to the Central Banks’ Liquidity?
A theme that has been quite central in the debate on lender of last resort (and which was already discussed during the long-term capital management or LTCM crisis in 1998) is what kind of institutions should be counterparties of central banks in regular operations to inject or drain liquidity. Again, the contrast between the ECB and the Fed is interesting (although it is worth recalling that in the European financial system banks perform a more central role, which no doubt simplifies things). In the Euro zone, the ECB operates with a relatively broad range of counterparties, while the US system hinges on the transmission of liquidity via a relatively small number of market makers. When transmission between these market makers and other systemic operators who do not have access to the central bank’s facilities failed, the Fed found that the liquidity it was injecting was not reaching the damaged segments of the financial system, and it was therefore obliged to extend access to its discount windows, in particular to investment banks and, more recently, to government-sponsored mortgage agencies (Fannie Mae and Freddie Mac).

This has ignited a fascinating debate. Who should have access to the central banks’ liquidity? Institutions that are supervised by the central bank? Institutions that are supervised by a public body, even when this is not the central bank? Institutions that generate means of payment? (and, in this case, what are ‘means of payment’?) Institutions protected by deposit insurance? Institutions whose lack of liquidity generates systemic risk?

In short, these questions concern the very raison d’être of banking supervision and the ultimate justification of the role of LOLR. Banks enjoy certain privileges linked to their role as generators of means of payment and their vulnerability to banking panics, such as access to the central bank’s liquidity, deposit insurance and entrance barriers to the banking business. These privileges have traditionally been compensated by stringent supervision and financial regulation. However, in a world in which intermediaries are increasingly being dispensed with, where the concept of means of payment is not clear (indeed not even the concept of banks is clear), where panics can be unleashed in specialised segments of financial markets and where supervision may be the task of the central bank or another body, the role of the LOLR is hugely complicated, among other reasons because it is more difficult to distinguish between liquidity problems and solvency problems at individual institutions and because in the case of systemic crises and loss of confidence in traditional counterparties there is no guarantee that the liquidity will reach the affected segments of the financial system.

This debate has given rise to a profound reflection on the US model of supervision and regulation, leading to a proposal by the Treasury to reform it (US Treasury, 2008) which must nevertheless complete a long legislative process in the next few years.

Central Bank Liquidity Injection Instruments
A similar debate is raging in regard to liquidity injection instruments. No model has worked properly in this kind of crisis, and all central banks have had to adapt to some extent. Accordingly, they have injected liquidity in longer than usual maturities (for example, the ECB has injected significant amounts to three months, when the normal period is one week), because it was in these periods where the problems were concentrated and in order to afford stability to financial institutions that were in trouble.

The broader the scope of a central bank’s normal instruments for injecting liquidity, the less need it will have to resort to exceptional instruments. In principle, the ECB had the broadest range of instruments, which explains why it has had to introduce fewer novelties, but this did not prevent it from resorting to ad hoc interventions, like the other central banks. The facilities and mechanisms designed to tackle this kind of problem have scarcely been used, and this should lead to a reflection on just how suitable they are. Specifically, the ECB’s credit facilities and the Fed’s discount window, designed for crises at individual institutions, at a penalty interest rate, have proved inadequate because, on the one hand, a massive injection at a higher interest rate would have triggered additional tensions in interbank rates and, on the other, because the banks were reluctant to use them because of the ‘stigma’ associated with them.

The precedent set by the US’s Savings & Loans Crisis in the early 1990s points in the same direction: penalty interest rates in LOLR operations does not make sense in a systemic liquidity crisis, because it can even compound it and encourage even more risky strategies at institutions in the worst positions (gamble for resurrection).

International Harmonisation of Lending of Last Resort
The international dimension of the LOLR policy is one of its most controversial and delicate aspects. As illustrated by the case of the rollover to Y2K, if the central bank is reasonably sure that it is a liquidity problem, and not a solvency problem, it should not have any problem lending to subsidiaries (which in principle should be treated like national institutions) and even to branches of foreign institutions, especially if their collateral is high quality. However, in cases of LOLR, collateral is often not optimum quality and there is usually a reasonable margin of doubt as to whether the central bank will eventually be forced to incur losses.

In principle, the responsibility of the home country in supplying liquidity is appropriate in the case of branches, not subsidiaries. In a continuum of liquidity versus solvency problems, the closer to the extreme of insolvency we move, the greater the reasons for the home country supervisor, regulator or central bank to act (and vice-versa, the closer we get to a problem of illiquidity, the more reasons there are for authorities in the host country to act).

In highly integrated international financial markets, where the major institutions have access to various lenders of last resort, in various currencies, to meet their needs, there can be something of a free-rider problem, if, for example, a central bank in a country whose banks have a significant foreign presence is more cautious, and this forces its institutions to access loans in their destination countries, shifting access to liquidity for local institutions, or making it more expensive (if there are quantitative limits on the provision of liquidity). Without advocating a standardisation of LOLR policies, which would be unrealistic, some harmonisation, for example, of collateral policies would help mitigate this problem (see IMF, 2008).

In this connection, it is interesting to compare the changes in the Bank of England’s and the Fed’s collateral policies with those of the ECB. The Bank of England and the Fed, whose definition of collateral was narrower, have broadened it to meet the liquidity requirements of their markets; the ECB (with a broader definition) recently made it more restrictive (see Afi, 2008), in part obliged by abuse on the part of international banks, reflecting pressure for convergence deriving from international arbitrage between financial centres. Accordingly, although convergence has not been the result of legal harmonisation that some international bodies, like the IMF, called for, a certain de facto convergence has taken place.

The most complex cases of LOLR in its international dimension arise when the problems affect an institution that is systemic in the host country, but not in the home country. In these cases, if there are doubts as to the solvency of the institution, and if it does not have top quality collateral, the central bank of the host country might assume a risk which is not necessarily backed by the central bank of the home country. It is crucial in these cases there be adequate communication between the authorities of both countries, central banks and supervisors (which is more complicated the more authorities are involved on each side).

The latest crisis has evidenced the importance of cooperation between central banks in injecting liquidity in currencies other than their own. The swaps lines agreed in December 2007 between the Federal Reserve, on the one hand, and the ECB and central banks of England, Canada and Switzerland, on the other, although perhaps not used as envisaged, transmitted a positive message of coordination at the time and broadened the range of instruments available to central banks at a very necessary time.

Injecting Public Money
Although central banks take every precaution to ensure that there is no underlying insolvency problem and they demand collateral with adequate discounts, any action by the LOLR entails the risk that a liquidity problem may become one of solvency, and that the collateral may suddenly slump in value. In these situations, the central bank must have an exit strategy.

Who decides on the use of public funds? Ultimately, the decision is for the government or Parliament, but if the origin is in a loan from a central bank, there must be adequate coordination between the latter, the financial supervisor (if this is another institution) and the Treasury. The case of Northern Rock has evidenced a number of deficiencies in the decision-making process which led to its eventual nationalisation (see FSA, 2008). There is undoubtedly no clean way out of this kind of situation, but the experience of the last few years may shed some light at least on how to avoid making certain mistakes.

It may be rational to bail out a bank rather than letting it fall, even in a solvency crisis. According to research by Goodhart and Shoenmaker (1995), covering 24 countries, of 104 failed banks, 73 were bailed out and 31 were liquidated. It would seem that rescuing ailing banks is the rule, rather than the exception. But this rationale does not prevent this kind of pattern from generating a moral hazard. In Bagehot’s words, ‘any aid to a present bad Bank is the surest mode of preventing the establishment of a future good Bank’.

The case of Northern Rock was not the only case of public aid to banks exposed directly or indirectly to the subprime crisis. In Germany, IKB (which was majority-owned by a public bank from the state of North-Rhine-Westphalia, KfW Bank) was rescued using public money at an estimated cost of €800 million. The intervention in Fannie Mae and Freddie Mac in the US was slightly different, because they were ‘sponsored’ institutions implicitly guaranteed by the government. The case of the AIG, possibly the largest financial rescue in history, broke all the rules of LOLR as we knew it, since an insurance company was never envisaged as subject of this kind of support. The turmoil created by this case –and the bankruptcy of Lehman Brothers a few days earlier– triggered a series of bail outs in the US and in several European countries, some of them targeted to specific ailing institutions, others to certain markets or types of securities, others to depositors or to mortgage debtors. It is too early to draw lessons from this wave of rescue operations, except that they will probably change dramatically the prevailing consensus on LOLR.

An interesting side to these bail-out operations is antitrust policy, which is particularly significant in Europe, where the single financial services market implies cross-border competition which sometimes might conflict with the national dimension of LOLR operations. Some public aid has triggered European Commission enquiries into possible unfair competition.

Which Institutions are Systemic?
According to the traditional doctrine on bailing-out problem banks, those which are ‘too big to fail’ enjoy a certain implicit government guarantee. It has been argued that this policy runs counter to the principle of a level playing field, since it implies a competitive disadvantage for smaller institutions. Although the too-big-to-fail policy has not, in general, been formulated explicitly by the authorities, it is priced in by financial markets and rating agencies.

However, in the latest crises the casuistry has become much more complicated. Some institutions, while not very large, seem to have become too complex to fall, too intertwined to fall, etc.

Transparency versus Constructive Ambiguity
The question of uncertainty is central in the debate on LOLR. According to the traditional view, the central bank must not commit ex ante to a specific LOLR policy, because this would exacerbate moral hazard. This is the policy of ‘constructive ambiguity’.

All the central banks, in their role as LOLR, need to practise ex ante constructive ambiguity to some extent, which consists in not tying their hands in the event of future crises of which all we know with any certainty is that they will surprise us somehow. However, a more difficult issue is how transparent a central bank should be during a crisis. On the one hand, there are times (and the 2007-08 crisis is a good example) when markets need to know with some precision the conditions of support from the central bank, especially when the turmoil originates from a generalised lack of confidence and the action of the central bank is aimed at correcting this. Furthermore, accurate information on central bank support to certain institutions can be a double-edged sword, since, in certain conditions, it can help restore confidence (as in the case of Bear Stearns), but in others it can undermine confidence and precipitate the bank’s fall (as in the case of Northern Rock). The factors that lead to one or another scenario are complex, but perhaps the most decisive one is the perception of resolute help by the authorities, which leads to a swift and definitive solution to the problems.

LOLR policy must necessarily be transparent ex post, once the crisis is over, because of the need for accountability in the use of public funds, especially in the event of capital injections, but also in the genuine cases of liquidity injections. The extent to which these actions generate a moral hazard is a significant question, but it must not be exaggerated: any insurance generates a moral hazard, but this does not mean that it is not rationale and that its cost is not often much lower than the systemic risk which was averted.

Conclusions: The traditional instruments available to central banks have not worked in tackling the subprime crisis. They have all been obliged to adapt their instruments to a greater or lesser extent, although the broadest and most flexible schemes (in areas such as the maturity of interventions, the type of collateral acceptable or the range of central bank counterparties) have proved to be the most adequate. Consequently, there has been some convergence towards the ECB’s intervention mechanisms, which was the broader among the main central banks.

The classical prescription that lender of last resort operations must be granted at a penalty rate has proved inadequate in a systemic, global and prolonged crisis.

The distinction, within the financial system, between flows channelled directly by markets or those that are channelled via financial intermediaries, and, among these, between banks and other intermediaries, is increasingly unclear. This has far-reaching implications on financial regulation and supervision, and on the design and scope of the financial system’s safety nets (including the lender of last resort). There is now an international debate in this regard, which looks set to continue in the next few years.

Some international harmonisation of some aspects of emergency liquidity injection is necessary, to prevent arbitrage by trans-national institutions and free-rider behaviour.

Although the role of the lender of last resort always involves some element of ‘constructive ambiguity’, systemic crises such as the present one require considerable transparency in central banks’ actions, as well as a commitment to clear rules of conduct, in order to restore market confidence.

Santiago Fernández de Lis
Partner of Analistas Financieros Internacionales (AFI).


Bagehot, W (1873), Lombard Street: A Description of the Money Market, H.S. King, London.

Fernández de Lis, S. (2008), ‘La función de prestamista de última instancia a la luz de la crisis financiera reciente’, Análisis Financiero Internacional, nr 131, first quarter.

International Monetary Fund (2008), ‘The Recent Financial Turmoil. Initial Assessment, Policy Lessons and Implications for Fund Surveillance’, April.

International Monetary Fund (2008), ‘Global Financial Stability Report: Containing Systemic Risks and Restoring Financial Soundness’, April.

FSA (2008), ‘Internal Audit Review of its Supervision of Northern Rock’, March.

Geithner, T.F. (2008), ‘Actions by the Federal Reserve of New York in Response to Liquidity Pressures in Financial Markets’, Statement before the US Senate Committee on Housing, Banking and Urban Affairs, 3 April.

Goodhart, C., & D. Shoenmaker (1995), ‘Should the Functions of Monetary Policy and Bank Supervision be Separated?’, Oxford Economic Papers, vol. 39.

Guttentag, J., & R. Herring (1983), ‘The Lender of Last Resort Function in an International Context’, Princeton Essays in International Finance, nr 151, May.

Taylor, J.B. (2007), ‘Housing and Monetary Policy’, National Bureau of Economic Research, Working Paper nr 13.682.

UK Treasury (2008), ‘Housing Finance Review: Analysis and Proposals’.

US Treasury (2008), ‘Blueprint for a Modernized Financial Regulatory Structure’, March.