Political foundations of the lender of last resort

Charles Calomiris, Marc Flandreau, Luc Laeven

19 September 2016



The recent Global Crisis has prompted the major central banks to conduct lender of last resort (LOLR) operations on an unprecedented scale and stirred up a debate on the constraints and boundaries of LOLR policies (see Bindseil 2014 for an overview of such operations in the recent crisis). For instance, the Federal Reserve, the ECB and the Bank of England have each assisted banks with special loans and asset purchases designed to bolster banks’ positions, expand the supply of liquidity, and reduce the risk of deposit withdrawals.

Yet, the degree to which central banks provided LOLR assistance to banks during the crisis was not uniform, nor was structure of such assistance the same across countries. Interestingly, such differences have been commonplace throughout history (see Bordo 1990 for an early account of the LOLR function throughout history; and Bignon et al. 2012 for a modern appraisal). For one, while central banks empowered to act as LOLR were commonplace in Western Europe by the middle of the 19th century, central banks were not created in the US until 1913, in Canada until 1935, and in Australia until 1959. The techniques of lending of last resort varied across countries and were deeply influenced by local conditions and institutions. Collateral rules for LOLR loans, including haircuts, have varied markedly across institutions and over time. In some cases only banks were eligible for emergency lending, in other cases such as in the influential British experience, lending was extended also to nonbanks. Finally, LOLR assistance sometimes has not been limited to loans, or to central bank operations. Government assistance through credit guarantees, preferred stock, and common stock investments has been a feature of LOLR assistance in response to particularly severe financial system shocks as early as the late 19th century.

History of the lender of last resort

In a recent paper (Calomiris et al. 2016), we examine the history of the development of the LOLR throughout the world, and explore how politics and economics interacted to produce the heterogeneous evolution of LOLR structures and actions around the world. Do such differences merely reflect differences in economic fundamentals that LOLR respond to, or also differences in the operational frameworks of central banks and political support for government assistance?

We define LOLR as central bank or government assistance to financial intermediaries in the form of emergency loans, guarantees, or asset purchases (including preferred or common stock purchases) to provide the needed liquidity or financial strength to end runs on short-term debt claims. Those actions allow financial intermediaries to continue to provide transaction services through the payments system and to provide credit to borrowers without access to capital markets.

Our historical account shows that differences in the structure and function of lenders of last resort reflect major political obstacles to establishing LOLRs and adopting effective rules for LOLR policy, and cannot be explained by economic differences alone. This was the case in early 19th century Britain, where the institutional changes that gave the Bank of England LOLR powers and responsibilities, following a succession of banking crises, were controversial and contested. In the US, the development of a LOLR was delayed as a result of political opposition, and when the Federal Reserve System was created in 1913, its structure and powers were circumscribed by restrictive legislation. The Fed’s powers were narrowly confined to engaging in collateralised rediscounts and advances on certain classes of assets with member banks.  In contrast, the Bank of England was permitted ample room for improvisation.

The experiences of Canada and Australia also illustrate unique central bank chartering outcomes, which reflect their own political histories. Canada’s classically liberal political environment eschewed central banking until 1935. Instead, Canada relied on interbank coordination to avoid banking crises. The establishment of the Bank of Canada in 1935 reflected monetary goals rather than any perceived failings due to the absence of a LOLR. Australia did not create a full-fledged central bank until 1959, which was the culmination of a protracted political struggle over the appropriate allocation of power over money and credit.  More recently, political constraints that reflect the allocation of political powers within the euro area have played an important role in defining and limiting LOLR actions of the ECB to deal with banking crises within the euro area.

The LOLR is a locus of political power, and as such, its creation should be viewed as the outcome of a political bargain (Calomiris and Haber 2014). It is therefore not surprising that countries differed in their propensity to create LOLRs, and in the powers with which they chose to endow them. LOLRs began as collateralised lenders empowered and required to provide credit to banks that were otherwise unable to fund their needs during crises, but LOLRs’ statutory powers changed over time in varied ways.

We trace changes over time in the approaches used by central banks and governments to deal with financial crises from the late 19th century to the late 20th century. We identify a shift in the scope of LOLRs away from a narrow reliance on collateralised lending to an approach including also other forms of support, including credit guarantees, preferred stock assistance, and other mechanisms. We relate this shift in part to the need to expand LOLR activity to a broader set of interventions in the case of systemic banking crises.

Although the mechanisms and reach of LOLRs expanded to include a wide variety of tools, which were employed to deal with varied circumstances, until the 1980s, most countries managed to construct policies that dealt with systemic threats while avoiding blanket protection of all banks’ liabilities. Even though LOLR assistance evolved to include approaches other than collateralised lending, historical assistance was selectively used only to address systemic threats, and when assistance was provided, it adhered to what we refer to as ‘Bagehot’s Principles’, building on Bagehot’s (1873) treatise:  central banks were encouraged to focus on the health of the financial system, rather than on the fate of individual banks. Failure of financial institutions was permitted unless there was a credible systemic risk associated with their failing. During episodes of systemic crises, LOLRs would take on some default risk as a necessary part of their role in assisting the banking system, but only within limits – banks as a whole had to bear most of the risk from such assistance. The participation of banks in risk sharing ensured that assistance would be selective.

Although historical LOLR assistance did not follow an explicit rule, because its structure generally adhered to Bagehot’s principles, assistance minimised adverse fiscal and moral hazard consequences. In many countries, including Britain and France, the structure of LOLR operations was explicitly intended to prevent severe fiscal consequences. While effective interventions necessarily involved risk taking by the LOLR, they usually turned out to be profitable – at least when measured on an ex post, cash-flow basis – because support was provided at a high price and with limited risk, taking advantage of the central bank’s monopoly position in the provision of liquidity.

After WWII, and especially after the 1970s, generous safety net protection became the norm, and in some cases offered unlimited depositor protection (at least ex post). Unlimited protection eliminates the risk of depositor loss and prevents any bank of significant size from failing, regardless of whether the bank poses a true systemic risk. Such protection is generally achieved via a combination of deposit insurance and ad hoc government bailouts of banks through injections of taxpayer funds. Protecting risky banks from the discipline of deposit withdrawals keeps bank credit flowing, which can be particularly beneficial to politicians anticipating an election, but such protection entails social costs in the form of greater risk-taking and large potential fiscal consequences due to long-term financial losses of protected banks, and output losses from the financial crises that protection encourages.

A comparison across countries

We perform a detailed comparison of 40 countries’ statutory provisions for central bank lending in 1960, and follow the changes in LOLR legislation in 12 of those countries from 1960 to 2010. We measure differences in central banks’ LOLR powers across several dimensions and consider possible explanations of those differences. We find that countries differ greatly in the extent of their LOLRs’ statutory powers. Those powers change little over time, except in response to crises. Countries with relatively powerful LOLRs in 1960 – in particular, those whose LOLRs enjoyed the power to issue guarantees – tended to be less generous in their level of deposit insurance coverage as of 1980. These findings suggest there may be some substitutability between LOLR activities and depositor protection.

Our historical analysis shows that, in general, there has been a lack of clear rules established by government that determine what sort of assistance can be supplied by the LOLR, and the process that determines how assistance would be provided. Instead, assistance by central banks and governments usually has been provided through ad hoc responses to events.

Obviously, rules matter because they affect incentives of market participants and thus can limit moral hazard. If banks know that assistance will be limited to certain circumstances and provided according to pre-established rules, that creates an incentive for banks to manage risk and maintain liquidity and capital to protect themselves from risks that are not protected. Furthermore, if market participants are aware of a commitment by the government or the central bank to provide LOLR assistance to address systemic risks, the expectation of assistance can help to stabilise the financial system by acting on market participants’ expectations.

Concluding remarks

We conclude that the LOLR function should strike a balance between the need to respond to severe systemic shocks in a flexible and timely manner and the desire to mitigate moral hazard through pre-established ruled that set limits on assistance. We also recognise, however, that failures to achieve the proper balance reflect the central reality of LOLR design, which is that LOLRs are the outcomes of political bargains.

Authors’ note: The views expressed here are our own and should not be interpreted to reflect the views of the ECB.


Bagehot, W. [1873] (1962), Lombard Street: A Description of the Money Market, Homewood, IL: Richard D. Irwin.

Bindseil, U. (2014), Monetary Policy Operations and the Financial System, Oxford: Oxford University Press.

Bignon, V., M. Flandreau, and S. Ugolini (2012), “Bagehot for Beginners: The Making of Lender-of-Last-Resort Operations in the Mid-Nineteenth Century.” The Economic History Review, 65, pp. 580–608.

Bordo, M. D., (1990), “The Lender of Last Resort: Alternative Views and Historical Experience,” Economic Review, Federal Reserve Bank of Richmond, January/February, pp. 18-29.

Calomiris, C. W., M. Flandreau, and L. Laeven (2016), “Political Foundations of the Lender of Last Resort: A Global Historical Narrative,” CEPR Discussion Paper No 11448. 

Calomiris, C. W., and S. H. Haber (2014), Fragile By Design: The Political Origins of Banking Crises and Scarce Credit, Princeton: Princeton University Press.



Topics:  Global crisis

Tags:  lender of last resort, LOLR, crisis, Federal Reserve, Bank of England, ECB

Henry Kaufman Professor of Financial Institutions, Columbia University Graduate School of Business

Professor, Graduate Institute, Geneva and CEPR Research Fellow

Director-General of the Directorate General Research, European Central Bank and CEPR Research Fellow