We are told that the phrase “lender of last resort” (LOLR) originated as a lawyer’s phrase.1 However, the term certainly does not aspire to legal precision: LOLR is “a term often used and less often defined.”2 This imprecision is understandable. The LOLR function has evolved with time, and today’s central banks face problems very different from those of the nineteenth and early twentieth centuries.

We are told that the phrase “lender of last resort” (LOLR) originated as a lawyer’s phrase.1 However, the term certainly does not aspire to legal precision: LOLR is “a term often used and less often defined.”2 This imprecision is understandable. The LOLR function has evolved with time, and today’s central banks face problems very different from those of the nineteenth and early twentieth centuries.

Although one can only be grateful that the LOLR function is not embalmed in law, the imprecision of the concept is chastening. Everybody agrees that the LOLR function has something to do with liquidity and financial crises. Beyond this point, all is controversy. To avoid unnecessary controversy and premature obsolescence, we will not attempt to define the LOLR function here. Instead, we will take a historical approach, discussing the role of central banks in older and more recent banking crises. This discussion will avoid the so-called international LOLR role appurtenant to sovereign debt crises.

The discussion starts with the classical LOLR role of the gold standard days and proceeds to the early years of the Federal Reserve System. It then shifts to recent history: events since 1970 until the present. It concludes with some possible lessons this history has for problems of the future.

Before the Federal Reserve Act

The lender-of-last-resort function existed well before it was recognized as such, much like Molière’s gentleman who realized that he had been speaking prose all his life. Although in the first half of the nineteenth century several institutions existed that we would now call central banks, they were reluctant to recognize that they had unique roles in times of financial crisis. We will not discuss this pre-history, but instead begin with Walter Bagehot, who first articulated a theory of central bank action. Bagehot’s ideas have heavily influenced subsequent concepts of the LOLR role, and are still worth reading today.

The Bagehot Formula

In his 1873 classic, Lombard Street, Walter Bagehot was the first to link a particular institution—the Bank of England—with a unique set of functions, including the LOLR role. However, the Bank of England had exercised these functions years before Bagehot described them. Bagehot was simply trying to get the privately owned Bank of England to acknowledge that it was a different kind of bank—a central bank—and that it had LOLR responsibilities.

Bagehot’s classical LOLR role was rooted in the gold standard. As today, most bank money was bank credit. However, gold standard banks—including central banks—were obligated to redeem bank credit for gold upon demand. Whatever the monetary policy virtues of the gold standard, this is an unstable way to design a payment system. There was much less gold than money, and the gold stock was finite in the short run. If everybody tried to redeem his bank credit money for gold at the same time, the money supply would disappear. Central banks had plenty of gold for ordinary operations, but there was never enough to meet all possible claims in times of panic.

Bagehot’s LOLR therefore had a difficult choice in a crisis—which at that time was a mass demand for gold. The more liquidity—gold—it injected into the market, the less liquidity the LOLR itself had. The less liquidity it injected into the market, the less confidence in the market, and the greater the demand for gold. An effective LOLR therefore had to hold an enormous quantity of gold in reserve, if it were to keep credibility in times of crisis.3 Hence the name of many central banks, such as the Federal Reserve Bank of New York.

Bagehot’s LOLR implied a certain monetary policy. A prudent LOLR would stockpile gold during economic good times to build its reserves against the panic certain to come. But this accumulation of gold also served as a countercyclical monetary policy. A central bank could only attract gold through high interest rates. This would tend to slow an overheated economy. In a kind of virtuous circle, this would discourage excessive speculation and therefore make a financial panic less likely.

The U.S. Experience

Between the 1836 abolition of the Second Bank of the United States and the Federal Reserve Act of 1913, the United States did not have a central bank. The United States experienced several financial panics during the late nineteenth and early twentieth centuries: in 1873, 1884, 1890, 1893, and 1907.4 Most of the time, somebody—or some group—assumed an ad hoc LOLR role. Usually, the LOLR came from the private sector—generally the clearing houses. Sometimes, the Treasury got into the act. Once—in the Panic of 1907—the LOLR was even a private citizen, J.P. Morgan. (The New York bank is his namesake.)

One does not have to be an advocate of privatization to realize that private LOLRs can work. The LOLR—although acting in the public interest—does not necessarily need a governmental character. The LOLR’s main tools are its liquidity, its prestige, and its experience. None of these characteristics are those one would traditionally consider to be governmental. However, the LOLR role is somewhat incompatible with what many consider the principal objective of a private business enterprise—generating economic returns for its owners.5 A LOLR could not always seek to maximize its profit,6 because it had to subordinate that objective to a more public objective—to provide for financial stability. Accordingly, an LOLR could not be as profitable as ordinary commercial banks. The LOLR had to stockpile a large quantity of gold for emergency liquidity. Gold, however, is a poor bank asset, because it bears no interest.

The ad hoc American LOLRs did not stockpile a large reserve of gold, but supplemented their reserves by other expedients, such as restricting the convertibility of bank money to gold. This practice was quite common in eighteenth- and early nineteenth-century Europe,7 as well as nineteenth-century America. However, restricted convertibility was inconsistent with Bagehot’s LOLR role, based on an adequate reserve gold supply, which maintained convertibility at all times. Nevertheless, the ad hoc U.S. system seemed to work, at least in the sense of preserving a working banking system at most times.

The Federal Reserve Act of 1913

However, the ad hoc United States system did not work well enough. Recurrent financial panics—and the concomitant absence of a formal LOLR—were probably the major driving force behind the foundation of the Federal Reserve System in 1913.

[T]he national banking system, among other defects, fails to afford any safeguard against panics and commercial stringencies or any means of alleviating them. This fact has received more attention than has thus far been given to any other in the whole range of the banking and currency discussion, and there has been more effort to apply some legislative remedy to this than to any other condition.8

The remedy was a real central bank, which could serve as an emergency reserve of gold, much like the European central banks. Like the European banks, lending was discretionary to the Federal Reserve Banks,9 although the Banks shared their autonomy to set rates with the Board. However, Congress charged the nascent Federal Reserve Banks with other responsibilities. Some of these additional responsibilities were complementary to the LOLR role, such as responsibility for supervising member banks. Congress understood that bank examination was a key component of the clearing house (i.e., LOLR) role and for that reason granted specific examination authority to both the Board and Federal Reserve Banks.10 However, some of the Federal Reserve’s new responsibilities conflicted with the LOLR role.

Although Bagehot’s classical LOLR was complementary to the monetary policy of the era, the Federal Reserve System’s original monetary policy conflicted with its LOLR role. The Federal Reserve Act of 1913 was built on the gold standard. However, it had a more complex monetary policy, which was a hybrid of the gold standard and the “real bills doctrine.” The real bills doctrine required that a central bank implement monetary policy by lending only against short-term self-liquidating commercial paper—so-called “real bills.” The original Federal Reserve Act implemented the real bills doctrine through statutory rules that restricted eligible collateral. However, the original Federal Reserve Act did not distinguish LOLR lending from monetary policy lending. As Bagehot noted long ago, the LOLR function calls for flexibility in collateral, not narrow “eligibility” rules:

[We lent money] by every possible means and in modes we had never adopted before; we took in stock on security, we purchased Exchequer bills, we made advances on Exchequer bills, we not only discounted outright, but we made advances on deposits of bills of exchange to an immense amount, in short, by every possible means consistent with the safety of the Bank, and we were not on some occasions over-nice.11

By the mid-1920s, the Federal Reserve System intellectually abandoned the real bills doctrine,12 shifting to open-market operations as its main monetary policy tool. However, the statutory constraints remained until the Great Depression of the 1930s.

The Great Depression and Postwar Years

The Great Depression caused Congress to revisit the Federal Reserve System, in a series of statutes, most notably the Banking Act of 1935. These statutes collectively transformed the Federal Reserve’s LOLR role.

First, Congress effectively abolished the gold standard, although Federal Reserve Banks were obligated to hold gold for years afterward. The abolition of the gold standard transformed the LOLR function. First, it took much of the risk out of LOLR lending. Without the gold standard, an LOLR could produce unlimited liquidity—at least in the short-term characteristic of financial panics. The LOLR’s only risk was of counterparty insolvency: a risk avoidable by good collateral. Second, market participants understood this. If the LOLR could not go broke and was willing to lend, the system need never run out of liquidity. To the extent that the LOLR was willing to lend and enough collateral was available, old-fashioned systemic panics seemed inconceivable.

Second, Congress eliminated the old statutory restrictions on collateral, effectively recognizing the intellectual demise of the real bills doctrine. The new Section 10B of the Federal Reserve Act allowed Federal Reserve Banks to accept any collateral that they deemed acceptable. This was originally an emergency measure. However, as the real bills doctrine faded away and the Great Depression continued, this emergency power became permanent.

Finally, Congress granted Federal Reserve Banks emergency powers to lend to nonbanks, although it restricted the collateral that Reserve Banks could accept. In the 1930s, Federal Reserve Banks were even empowered to lend to small business, although the statutory authorization for this expired in the late 1950s. This particular extension of the LOLR idea died with the Great Depression, but Reserve Banks retained their emergency powers to lend to any “individual, partner, or corporation,” under “unusual or exigent circumstances.” We will come back to this Doomsday authority later. Suffice it now to say that the Federal Reserve has been altogether parsimonious in its use of this power to make loans to nonbanks.

Modern History

For about 30 years after the Great Depression, the LOLR role seemed to have disappeared. Financial panics simply did not happen, at least in industrial countries. The old-style financial panic—a run for gold—could not happen, because the gold standard no longer existed. Major banks did not become insolvent, and major economies avoided severe depressions.

And then things changed. We will mark the turning point in 1974, with the Herstatt and Franklin insolvencies. The selection of this date and these events is somewhat arbitrary, because several things happened around the same time. The world abandoned fixed exchange rates in the early 1970s, after a decade of pressure. Some of the older industries—such as railroads or steel—came on financial hard times. The notion of a sovereign defaults first became conceivable, and then the conceivable became the anticipated. Bank balance sheets became both less liquid and more complex. (They also became less relevant with the growth of off-balance sheet items.) Banks became less regulated and more competitive. Bank capital reached an all-time low, the end-product of a secular decline since at least World War I. Domestic interest rates became very volatile.

Finally, although bank capital and liquidity had decreased, the dollar value of bank payments—both domestic and cross-border—kept increasing. This stressed liquidity in a way not seen since the gold standard. As the volume and velocity of payments increased, for every major bank the day’s incoming payments became the only possible funding source for the day’s outgoing payments. If an appreciable amount of the incoming payments was disrupted for any reason, the outgoing payments could not be made.

This is enough generalizing. A specific list of events is far more useful. This list mainly draws from recent U.S. experience.

1971—Penn Central

When this old railroad defaulted on its commercial paper, the commercial paper markets panicked. Other issuers of commercial paper had a difficult time placing or rolling over their paper, in circumstances much like an old-fashioned run on a bank. The situation loosely fit Bagehot’s classic criterion—a general liquidity crisis—but this was a liquidity crisis among nonbanks.

The Federal Reserve System used classic crisis-management tools to resolve this crisis: moral suasion and systemic lending. It notified banks that it would provide discount window credit to the banks that accommodated former commercial paper issuers. The issuers shifted to bank loans and crisis was averted. Some say that the Penn Central default was a first sign that disintermediation had been blurring the boundaries between banks and nonfinancial firms. Others say that banks performed their classic role as a transmission belt for liquidity: moving it from the LOLR to where it was needed in the nonbank sector.

The Penn Central insolvency raised the question of the scope of the LOLR function. Did the LOLR only apply to banks? If it applied to non-banks, was it limited to threats to the financial system posed by nonbank failures? Or was the LOLR function a general public-policy tool, used as an adjunct to corporate workouts and other situations? During the decade of the 1970s, the third possibility was explored and occasionally large insolvencies were averted by use of governmental loan guaranties, as with Lockheed and Chrysler. However, the Federal Reserve declined to use its lending power to stave off any industrial insolvencies during this period. By the 1980s, many large U.S. industrial firms entered bankruptcy proceedings, with little thought of guarantees or the like, and little or no adverse effect on financial markets.

1974—Herstatt and Franklin

This was an inauspicious year, boasting two major liquidity crises, both triggered by foreign exchange losses at banks. These crises illustrated the need for a LOLR role.

The German supervisors closed Bankhaus I.D. Herstatt on June 26, 1974 at the end of the German business day, after Herstatt had settled the deutsche mark side of its foreign exchange transactions, but before the dollar side was settled. Overnight, the banking system rediscovered settlement risk and liquidity crisis. The liquidity crisis was not perhaps as severe as the crises of the gold days, but was bad enough. Foreign exchange trading slowed to a trickle and took several years to recover its pre-1974 levels. All but the top-tier institutions were shut out of the foreign exchange market. New York Clearing House banks would not make payments without intraday cover.

The Franklin National Bank, an aggressively (and poorly) run domestic institution, was about to collapse in May. The collapse was triggered by an announcement of losses from unauthorized foreign currency trading. (Franklin had about ten times the foreign exchange exposure of Herstatt.) The Federal Reserve Bank of New York kept Franklin on discount-window life support for five months, averting amplification of the Herstatt crisis. In October, Franklin was purchased in an auction, with the insurer paying off the liabilities to the Federal Reserve Bank.

1980—Monetary Control Act

Congress had not appreciably changed the LOLR function since the Great Depression, although the Federal Reserve Banks lost their small business lending authority in the 1950s. But in 1980, Congress considerably expanded the potential scope of LOLR lending by permitting Federal Reserve Banks to lend to all depository institutions, rather than just member banks.

1984—Continental Illinois, Bank of New York

The year 1984 was another bad year for banks. Although the Continental Illinois crisis resembled the Franklin crisis, the Bank of New York (“BONY”) crisis ushered in a new chapter of LOLR lending.

The Continental Illinois crisis was a reprise of Franklin. As with Franklin, the Federal Deposit Insurance Corporation (FDIC) recapitalized a large insolvent bank with a bridge loan from the Federal Reserve Bank of Chicago. However, the FDIC did not quickly sell the institution, but rather held its stake for a number of years and made a tidy profit. Continental Illinois is significant in two respects. First, it illustrates the distinction between balance-sheet insolvency and “goodwill insolvency.” Because the FDIC made money on an insolvent institution, an accountant would say that it had enough goodwill on its balance sheet to fill the asset gap.13 Second, Continental Illinois was a harbinger of the Savings and Loan (S&L) crisis of the late 1980s and early 1990s. During the S&L crisis, other institutions, such as the Bank of New England, received similar liquidity support.

The events of BONY are easy to describe; the implications are profound. One day, it suffered a temporary computer failure, which enabled it to make payments, but not receive them. This could have been a systemic disaster, because BONY was one of the “clearing banks” for government securities. A $22.6 billion overnight loan from the Federal Reserve Bank of New York permitted BONY to fulfill its payment obligations for the day, while it fixed its computer. BONY represented at least two new wrinkles in the LOLR function.

  • First, it marked a return to the past: severe illiquidity coupled with unquestionable solvency. This combination was common in the gold era, when all banks were always at risk of having insufficient gold. However, after the gold era, especially with modern capital markets, illiquidity became increasingly related to dubious solvency. BONY showed that this new rule has exceptions, and operational problems could create illiquidity just as effectively as a catastrophic loss to the balance sheet.

  • The second new wrinkle was in BONY’s clearing bank function. Because BONY was a clearinghouse, the entire system relied on BONY. The international banking system has become increasingly dependent on private clearinghouses, such as BONY: Euroclear, Cedel Bank, CHIPS, Multinet, the new CLS Bank, etc. Most of these institutions were created since 1970.

1987—The Stock Market Break

The October 1987 stock market break strained the clearing and settlement systems. This threatened a systemic liquidity crisis, similar to the BONY case, but was not institution specific. Systemic problems were averted by a large and temporary injection of liquidity through open-market operations. Here again, we can witness the transmission-belt phenomenon. The Federal Reserve provided liquidity to the banking community, which transmitted it onward to other institutions in need of credit. The market break experience was very similar to Bagehot’s classic formula in which the LOLR would lend indiscriminately to any creditworthy counterparty with good collateral. The response in the 1987 stock market break shows that the expression “lender of last resort” is a bit of a misnomer. A purchase of securities by a central bank provides liquidity as easily as a loan. The more meaningful distinction is between institution-specific liquidity and general market liquidity.


The Drexel insolvency illustrates a significant point: today’s LOLR function need not involve any lending, or even any credit. The Drexel collapse was quite distinctive, because Drexel was still balance-sheet solvent when the market lost confidence in it. Nevertheless, liquidity to Drexel threatened to dry up, and counterparties had a difficult time settling their positions with Drexel. One specific problem involved liquidation of various physical securities. The Federal Reserve Bank of New York established an emergency facility for physical delivery-versus-payment transfer, known in the community as “Stone-Age clearing.” Parties without confidence in Drexel could nevertheless transact through this clearing facility. This facility was ultimately unnecessary, but shows that a market disturbance can sometimes be smoothed with calming words and conduct from the LOLR.


In 1991, Congress looked back at the savings and loan crisis of the 1980s, and enacted the Federal Deposit Insurance Corporation Improvement Act (FDICIA). Most of its reforms were supervisory, most notably the “prompt corrective action” requirements. However, FDICIA modified discount window practice in two ways. First, it made Federal Reserve Banks responsible for Continental Illinois or Franklin-style lending. FDICIA applies when the Federal Reserve extends credit to an undercapitalized institution pursuant to Section 10B of the Federal Reserve Act. If it does so, it must reimburse the FDIC for any capital erosion while the institution is on life support. Second, Congress dropped the archaic collateral limitations on lending to nonbanks. However, loans to individuals, partnerships, and corporations still require specific approval by the Board of Governors of the Federal Reserve System, based on a finding of “unusual and exigent circumstances.”


The 1995 Barings insolvency is a nice counterpoint to the Herstatt insolvency of 1974 and illustrates the increasing significance of clearing houses. Both the Barings and Herstatt insolvencies were unexpected and affected midsized banks. But Barings’s counterparties had learned from the Herstatt experience and were comfortable with their exposures to Barings. The only uncertainty in this insolvency was whether the Singapore clearing house could accommodate itself to the insolvency. Fortunately, the market had sufficient confidence in the Singapore clearing house to proceed, and a major crisis was averted without explicit use of LOLR facilities.

It is probably too early to discuss the recapitalization of Long Term Capital Markets (LTCM) in any detail. However, the LTCM recapitalization is worth mentioning, at least to show that the trading-related liabilities of many financial institutions have a very short tenor, default clauses often contain hair triggers, and liquidity is always at risk of drying up.

Challenges of the Future

If the future resembles the recent past, the LOLR function will retain its significance. Three themes will likely dominate the next few decades of this function. First, market liquidity will continue to remain a major and evolving problem. Second, we will continue to grapple with the problems of institution-specific lending. Finally, international financial crises are unlikely to disappear. Because this third theme is outside the scope of discussion, this section will concentrate on the challenges of market liquidity and institution-specific lending.

Market Liquidity and Payment Systems

Before Herstatt, one could think that systemic liquidity crises disappeared with the gold standard. But after Herstatt—and BONY, the 1987 crash, and Barings—we know that systemic liquidity crises are inherent in modern banking.

In the old days, the problem was obvious: there was only so much gold to go around. The problem is by no means as simple today. There is plenty of credit in the system, and central banks can add a potentially unlimited amount, at least for short periods. However, if a major insolvency or operational failure occurs, the credit might not circulate, and end-of-day payment obligations may not be met. In response to this, other parties may protect themselves by not making their own payments, leading to a payment gridlock resembling the old days of insufficient gold. The old solution to payment gridlock—the LOLR role—remains as relevant today as it was in the nineteenth century.

During the past 25 or 30 years, payment systems have evolved tremendously. The weaknesses of the 1970s payment system have largely (but not completely) been fixed. Unfortunately, this does not mean that the payment system is now perfect. It merely means that payment systems continue to evolve dynamically. As part of this evolution, bankers can understand and fix the current problems with the system. However, the dynamism of this evolution creates new problems, which we must continue to understand and fix. In other words, work as hard as we might, the banks are always a little bit behind the curve. At least based on our post-1970 experience, some payment system risks will always be new and will always cause surprises, possibly leading to liquidity crises. As long as payments continue to evolve, the LOLR function will remain necessary.

The BONY computer problem and the Barings insolvency are good cases in point. If the solvency (or liquidity) of a clearing house is in doubt, financial panic is possible, no matter how well individual firms are protected against the insolvency of other individual firms. Although international standards ensure that clearing houses can withstand considerable stress, we have seen that markets may become skeptical, especially in new cases. Such skepticism may create a liquidity crisis. Such doubts may require LOLR intervention.

Institution-Specific Lending

Institution-specific lending has always been more controversial than market liquidity lending. Institution-specific lending obviously has a much greater potential to create moral hazard. Institution-specific lending is also much more likely to raise suspicions of favoritism and the like, especially because the LOLR function is discretionary.

It is often difficult to distinguish between institution-specific lending and provision of market liquidity. Sometimes, the best way to provide liquidity to the market is through a central institution. The overnight loan to BONY is a recent example, although a history buff might also think of the 1882 loans from the Banque de France to the French bourses.14 At other times, an institution is saved to avoid market panic, such as perhaps was the case with the 1974 Franklin National Bank. It is hard, in such cases, to distinguish between saving the system and saving the institution. Institution-specific lending is risky to an LOLR. The risk is not so much to the LOLR’s balance sheet as it is to the LOLR’s legitimacy. Institution-specific lending may create a public perception of unfairness, where friends are saved and others lost.15 If such a perception becomes sufficiently strong, the LOLR may lose its special legal status, at least in a democratic society where a legislative response to the will of the people is the tradition.

However, thanks to the lessons of the past two decades, banks are more capable of monitoring and managing the credit risk of their counterparties. We have all become accustomed to the insolvency of industrial firms. Even bank insolvencies have lost much of their power to shock. Not only are counterparties more sophisticated, but the worldwide movement toward enforceable payment-system netting has decreased the impact of insolvency on counterparties’ positions. Consequently, market panics are less likely to occur, or at least less likely to be triggered by an insolvency. The last two insolvencies of major international banks—Barings and BCCI—did not require the LOLR function.

We do not know if this means that institution-specific LOLR will disappear in the future. Certainly, it has become harder to justify emergency lending treatment to individual banks, especially in countries with well-developed capital markets and insolvency law. However, at least one caveat is in order. As discussed above, the LOLR function appeared extinct in the 1950s. But times changed. Today, we know that LOLR lending—or at least the existence of the LOLR alternative—plays a central role in our dynamically evolving payment systems. As the role of central banks continues to change, the scope of the LOLR function will change with it.

Appendix: Legal Authority for Federal Reserve Liquidity Activities

The three key sources of LOLR authority are found in Sections 10B, 13(3), and 14 of the Federal Reserve Act. Section 10B authority is also used for routine lending activities; Section 14 authority is also used for monetary policy implementation.

Other discounting and lending powers may be found in the following sections of the Federal Reserve Act: Section 10A (emergency advances to member bank groups); Section 11(b) (discounts of other Reserve Banks’ paper); Section 13(2) (original statutory authority to discount against eligible real bills); Section 13(4) (discount authority for sight drafts); Section 13(6)–(7) (discount authority for eligible bankers’ acceptances); Section 13(8) (advances secured by certain collateral); Section 13(12) (discount of certain bankers’ acceptances); Section 13(13) (advances on government securities); and Section 13A (discount of agricultural paper). In very special cases, one of these powers might become relevant to LOLR lending.

Appendix Table

article image



The themes discussed this morning were very interesting and underscore the importance of bank supervision. We looked at fundamental supervision, the bank restructuring process, and then the lender-of-last-resort function of central banks—the trinity of issues relating to orderly financial markets.

Mr. Feldberg’s paper dealt with the fundamentals of supervision and how we execute our basic mission. Of course, bank supervision is necessarily a judgment business. You get no awards for being right. There is a perception that either regulators are too loose and allow situations to develop such as that which resulted in the closure of many Saving and Loans and Commercial banks in the late 1980s, or supervisors too harshly adhere to regulations that inhibit economic growth. Bank supervisors will always be viewed as acting too early or too late and ultimately costing somebody some money somewhere—either the taxpayer or the bank or the shareholders.

There has been a lot of pressure for bank supervisors to upgrade their principles and concepts. Part of it began with modification of U.S. laws as a reaction to the BCCI failure. U.S. lawmakers believed that we should ensure that comprehensive consolidated supervision was practiced by institutions wanting to come into the United States and enjoy the fruits of the U.S. market. Our experience showed that this requirement was a little bit too tight. It was very difficult, as a supervisor, to make determinations on what consolidated comprehensive supervision is in a format that could be judged cross-border. It is good public policy to make sure that affected parties are clear about the standards under which they are to be judged. Out of this initial effort grew the introduction by the Basle Committee on Banking Supervision of 25 core principles. There are seven areas in which these basic principles are directed. First and foremost, however, as a precondition for effective banking supervision, there has to be a reasonable environment within which to work—good legal framework, efficient bankruptcy laws, and clear regulation.

The set of core principles relates to licensing, prudential supervision, formal powers of bank supervisors, and cross-border banking. The principles relate primarily to the technical side of the business—how you go about the business of supervision. What expectations should you have as a bank supervisor? What should you expect the banks to do in following safe practices? Other areas of the core principles look at information requirements. It’s good to have strong information flows as a bank supervisor and you can never get too much information.

Formal powers of supervisors are an important element. You can’t deal with inadequate powers of bank supervision when the financial system is in a crisis mode. It just doesn’t work. And those who have already thought out the responsibilities they wish to give to the supervisor are going to be ahead in resolving financial crises because they do not have to debate these issues when quick action is required.

Another area referred to by Mr. Feldberg in his talk was cross-border banking. Three of the principles are devoted to this area and to assimilating the different practices around the world into some kind of orderly context.

The development of these core principles was a major contribution not only in the Group of Ten countries but also in other countries. Russia and China, for example, both spoke strongly for a need for some kind of universal standards that are not just best practices but more rule-oriented so as to convince their policymakers of the need to strengthen the supervisory framework.

The next phase in implementing the core principles, as Mr. Feldberg pointed out, is the self-evaluation all supervisors are asked to perform. I am always a bit skeptical when people are asked to evaluate themselves. We have asked the regional chairman of the supervisory groups to play an important role in overseeing the responses by their countries so that the responses are neither overly critical nor overly optimistic.

Some issues are clearly still on the table. One is whether legal compliance with all these core principles is required or whether longstanding practice is sufficient. Is it permissible, for example, to claim this has been my practice as a bank supervisor, but I don’t have a legal context? I don’t know the answer. We developed these principles in bold print and following the bold print is a description of what elements should be present in achieving compliance. Discussions will no doubt center on what compliance means. Do you need only to follow the principle itself or do you need to incorporate all the issues raised with respect to the lighter type in achieving full compliance? Mr. Feldberg introduced the issue of how to use these principles in a practical way. The IMF and World Bank have said they will use the core principles in their work in evaluating the strength of bank supervision in member countries. The Basle Committee has so far been reluctant to evaluate individual countries. The Canadians have made some suggestions along the lines of a peer group review. It is not clear, however, how the evaluation process will proceed, but some form of evaluation will no doubt be necessary. What time period for implementation is reasonable? No deadline has been set. The principles have yet to be endorsed fully by all the world but we expect that to happen in Sydney, Australia, in 1998. At the time of this Biannual Conference of Bank Supervisors, discussion will no doubt focus on timeframes for implementation. I foresee a three- to five-year time horizon. The question of whether sanctions ought to be imposed for noncompliance will also be on the agenda. That said, it is less than clear that if you follow all of these principles, whether the market will reward banks by charging lower rates when they borrow in the international market. There may be sanctions against those banks from countries that don’t adhere to the principles. As a companion to the core principles, the Basle Committee issued a compendium to help supervisors know best practices. The compendium will be updated periodically.

While the core principles are key initiatives in bringing order to bank supervision globally, a lot of issues remain in the broader context. For example, parallel banks are more popular than ever, especially in Latin America. Parallel banks are commercial banks owned by the same set of shareholders but are not connected institutionally other than through the common shareholders. Most also have a presence in offshore centers. Offshore centers generally don’t provide as healthy a degree of bank supervision as one would like. Such centers traditionally have high barriers against adequate flows of information to bank supervisors. Historically, the United States has had poor experience with parallel banks with an offshore presence. They are largely unsupervisable and may be one of the weakest links in the chain of supervision.

At the end of Mr. Feldberg’s talk, he raised the principle of independency of supervision authority. You could, as some do, put the authority in a central bank. I happen to like it there because I live in a central bank. The primary objective, however, is to ensure that bank supervision resides in a reasonably independent environment. The movement has reluctantly been toward assigning supervision to an independent service agency. England, Japan, Korea, and Australia are but a few examples.

Looking at the second theme—bank insolvency and liquidation—I think we had an excellent presentation. The most effective agencies in supervision are those that have well-defined rules on insolvency and a historic framework of effective liquidation procedures. The economy is most resilient when banks are allowed to function in their capacity as risk allocators. That said, it is important to have in place a strong culture of compliance with safe and sound prudential standards and the legal framework to ensure compliance.

Supervisors, as part of their duties, are responsible for making judgments on the health of individual institutions and the banking system as a whole. These judgments are sometimes difficult but are necessary to ensure that sick and weak institutions take the proper remedial action and, if they don’t, are forced into prompt liquidation. If not resolved promptly, problem banks become much more expensive to liquidate.

There was a very good discussion on insolvency. Of course, there is a lot of technical literature on the subject of determining bank insolvency. But the short answer is that a bank is insolvent when the supervisor says it’s insolvent. This has to do, of course, with evaluation of the asset quality, which is so critical to the job of bank supervision.

Prompt corrective action is a topic debated in many countries. Again, a good discussion this morning. At first, U.S. supervisors didn’t like prompt corrective action when we got it. We thought it might inhibit our flexibility. We always like to think we’re very pragmatic, very smart people and, therefore, we make the best judgments. But I happen to like the framework of prompt corrective action. It allows you the legal framework to ensure strong adherence to capital adequacy rules and a graduated response to different levels of impairment.

The problem with prompt corrective action is that it has not been tested under battle conditions. The law was put in place while the financial system in the United States was very healthy. We have had to use this on only a few occasions. And we don’t know what it is going to be like if we go to a period when we have 300 or more bank failures a year, as happened not too long ago.

One other issue with prompt corrective action is confiscation of private capital. If capital has not been fully exhausted, but the institution is closed, is not that taking away residual shareholder’s value? Experience has shown, however, that losses in liquidating a problem bank’s assets usually run around 10–12 percent. Thus, residual capital in a problem bank can erode in a very short time.

The public benefit of prompt corrective action is curtailing certain activity as the bank is weakened. As the bank penetrates each of the capital thresholds, the corrective action bites more and more and the law requires the bank to withdraw from certain markets, especially the national deposit market for large brokered funds. Banks prefer, of course, to grow out of problems by making more questionable loans or expanding into new business lines or territories. This just doesn’t work at the end of the day.

Another aspect we talked about was insolvency and restructuring, as it relates to changes in bank management. That’s the hardest thing to do. What do you do about making changes in management? It is often hard to find a suitable candidate to take over management. There are just not enough good bankers in the world. The Board of Directors itself is usually also reluctant to replace management. Management is the last to acknowledge that there is a life-threatening problem in the institution. There is usually a five-year life cycle of a problem bank, from beginning to end.

The lender-of-last-resort discussion was an excellent primer by Mr. Baxter—strong reasons were given for having bank supervision in the central bank and this, in my mind, is tied with the lender of last resort.


According to Kindleberger, the “expression comes from the French dernier ressort, the legal jurisdiction beyond which it is impossible to take an appeal.” Charles P. Kindleberger, Manias, Panics, and Crashes 161 (Basic Books, 1978).


Joan Edelman Spero, Council on Foreign Relations, The Failure of the Franklin National Bank 122 (1980).


Two chapters of Bagehot’s classic book were devoted to this subject. See Chapter VII: “A More Exact Account of the Mode in Which the Bank of England Has Discharged Its Duty of Retaining a Good Bank Reserve, and of Administering It Effectually,” and Chapter XII: “The Principles Which Should Regulate the Amount of the Banking Reserve to Be Kept by the Bank of England,” in Walter Bagehot, Lombard Street: A Description of the Money Market (1873).


Carter Glass & H. Parker Willis, House Committee Report, Changes in the Banking & Currency System of the United States, H.R. Rep. 69, 63d Cong., 1st Sess., at 4 (1913).


The Business Roundtable, Statement on Corporate Governance 1 (Sept. 1997).


For a more sophisticated treatment, see Charles Goodhart, The Evolution of Central Banks (MIT Press, 1988).


Charles P. Kindleberger, A Financial History of Western Europe 278–80 (Allen & Unwin, 1984).


Bagehot, supra note 3, at 25 (citing Mr, Harman of the Bank of England on its response to the 1825 panic).


Robert Craig West, Banking Reform and the Federal Reserve: 1863–1923 at 195–204 (Cornell University Press, 1977); Lester V. Chandler, Benjamin Strong: Central Banker at 195–98 (Brookings Institution, 1958).


During the Great Depression, the Reconstruction Finance Corporation recapitalized many insolvent banks. Helen A. Garten, A Political Analysis of Bank Failure Resolution, 74 B.U.L. Rev. 429 (1994). This profitable program used private co-investors.


Goodhart, supra note 6, at 121.


Kindleberger, supra note 1, at 174–75.