In June 1999 the Basel Committee, which sets prudential standards for international banks, put forth a proposal for reforming bank capital standards. This monograph is the Shadow Financial Regulatory Committee's critical and constructive resopnse to that proposal. The topic is an important one, as illustrated by the recent waves of banking crises in both developed and developing economies.

The Shadow Financial Regulatory Committee is a group of independent experts on the financial services industry and its regulatory structure.

The purposes of the committee are: first, to identify and analyze developing trends and continuing events that promise to affect the efficiency and safe operation of sectors of the financial services industry; second, to explore the spectrum of short- and long-term implications of emerging problems and policy changes; third, to help develop private, regulatory, and legislative responses to such problems that promote efficiency and safety and further the public interest; and, finally, to assess and respond to proposed and actual public policy initiatives with respect to the impact on the public interest.

The results of the committee's deliberations are intended to increase the awareness and sensitivity of members of the financial services industry, public policymakers, the communications media, and the general public to the importance and implications of current problems, events, and policy initiatives affecting the efficiency and safety of the industry and the public interest.

Members of the Shadow Financial Regulatory Committee are drawn from academic institutions and private organizations and reflect a wide range of views. The committee is independent of any of the members' affiliated institutions or of sponsoring organizations. The recommendations of the committee are its own. The only common denominators of the members are their public recognition as experts on the industry and their preferences for market solutions to problems and the minimum degree of government regulation consistent with efficiency and safety.

The American Enterprise Institute for Public Policy Research is proud to support the activities of the Shadow Financial Regulatory Committee.

Christopher DeMuth
President
American Enterprise Institute

Introduction and Executive Summary

From its inception in 1986, the Shadow Financial Regulatory Committee has consistently urged financial policymakers in the United States to promote competition in the financial marketplace while ensuring the safety and soundness of depository institutions in particular. Since its founding, the SFRC has issued more than 160 statements that address those objectives. One of the themes that runs consistently through many of the statements is that sound policy requires the right blend of regulation, supervision, and market discipline to provide the proper incentives for commercial banks and thrift institutions to avoid excessive risks and to protect taxpayers, who ultimately stand behind the government funds that insure the deposits of those institutions.

In this monograph, the SFRC brings that perspective to the process of refining and extending international bank capital standards that have been in place since 1989 and that the Basel Committee on Banking Supervision--the international body of bank supervisors from the G-10 countries plus Luxembourg and Switzerland that sets those standards--proposed in June 1999 to modify.

We concentrate our analysis and recommendations on large banks--and not their holding companies--because they are the institutions that are the focus of the Basel standards and because some believe that the failure of those institutions poses the greatest risks to the financial system. Furthermore, while we address our analysis and recommendations to the Basel Committee, we believe that policymakers in the United States should adopt our proposals, whatever further actions the committee itself may take.

The standards that are the subject of this monograph--and the Basel Committee's recommendations--came into being in the late 1980s, primarily in response to concerns about the fragility of large international banks and the potential consequences of that fragility for the global economy. Those difficulties arose initially in the United States because of the less-developed-country debt crisis and later because of excessive lending for commercial real estate development. At the time, banks in the United States and some other industrialized countries already were subject to national standards governing the minimum amounts of "capital" they were required to maintain to absorb losses and, thus, to protect deposits or any entity that insured depositors from loss. ⁽¹⁾ But because of large lending losses, it was widely perceived that many of the large international banks were too thinly capitalized.

Policymakers in the countries belonging to the Basel Committee responded in the late 1980s by setting minimum capital rules for international banks, for two reasons. First, because large international banks were active in a number of countries and were linked through payment systems and interbank deposits, regulators feared that the failure of one or more of those institutions in one country would adversely affect the financial welfare of other institutions in other countries. Second, governments of each country were reluctant to strengthen capital standards that applied only to home-country institutions because they feared that doing so would disadvantage their domestic banks when competing with banks from other countries.

The most significant feature of the Basel standards from their inception is that they have required banks to maintain more capital to support those assets or activities perceived by the committee to carry greater risks. As a result, the standards assign different assets or contingent liabilities in different risk classes or buckets, assign risk weights to those buckets, and then require banks to maintain capital equivalent to fixed percentages of their total risk-weighted assets and off-balance-sheet commitments. The general standards have continued to evolve since they were adopted in 1988, as have the capital regulations of individual countries. ⁽²⁾

The Basel Committee's latest proposals for change, issued in June 1999, are the most sweeping alterations of all. In brief, the committee proposes a more refined system of risk weights and the use of ratings by private credit-rating agencies to assign risk weights to classes of assets or activities. In addition, the committee has considered permitting the banks' own internal risk-rating systems to play a greater role in determining capital requirements and has encouraged national regulators eventually to allow banks to use internal models to set their own capital standards. One of the major objectives of those proposals is to apply more market-based assessments of risk in the setting of bank capital standards.

While the Basel Committee should be applauded for seeking that objective, the SFRC believes that the June 1999 proposal is deficient in several respects. In particular, the standards erroneously continue to rely on arbitrary risk weights for computing required bank capital. In addition, the standards ignore the fact that bank risk is more properly measured by an institution's overall portfolio than by the sum of its individual assets and other off-balance-sheet commitments. The standards also distinguish improperly among different types of capital by creating two different "tiers" of required capital. Finally, the committee's latest proposals to assign assets to different risk classes on the basis of private credit-rating agencies and to rely increasingly on banks' internal models of risk are flawed in various ways that we outline in detail below.

Accordingly, the SFRC urges the Basel Committee--as well as U.S. bank regulators--to take a more direct approach to injecting greater market discipline into the setting of capital standards for large banks. Specifically, we urge the adoption of a series of independent, but mutually reinforcing, recommendations.

The current risk-based capital requirement should be replaced by a simple, but higher, minimum leverage requirement.

The current distinction between "tier 1" and "tier 2" capital should be eliminated. Instead, banks should be allowed to meet that leverage requirement with an unlimited proportion of explicitly uninsured and suitably structured ("qualifying") subordinated debentures, with the mix of debt and equity being governed by the market as is true for corporations generally.

Capital should be measured by the difference between the market values of bank assets and liabilities, not their historical values.

Large banks not only should be allowed to meet a specified proportion of their capital requirements with new issues of subordinated debentures, but should be required to do so--subject to certain additional restrictions that we outline further in this monograph.

The current system of early intervention and resolution, incorporated in U.S. law by the Federal Deposit Insurance Corporation Improvement Act of 1991 (FDICIA), should be strengthened--in particular, by tying the required specific interventions to the market signals provided by the prices and yields on bank subordinated debt.

Those recommendations would apply greater market discipline to banks' risk taking and reduce burdens on bank supervisors, although supervisors still would need to determine whether banks are complying with capital requirements and take prompt action against banks that fail to do so. A subordinated-debt requirement would achieve that objective by providing supervisors with valuable information concerning bank risk and would thereby make it more difficult for them to forbear when intervention is necessary. We emphasize that the role of supervisors continues to be important; a subordinated-debt requirement is designed to make their jobs easier and to strengthen the discipline they already provide. Moreover, a subordinated-debt requirement would provide stronger incentives for banks to disclose more information than they currently do about their portfolio risks in a timely manner to the public--and in particular to the holders of subordinated debt.

We proceed as follows. Section 1 reviews the role of capital, in theory and in practice, before the establishment of deposit insurance. Section 2 discusses the motives for regulating capital in the presence of a government safety net and reviews relevant financial history in the United States and elsewhere. Section 3 summarizes the development of international bank capital standards, their shortcomings, and the problems with the reforms that the Basel Committee proposed in June 1999. In section 4 we propose an alternative set of reforms and explain the rationale for them. Section 5 describes how our proposed subordinated-debt requirement could be implemented. We conclude in section 6.

1

Bank Capital before Federal Deposit Insurance

The primary function of capital at a bank, as at any firm, is to absorb losses. If capital is insufficient to cover losses, unsatisfied claims by depositors or other debt holders would lead to insolvency. The amount of capital that a firm maintains should be determined by, among other factors, the probability that losses of specific magnitudes will be incurred. The greater the probability of large losses, the greater should be the amount of a firm's capital in relation to its other liabilities. In the absence of government guarantees, market forces would cause a bank's capital to vary with the risk of its assets, liabilities, and off-balance-sheet positions.

The fact that capital is available to absorb losses means that it acts as a buffer shielding senior claimants, such as depositors or their insurers, from the risk of loss and thus enables banks to attract uninsured depositors who are averse to risk. Capital also creates the appropriate incentives for the firm's managers. So long as banks maintain sufficient capital, the suppliers of capital will have an incentive to limit bank risk taking, since they would bear the initial cost of negative outcomes. ⁽³⁾

Before governments began protecting banks' depositors from loss, banks were subject to market discipline much like other corporations. Except for minimum capital requirements at the time a bank was chartered, the amount of its capital was determined by the market. If depositors believed that a bank had insufficient capital to protect the par value of their deposits, they could withdraw their funds, frequently on demand. That threat encouraged banks to maintain sufficient capital, commensurate with their portfolio risk, to ensure the continued confidence of their depositors so as to avoid runs.

In the years before the Great Depression--or before federal deposit insurance was enacted in 1933--banks in the United States failed, on average, at about the same rate as other firms, but losses to depositors at failed banks were lower than losses to creditors of other failed companies. ⁽⁴⁾ The liquid nature of bank liabilities that allows many depositors to remove their funds immediately ("run") upon signs of financial distress, accelerated the resolution process for troubled banks and thus limited losses to depositors who did not or could not run. The smaller losses reflected the preference of bank depositors for low-risk, liquid claims--an essential and special aspect of deposits as compared with debt claims on other firms. ⁽⁵⁾ Unlike other insolvent corporations in the United States, insolvent banks were resolved not through the usual legal bankruptcy process, but instead by their chartering agencies--the comptroller of the currency for national banks and the state banking agencies for state-chartered banks. Market forces in the form of depositor runs--and, at times, even the prospect of depositor runs--generally caused troubled banks to suspend operations. The authorities then resolved insolvent banks relatively quickly, before they could generate additional losses, unlike the bankruptcy process for other firms. ⁽⁶⁾

2

The Need to Regulate Bank Capital When Deposits Are Insured

Where a government safety net exists, one can no longer rely on market forces to establish the appropriate level of a bank's capital. Accordingly, in the United States, capital requirements for banks have become especially important since 1933, when the government first introduced federal deposit insurance. Initially, the insurance program covered accounts up to only $2,500. Over the following five decades, the insurance ceiling was intermittently raised, the last time being in 1980, when the coverage was lifted to $100,000, from the previous $40,000.

Even when the amount of insurance coverage is limited by law, depositors with amounts in excess of the limit may expect government protection de facto, especially if their banks are sufficiently large. In that event, it may be widely believed that policymakers fear that imposing losses on uninsured depositors at the failed bank could trigger runs by similarly situated depositors, even in healthy banks. In fact, in recent years depositors of banks in almost all countries, including the United States, have been "bailed out" by their governments, whether or not they were legally protected by deposit insurance. ⁽⁷⁾

In short, the central problem for bank regulation is that, although it may promote financial stability in the short run, deposit insurance (formal or informal) provided by a credible government authority tends to reduce banks' incentives to maintain adequate capital and endangers stability in the longer run. Depositors, whose accounts are fully insured and paid promptly upon default, do not discount the gross returns offered to them by banks for the risk that the institutions might become insolvent. Rather, insured depositors regard their claims on banks as riskless--equivalent to claims on the federal government--and they discount those claims at the risk-free rate. ⁽⁸⁾

As a result, depositors have less incentive to monitor banks' activities or to discipline banks because some or all of their deposits are protected from loss. In turn, protected banks face strong incentives to allow their capital ratios to fall and their portfolio risk to rise, if doing so increases the value of the implicit safety net subsidy they receive from not having to pay depositors to bear default risk when the guarantee is underpriced. Deposit insurance also increases the banking system's tolerance for incompetent and dishonest bankers, who unwittingly increase risk or operate their banks in self-serving ways, since the insurance reduces depositors' incentives to discriminate among banks according to their managers' competence or probity.

The combination of all of the aforementioned incentives for banks to increase their risk of insolvency without bearing the full cost of their errors--all due to deposit insurance--is known as the moral hazard problem. ⁽⁹⁾ In fact, the United States has seen many manifestations of moral hazard in the banking industry throughout its history--even before federal deposit insurance was introduced in 1933. Before that time, state-level deposit insurance systems operated in some states during and immediately after World War I. The surge in the relative prices of agricultural products during the war led some to believe--or hope--that a permanent shift in the price of those products had occurred. Deposit insurance in the states that offered it empowered those optimists by allowing them to charter and operate banks and raise insured funds, which they used to supply credit for cultivating marginal lands in the expectation of high future prices. Those banks also maintained higher leverage and higher loan-to-asset ratios than were the norm for noninsured institutions. When agricultural prices declined, bank failures were widespread, and state deposit insurance funds suffered enormous losses. ⁽¹⁰⁾

The problem of moral hazard resurfaced in the 1980s, when the United States experienced more bank failures than at any time since the Depression. That happened for a number of reasons. In the early part of the decade, interest rates and oil prices soared and sent the U.S. economy and other economies around the world into recession. Many large "money center" banks, in particular, suffered significant losses in their portfolios when their loans to less-developed countries proved not to be fully recoverable. Deposit insurance premiums were not increased, however, to reflect the increased risk among those institutions, nor were the banks required to recognize the full extent of their losses. As a result, certain of those weakly capitalized institutions, with the benefit of deposit insurance, took additional risks, primarily in commercial real estate lending, which later in the decade also proved to be highly costly. Other banks--many of them smaller institutions in states that did not allow their geographic diversification--suffered from a subsequent decline in oil and agricultural prices. Furthermore, the entire banking industry throughout the 1980s was subject to increasing competition from foreign banks, money market mutual funds, domestic finance companies, and domestic securities firms; that competition generally eroded banks' franchise values. The net result was a steady increase throughout the decade in the annual numbers of commercial bank failures that totaled almost 1,500--10 percent of the industry. The number of failures reached a post-Depression high of 206 in 1989, compared with an average of less than 10 per year from 1941 through 1981. ⁽¹¹⁾

The savings and loan (S&L) debacle of the 1980s is perhaps the most infamous and best-known example of how underpriced deposit insurance, in combination with regulatory forbearance from capital standards, can lead to excessive risk taking. In the absence of government deposit insurance, it is unlikely that any significant amount of short-term funds would have been placed with institutions allowed by law to operate with as little capital as 6 percent of assets but also restricted by law to investing primarily in long-term, fixed-interest obligations (mortgages). An increase in interest rates would cause the economic value of those assets to decline and, if the decline were greater than 6 percent, the corporations would be insolvent and the creditors--in this case, depositors--would incur losses. That is precisely what happened from 1979 through 1981, when a sharp increase in interest rates rendered most S&Ls economically insolvent. ⁽¹²⁾

Nevertheless, depositors kept their funds in the S&Ls for a simple reason: deposit insurance. Depositors rightly believed that the federal government would fulfill its promises to guarantee payment of their deposits. But instead of paying off depositors of economically insolvent thrifts, regulators allowed the institutions themselves to remain in business because the losses due to the interest rate spike far exceeded the meager resources of the thrift insurance fund, the Federal Savings and Loan Insurance Corporation. In effect, Congress and the regulators gambled not only that interest rates would come down, but that the institutions would not in the meantime take additional risks. Congress and the regulators were wrong. By allowing weak and insolvent thrifts to continue operating, policymakers in fact invited those institutions to assume the risk of regional recessions and even to "gamble for resurrection." When recessions occurred in the Southwest and New England and a large number of the gambles for resurrection turned sour, thrift institutions suffered even deeper losses. Ultimately, in 1989, U.S. taxpayers were called on to pay substantially all the costs of removing the insolvent institutions from the financial landscape and paying off depositors--a sum that eventually totaled about $150 billion.

Of course, policymakers have long been aware of the potential and actual costs of deposit insurance. President Franklin D. Roosevelt and many others recognized the problem before federal deposit insurance was adopted. Accordingly, legislators and regulators have tried various measures through the years to limit risk taking by depository institutions.

For most of the post-Depression era, those measures have largely taken the form of restrictions on bank and thrift activities--a reflection of the implicit (if not explicit) assumption that failures are due primarily to "overbanking" or "excessive" competition. For example, after the wave of bank failures during the Depression--9,000 during the 1929 through 1933 period alone--regulators granted few new bank charters, at least until the 1960s. Interest on time-deposit accounts was subjected to a ceiling, while banks were prohibited from paying interest on demand deposits. Congress also enacted various bills designed to constrain services that banks could offer or the assets they could hold that were alleged to be--but often were not--particularly risky. Examples include the Glass-Steagall Act of 1933, which largely separated investment and commercial banking, and the Bank Holding Company Act of 1956, which severely restricted activities that could be conducted in an affiliate of a bank. Furthermore, throughout much of the twentieth century, banks and thrifts have been limited in their ability to diversify their funding and lending by prohibitions on branching across state lines (and in many states, even on opening branches within the same state or county). In fact, the combination of those and other actions did reduce the number of bank failures almost to zero until the 1980s, although that result does not imply that the economy generally benefited from the combination of few failures and substantial restraints on banking.

In any event, market forces and technological advances in the 1980s and 1990s eventually induced policymakers to relax and ultimately repeal most of the Depression-era product and geographic restrictions on bank and thrift activities--a subject that is beyond the scope of this monograph. ⁽¹³⁾ In place of such restrictions, policymakers have gradually turned to regulating bank capital as the primary means for limiting risk taking by banks (and thrifts). That did not happen right away, however. Indeed, the initial reaction of Congress and regulators in the early 1980s to the thrift crisis was to avoid confronting the problem. As the capital accounts of S&Ls were depleted by losses from funding low-yielding mortgage loans with more costly deposits, policymakers actually reduced capital requirements directly by lowering the required ratio and indirectly by changing the rules governing recorded assets and liabilities. ⁽¹⁴⁾ In the Competitive Equality in Banking Act of 1987, Congress specifically provided "capital forbearance" for banks serving farmers that had experienced large losses.

In 1986, however, U.S. bank regulators took at least some initial steps to adopt common definitions of capital for commercial banks--and eventually their holding companies. The Federal Deposit Insurance Corporation (FDIC) and the Office of the Comptroller of the Currency moved first, followed by the Federal Reserve Board. They defined two classes of capital. Banks were required to maintain "primary capital"--shareholders' equity, perpetual preferred stock, reserves for loan and lease losses, some mandatory convertible debt, minority interests in consolidated subsidiaries, and regulatory net worth certificates--of at least 5.5 percent of total on-balance-sheet assets. The regulators also defined "secondary capital" as including limited-life preferred stock and subordinated notes and debentures, the latter being limited to 50 percent of primary capital. Together, primary and secondary capital, or "total capital," had to total at least 6 percent of total assets. As we discuss in the next section, the notion that not all bank capital was alike was copied in the international bank capital standards that were adopted shortly thereafter.

The common definitions of bank capital, however, did not prevent regulators (or Congress, in the case of agricultural banks) from granting troubled banks "forbearance" from meeting the capital requirements themselves. By the end of the 1980s, it was widely recognized that forbearance had contributed to the severe losses in both the banking and thrift industries. Accordingly, in 1991, Congress enacted the Federal Deposit Insurance Corporation Improvement Act, whose major objective was to compel bank regulators to enforce bank capital regulation. In particular, FDICIA required regulators to intervene promptly, at various stages well before a bank's capital is fully depleted, to prevent banks from taking added risks and to require them to raise additional capital. ⁽¹⁵⁾ The regulators were even instructed to assume control over banks that failed to comply with those requirements, before their book value equity was fully depleted. ⁽¹⁶⁾

That new system of "prompt corrective action" has been a success, at least so far, although it has not been tested by such severe shocks to bank asset values as those that occurred in the 1980s. ⁽¹⁷⁾ Nor is the system designed to deal with fraud, which often has been responsible for substantial losses. In addition, critics question whether regulators would be able or willing officially to recognize and act upon, in a timely fashion, significant systemwide losses to capital. That concern, in part, motivates continuing efforts--including this monograph--to improve further the design of regulatory capital standards.

In sum, the clear lesson of the U.S. experience since the Depression is that, as long as the government provides deposit insurance--de jure or de facto--and bears potential losses, it must also maintain an effective system of capital regulation to limit potential losses.

The United States is not alone in that regard. Governments in other countries also offer bank safety nets and thus share the same incentive problems that our country has experienced. For example, capital-asset ratios of banks in the European Community have decreased as governments have indicated that depositors are unlikely to experience losses at failed banks. ⁽¹⁸⁾ Over the past twenty years, an unprecedented wave of banking system insolvencies has plagued both developed and developing countries around the globe. In several cases--including the 1990s banking collapses in Venezuela, Japan, Korea, Indonesia, Thailand, and Mexico--the estimated banking system losses have been particularly pronounced. ⁽¹⁹⁾

In the process, economies have suffered two types of losses. Society as a whole loses on account of the misallocation of resources that weak and failing banks direct into unproductive investments so that through time gross domestic product declines below its potential growth rate. Furthermore, taxpayers suffer when they are saddled with the costs of paying for the cleanup of failed institutions and protecting the depositors. For example, in Japan, the world's second largest economy, the transfer costs alone have been estimated at about 15 percent of the country's GDP, and they are significantly higher in countries like Argentina, Indonesia, Korea, and Thailand as compared with savings and loan resolution costs in the United States of about 3 percent of GDP. ⁽²⁰⁾

3

International Regulation of Capital under the Basel Standards

Given the magnitude of past banking problems in many different countries, a key question is whether any international financial standards-setting body, such as the Basel Committee, can set standards that are adequate to the task. Nevertheless, the Basel Committee has attempted to do so, at least for the banking systems in its industrialized member countries, for a little more than a decade. We now discuss how those standards came into being, how they have been amended, what criticisms have been leveled against the standards, the most recent proposal for changing the standards, and the views of the SFRC about the Basel Committee's proposal. We ultimately conclude that the risk-weighting system that is at the heart of the standards is fundamentally flawed and may even have led to counterproductive behavior by banks that have attempted to "game" the system. As we explain at the end of this section, the most recent Basel proposal has not cured that and other problems.

The Development of the Initial Basel Standards

As U.S. bank regulators began to refine and consider tightening bank capital standards in the 1980s, they also grew concerned that unilateral increases in those standards in this country might place American banks at a competitive disadvantage relative to banks in other countries that were subject to more lenient capital rules. Special concerns were aimed at banks in Japan, which in the 1980s had grown very rapidly--along with that country's financial system--and which were then beginning to make major advances in the U.S. banking market. U.S. regulators feared that unless some attempt was made to coordinate capital standards across countries, individual countries might relax their standards as a means of enhancing the international competitive positions of their home country banks, or to protect those banks against competition by foreign banks in home markets.

There was also concern that the reliability of the global payments system--in which all international banks operated--required minimal international standards for all participating banks. In fact, the effort to establish international banking standards began shortly after the 1974 failure of the Bankhaus Herstatt, a West German bank whose unfulfilled foreign currency obligations to American and other banks caused serious dislocations in foreign exchange and international interbank markets. Consequently, in 1975 the G-10 countries plus Luxembourg and Switzerland formed the Basel Committee on Banking Supervision, whose initial mission was to develop principles for the supervision of internationally active banks.

The committee did not take long to focus as well on capital-adequacy standards and it thus transformed itself into a regulatory body. That happened in the 1980s, when a number of international banks suffered under the weight of nonperforming loans to less-developed countries, and prompted financial supervisors in the Basel member countries to grow increasingly concerned that further weakening in bank capital might threaten the stability of the global financial system.

While all member countries regulated the capital of their own banks, each had a different approach and definition of capital. Accordingly, the Basel Committee began in the 1980s to seek ways to promote international convergence of capital-adequacy measurement and standards, and to achieve major objectives including removing

incentives for excessive risk taking by banks in their loan and securities portfolios, extending capital requirements to off-balance-sheet positions, and eliminating differences in capital as a source of competitive imbalance in banking among the Basel member countries. The committee's members believed that those goals could best be accomplished by adopting minimum capital standards for internationally active banks.

The most difficult negotiations involved the definition of capital. All countries regarded shareholders' equity as capital, but disagreements arose over other components of regulatory capital. The Germans regarded the broadening by the Basel Committee of any definition beyond shareholders' equity as undermining the rigor of German capital requirements. France, which had a number of state-owned banks that would have found it difficult to increase shareholders' equity, argued for including a substantial amount of subordinated debt in the definition. The United States, which had counted loan loss reserves as part of regulatory capital, argued that such a practice should be continued. The Japanese, whose banks had substantial unrealized capital gains in securities holdings, argued that such gains should be counted as assets and, hence, as higher equity.

The 1988 Basel Accord Capital Standard and Later Amendments

The resulting compromise definition--reflected in the initial standards--owed more to banks' existing circumstances than to economic logic. Specifically, the Basel Committee established two kinds of capital: core or "tier 1" capital that was mainly shareholders' equity and noncumulative perpetual preferred stock and supplementary or "tier 2" capital that included subordinated debt (to please the French), some loan loss reserves (to please the United States), and 45 percent of unrealized capital gains on securities (to please the Japanese).

The committee also specified a risk-weighting framework to tie capital requirements to the perceived credit risks of assets and off-balance-sheet commitments. Government bonds of the countries that were members of the Organization for Economic Cooperation and Development (OECD) (which includes all members of the Basel Committee) were assigned a zero risk weight, all short-term interbank loans and all long-term interbank loans to banks headquartered in OECD countries a 20 percent risk weight, home mortgages a 50 percent risk weight, and most other loans a 100 percent risk weight. Off-balance-sheet exposures were converted into loan-equivalent values and also assigned risk weights.

The initial standards required internationally active banks to meet two minimum capital ratios, both computed as a percentage of the risk-weighted (both on- and off-balance-sheet) assets. The minimum tier 1 ratio was 4 percent of risk-weighted assets, while total capital (tiers 1 and 2) had to exceed 8 percent of risk-weighted assets. Market risks, such as interest or exchange rate risks, liquidity risks, and operational risks of banks were not addressed by the Basel standards of 1988. Those omissions were among the targets of critics of the initial standards.

In April 1993 the Basel Committee began an effort to refine the initial standards by proposing to require banks to hold capital against market risks in their "trading book"--or losses that might result from adverse changes in security, currency, and commodity prices and interest rates on securities held for sale. The required capital would be measured according to a so-called building-block approach that the European Union had adopted. Under such an approach, risks of individual types were added together and capital required against them. But several major banks found that proposal too primitive and too different from the way they managed their market risks internally.

The Basel Committee responded in a fashion unusual at the time for a body of regulators. Instead of continuing to insist on a "one-size-fits-all" standard, the committee proposed that bank capital required for market risk be based on only the supervised use of banks' internal market-risk models. That approach was adopted with the 1996 Amendment to the Capital Accord, which allows banks to use their internal models for measurement of market risk instead of the building block approach, subject to a number of qualitative and quantitative criteria, including successful back-testing of those models.

The Basel Committee has yet to institute standards for other risks--including interest rate risk in the "banking book" (primarily loans held until maturity), currency risk, liquidity risk, and operational risks--but the committee continues to search for the regulatory "holy grail." The quest not only is likely to be fruitless, but as the following review of criticisms indicates, will probably have adverse consequences.

Criticisms of the Basel Standards

The Basel standards have been subjected to a series of criticisms. We review them here, roughly in ascending order of severity.

At the simplest level, the standards have not achieved one of their central objectives: to level the playing field in banking across countries. Scott and Iwahara (1994), for example, compared the implementation of the Basel Accord in the United States and Japan and concluded that the accord had no impact on competitiveness. The authors also showed that other factors such as taxes, accounting requirements, disclosure laws, implicit and explicit deposit guarantees, social overhead expenditures, employment restrictions, and insolvency laws, also affect the competitiveness of an institution and, in principle, its appropriate capital-asset ratio. Consequently, imposing the same capital standard on all institutions that differ with regard to those other factors is unlikely to enhance competitive equity. On the contrary, uniform capital standards may widen rather than narrow competitive differences.

The standards have also been criticized for failing to assign "correct" risk weights and for failing to promote bank safety effectively. The Basel Committee itself has recognized the validity of many of those criticisms, particularly regarding the risk weightings. As we have noted, the risk weights do not attempt to take account of market risks, liquidity risk, and operational risks that may be important sources of insolvency exposure for banks. Although the risk weights attempt to reflect credit risk, they are not based on market assessments but instead favor claims on banks headquartered in OECD countries and OECD governments, and on residential mortgages. Furthermore, the risk weights fail to distinguish among gross differences in the credit quality of borrowers within a risk class. Thus, banks engage in substantial arbitrage among loans whose risks, as determined by the market, differ from the risk weights assigned by the Basel Committee. The problems are compounded by the fact that the Basel standards are computed on the basis of book-value accounting measures of capital, not market values. Accounting practices vary significantly across the G-10 countries and often produce results that differ markedly from market assessments.

Perhaps the most fundamental problem with the current Basel standards stems from the fact that they attempt to define and measure bank portfolio risk categorically by placing different types of bank exposures into separate "buckets." Banks are then required to maintain minimum capital proportional to a weighted sum of the amounts of assets in the various risk buckets. That approach incorrectly assumes, however, that risks are identical within each bucket and that the overall risk of a bank's portfolio is equal to the sum of the risks across the various buckets.

Such a conception of portfolio risk bears little, if any, relation to the true portfolio risk of banks. All activities or loans within a particular category do not have the same market-based credit risk. For example, not all mortgages are exactly or even approximately half as risky as all commercial loans (reflecting the assigned risk weights), and a loan to General Electric is not as risky as a loan to Guatemala or George's Pizza Parlor. Moreover, the aggregate risk of a bank is not equal to the sum of its individual risks; diversification through the pooling of risks can significantly reduce the overall portfolio risk of a bank. ⁽²¹⁾ Indeed, a well-established principle of finance is that the combination in a single portfolio of assets with different risk characteristics can produce less overall risk than merely adding up the risks of the individual assets.

The problems inherent in assigning risk weights in the Basel standards are compounded by the inappropriate division of bank capital into different "tiers." In the process, the Basel Committee implicitly favors equity over other forms of capital--specifically, subordinated debt. As we discuss at length in section 5, the preference for equity not only is unwarranted but also may be counterproductive since subordinated debt--which is included in tier 2 capital, but not in tier 1--often can be superior to equity from a regulatory standpoint.

Recent financial crises involving international banks have highlighted several additional weaknesses in the Basel standards that permitted, and in some cases even encouraged, excessive risk taking and misallocations of bank credit. Notably, Asian banks' short-term borrowing of foreign currencies was a major source of vulnerability in the countries most seriously affected by the Asian financial crisis. The current Basel standards contributed to that problem by assigning a relatively favorable 20 percent risk weight to short-term interbank lending--only one-fifth as large as the weight assigned to longer-term lending or to lending to most private nonbank borrowers. Putting aside the important issue of whether the standards should have assigned different risk weights for short-term lending to banks in the developed and in the developing world--a distinction not captured by the current system of weighting asset risks--it is clear that the much lower risk weight given to interbank lending than to other types of bank loans encouraged some large internationally active banks to lend too much for short durations to banks in Southeast Asia. Those banks reloaned the funds in domestic currency at substantially higher rates and assumed large foreign exchange rate risk. One would expect those distortions to be most pernicious for banks that are capital-constrained. Therefore, it is not surprising that Japanese banks, which have been weakly capitalized throughout the 1990s, had accumulated the heaviest concentrations of claims on faltering Asian banks.

The current standards also assign a zero risk weight to all sovereign debt issued by countries belonging to the OECD. Although sovereign debt was not at the center of the Asian financial crises, it played a central role in the earlier Mexican financial and currency crisis of 1994-1995. Significantly, Mexico and South Korea--both of which experienced substantial bank insolvencies--are now members of the OECD; thus, the bonds issued by their governments are subject to the zero risk weight.

The last two weaknesses in particular are by now widely accepted--even among members of the Basel Committee--and have created a sense of urgency for reforming international capital standards.

The Basel Committee's Proposed Reforms: Description and Evaluation

In June 1999 the Basel Committee released and invited public comment on a proposal outlining potential improvements in its existing system of capital regulation. The SFRC has prepared this monograph in response to that invitation.

The committee proposed three important modifications to the current credit-risk standards. The first feature would require that bank loan risk weightings reflect the ratings assigned to borrowers by such private credit-rating agencies as Standard and Poor's or Moody's. As part of that first reform, the range of the risk weights was also increased (from 0 to 100 percent to 0 to 150 percent). The second element would permit banks' own internal risk-rating systems to play a greater role in determining capital requirements. The third part of the proposal contemplates extending the current "internal models" approach to market risk to setting capital requirements for the bank as a whole. The SFRC believes that all three parts of the proposal have serious defects.

The enlargement in the number of risk buckets and the link to external assessments of credit risk reflect the commendable objective of improving the measures of credit risk in bank loan portfolios. Nonetheless, the first element of the proposal entails several obvious distortions. Differences in risk weights across some risk buckets are disproportionate to credit spreads for comparably rated corporate bonds and actual historical loss experience. For example, under the proposal, loans to AA-rated corporate borrowers will require one-fifth the capital of a loan to an A-rated corporate borrower, even though the historical loss rates are quite similar. Meanwhile, variations within some risk buckets remain large relative to variations across risk buckets. A-rated companies have the same risk weight as companies rated lower, including those below investment grade (BB+ to B), despite the fact that historical loss rates are vastly different. Finally, all unrated corporate borrowers are treated as favorably as A-rated borrowers and more favorably than borrowers rated below B. The disparity presumably reflects an attempt to enlist support for the proposed system from countries where most firms are not rated but is unlikely to represent the true credit quality of many unrated borrowers.

Even if the proposed risk buckets were less arbitrary than the existing risk buckets, the new proposal retains the flawed summation-of-risk-buckets approach to measuring the risk of a loan portfolio. Furthermore, the larger number of risk buckets appears to confuse precision with accuracy.

The proposal to rely on ratings agencies for assigning loans to risk buckets raises additional difficulties. As Altman and Saunders (1999) have shown, ratings agencies move slowly, and changes in ratings lag changes in actual credit quality, so that the ratings have a questionable ability to predict default. Indeed, the record of the ratings agencies before the recent Asian financial crisis was particularly poor.

Furthermore, the use of private credit ratings to measure loan risk may adversely affect the quality of ratings. If regulators shift the burden of assessing the quality of bank loans to ratings agencies, those regulators risk undermining the quality of credit ratings to investors. Ratings agencies would have incentives to engage in the financial equivalent of "grade inflation" by supplying favorable ratings to banks seeking to lower their capital requirements. If the ratings agencies debase the level of ratings, while maintaining ordinal rankings of issuers' risks, the agencies may be able to aviod a loss in revenue because investors still find their ratings useful. If incumbent firms do not succumb to those added incentives, new entrants are likely to arise to meet the demands for laxity. Indeed, because entities based in the United States or the United Kingdom currently dominate the ratings business, regulatory authorities in other countries would be strongly tempted to approve new domestic ratings agencies without necessarily having full regard for the quality of their ratings. In short, if the primary constituency for new ratings is banks for regulatory purposes rather than investors, standards are likely to deteriorate. ⁽²²⁾

The second part of the Basel proposal, greater reliance on banks' own internal risk ratings, may be an improvement, but the current proposal raises more questions than it answers. Specifically, the proposal does not indicate how regulators will evaluate the accuracy of banks' own internal credit-risk ratings or how they would be translated into capital requirements. Nor does it explain how it would achieve comparability across the variety of internal rating systems in different banks. ⁽²³⁾ Most important, the proposal does not explain how regulators will enforce the ratings that banks produce or impose sanctions if the ratings turn out to be inaccurate and capital is insufficient or depleted. In any event, even if an effective enforcement mechanism were in place, summing across risk buckets is just as deficient when the risk buckets are determined by internal ratings as when they are determined by external risk ratings or the current arbitrary regulatory distinctions.

Banks' own portfolio risk models solve the aggregation problem and, in principle, measure precisely the risk that should concern the regulatory authorities--the risk of loss for a bank's whole portfolio. The Basel Committee has taken note of recent advances in modeling portfolios of credit risk but has determined--correctly, in our view--that the state of the art is not sufficiently advanced to warrant relying on internal models to determine capital requirements.

Nevertheless, one could imagine a system in which regulators do not concern themselves with the validation of banks' own credit-risk models but simply require banks to commit an amount of capital to absorb all credit risks. Regulators would then levy heavy penalties on banks ex post for ex ante underestimation of their risks. That possible system has several problems, however.

One major hurdle is determining how to make any penalties credible. The staff at the Federal Reserve has proposed a similar "precommitment" approach for setting capital standards for market risks only. Critics have questioned whether regulators can credibly impose penalties on banks that fail to set aside sufficient capital for such risks; those critics suggest that "kicking banks when they're down" may be impractical or even counterproductive. Whatever view one takes about precommitment in the context of trading risks, the same objection should clearly apply with much greater force to any similar system for setting capital standards for bank activities in their entirety--or even for just the credit risks in the loan portfolio. Who pays the penalties when the bank itself is insolvent? To be sure, regulators may be able to take advantage of the prompt corrective action feature of FDICIA and assume control over an institution that has suffered large losses of capital.

Thus, it is conceivable that the gradual penalties embodied in the structured early intervention and resolution system--which we describe below--could help to make the precommitment approach, as applied to the entire bank, credible.

Nevertheless, other problems remain. Information about both the standards and bank compliance would continue to remain solely in the hands of regulators. As long as that is the case, regulators have the ability and incentive to engage in forbearance if the standards set by individual banks prove ex post to be excessively low.

Another key problem with relying on banks' internal models is the inconsistency among banks in measuring risk or even in defining such basic concepts as "default" or "loss." Furthermore, current risk-assessment models are hampered by a lack of sufficient historical data to provide reliable estimates of loan defaults. Indeed, the Basel Committee issued a report in January 2000 that highlighted those problems and found that even sophisticated banks do not express great confidence in the results their models produce. ⁽²⁴⁾

The SFRC believes that if the Basel Committee truly wants to take advantage of market information and discipline in influencing bank behavior, it should move in a different direction--one that relies, in part, on enforceable capital standards but that also makes much greater use of the market itself to discipline banks against taking excessive risks and regulators against pursuing forbearance. Below we outline such a system.

4

A Framework for Reform

The proposals we now describe have a simple set of objectives. We believe that policymakers should ensure that banks maintain sufficient capital to absorb almost all losses that they might incur. Furthermore, a mechanism must be in place that harnesses market forces so as to induce banking authorities to act promptly and effectively to bring banks into compliance with prevailing capital standards or at least to limit the losses that depositors, prudently run banks, and taxpayers may be forced to bear.

Specifically, we urge the adoption of five independent, but mutually reinforcing, recommendations. First, capital should be measured so that it substantially reflects market values and should be disclosed promptly at regular intervals. Second, banks should maintain a level of capital that is sufficient to absorb almost all losses that would be incurred by reasonably prudent management. Third, banks should be allowed to meet their capital requirements by issuing an unlimited amount of subordinated debentures--appropriately structured, among other things, to prevent banks from redeeming them before the banking authorities can act--so that the cost of capital to banks is no greater than it would be for corporations with debt that is not government-insured. Fourth, bank regulators in the United States should improve the current system of structured early intervention. Other countries that have not yet done so should adopt such a system. Finally, banking authorities should enhance market discipline on both banks and themselves by requiring large banks to issue and regularly reissue a special form of subordinated debt--a recommendation we explore in depth in section 5.

Measuring Capital

For regulatory purposes, bank capital should be the difference between the market values of assets and senior (insured) bank liabilities. Equity is the basic form of capital, but, from the standpoint of depositors, regulators, or the deposit insurer, anything that is effectively junior to their respective claims and absorbs losses serves as capital. ⁽²⁵⁾ If the standard of measurement applying to assets significantly overstates their market values, the protection and incentive will be limited or illusory.

Although market valuations are not yet widely incorporated in financial reporting, the SFRC believes that reasonably close approximations to market-value accounting are feasible and relatively inexpensive for financial institutions to adopt. ⁽²⁶⁾ Unlike nonfinancial firms, banks have relatively small investments in assets for which current market values are particularly difficult to measure or estimate, such as land, buildings, equipment, work-in-process, patents, and trademarks. In contrast, most assets on banks' balance sheets can be stated at or close to their current market values.

For example, securities (including derivatives) are now stated at market values either directly or in footnotes. ⁽²⁷⁾ Although most bank loans do not have readily ascertainable market values, particularly when those loans are made to small companies without publicly traded securities, a close approximation is available. After all, loan pricing reflects bank assessments using benchmarks from market yields on comparable risks. Under generally accepted accounting principles (GAAP), if correctly applied, however, banks are required to estimate an allowance for loan losses--a measure intended to reduce loans receivable to their net realizable value. In addition, the values of both bank loans and deposit liabilities can easily be adjusted to reflect changes in interest rates. ⁽²⁸⁾

Some other technical issues would need to be resolved if regulators were to use market values to measure bank capital. One of them relates to the fact that under GAAP, the costs of intangible assets that are developed rather than purchased (such as advertising, patents, employee training, and customer goodwill) are typically charged to expense rather than capitalized. The principal intangible assets held by banks are the core-deposit intangible and the charter value. The value of core deposits can be estimated with standard models that use the difference between a bank's deposit interest rate plus operating costs and the interest rate on purchased funds, coupled with the time pattern of deposit flows, to estimate the present value of the deposit relationship. Charter values are much more difficult to measure. For regulatory purposes, it is preferable to omit them, since the charter is likely to be essentially without value should a bank become insolvent. The equivalent loan amounts of off-balance-sheet contingent liabilities, such as loan guarantees, should also be included as assets against which regulatory capital is required, as is now done in the Basel standards. ⁽²⁹⁾

In short, a compelling case exists for regulators to measure bank capital on the basis of market rather than book values of assets and liabilities. We note that, in any event, large banks are likely on their own to improve their disclosure of useful information if required to maintain a certain amount of specially structured subordinated debt as a source of funds, as we discuss in section 5. This is so because investors whose funds cannot be withdrawn on demand and are not insured, explicitly or implicitly, are likely to demand higher interest rates on debt issued by banks that do not provide the information that enables investors adequately to assess bank risk on a timely basis. Consequently, a subordinated-debt requirement should generate added demand for and supply of meaningful accounting numbers that come closer than traditional GAAP to reporting economic values of assets, liabilities, and capital.

Minimum Required Level of Capital

Because deposit insurance greatly reduces the incentives for insured depositors to be concerned about banks' risk taking, banking regulators must require banks to operate with a sufficient amount of shareholders' capital to cover losses that banks might incur and to provide an effective incentive for banks to manage their operations in a prudent manner. We believe that the risk-weighted system of computing required bank capital should be abandoned in favor of a simple leverage ratio (which does not make use of risk weights) and that the total capital of a bank should be increased.

The Case for a Leverage Ratio. We earlier argued that the Basel standards' distinction between the two tiers of capital was driven more by the need to negotiate an international compromise than by economic logic. But that is not all. Although arguments can be made both for and against the current and proposed revised risk weights on different asset and activity categories, on balance, we believe that risk weights distort lending activity and are generally unrelated to market risk differences across and within categories of assets. Accordingly, we favor the elimination of regulatory risk weights on assets, on and off the balance sheet. The better course is for required capital to be calculated on the basis of the market values of bank assets and contingent liabilities.

To be sure, it is not obvious that uniform (and therefore, necessarily inaccurate) risk weights are less distortionary than multiple differential (and also inaccurate) risk weights. Nevertheless, a limit on leverage, without the complication of risk weighting, has the advantage of greater simplicity and is less misleading, since it does not purport to weigh the relative risks associated with broad categories of assets. Moreover, a straightforward leverage requirement reduces banks' incentives to manipulate required capital by shifting assets among risk-weight categories, when those shifts do not represent real changes in portfolio risk. Nor would banks benefit from making loans to weak banks and countries simply because a lower risk factor is applied to those loans.

Of course, a single "risk" category, like a set of risk categories, gives opportunistic bankers an incentive to shift investments to assets with higher risks. For example, bankers could transform consumer loans into securitized assets by selling th