Risk-Based Capital Guidelines; Implementation of New Basel Capital Accord

DEPARTMENT OF THE TREASURY

Office of the Comptroller of the Currency

12 CFR Part 3

[Docket No. 03-14]

RIN Number 1557-AC48

FEDERAL RESERVE SYSTEM

12 CFR Parts 208 and 225

[Regulations H and Y; Docket No. R-1154]

FEDERAL DEPOSIT INSURANCE CORPORATION

12 CFR Part 325

RIN 3064-AC73

DEPARTMENT OF THE TREASURY

Office of Thrift Supervision

12 CFR Part 567

[No. 2003-27]

RIN 1550-AB56

Risk-Based Capital Guidelines; Implementation of New Basel Capital Accord

AGENCIES:

Office of the Comptroller of the Currency, Treasury; Board of Governors of the Federal Reserve System; Federal Deposit Insurance Corporation; and Office of Thrift Supervision, Treasury.

ACTION:

Advance notice of proposed rulemaking.

SUMMARY:

The Office of the Comptroller of the Currency (OCC), the Board of Governors of the Federal Reserve System (Board), the Federal Deposit Insurance Corporation (FDIC), and the Office of Thrift Supervision (OTS) (collectively, the Agencies) are setting forth for industry comment their current views on a proposed framework for implementing the New Basel Capital Accord in the United States. In particular, this advance notice of proposed rulemaking (ANPR) describes significant elements of the Advanced Internal Ratings-Based approach for credit risk and the Advanced Measurement Approaches for operational risk (together, the advanced approaches). The ANPR specifies criteria that would be used to determine banking organizations that would be required to use the advanced approaches, subject to meeting certain qualifying criteria, supervisory standards, and disclosure requirements. Other banking organizations that meet the criteria, standards, and requirements also would be eligible to use the advanced approaches. Under the advanced approaches, banking organizations would use internal estimates of certain risk components as key inputs in the determination of their regulatory capital requirements.

DATES:

Comments must be received no later than November 3, 2003.

ADDRESSES:

Comments should be directed to:

OCC:

Please direct your comments to: Office of the Comptroller of the Currency, 250 E Street, SW., Public Information Room, Mailstop 1-5, Washington, DC 20219, Attention: Docket No. 03-14; fax number (202) 874-4448; or Internet address:

regs.comments@occ.treas.gov.

Due to delays in paper mail delivery in the Washington area, we encourage the submission of comments by fax or e-mail whenever possible. Comments may be inspected and photocopied at the OCC's Public Information Room, 250 E Street, SW., Washington, DC. You may make an appointment to inspect comments by calling (202) 874-5043.

Board:

Comments should refer to Docket No. R-1154 and may be mailed to Ms. Jennifer J. Johnson, Secretary, Board of Governors of the Federal Reserve System, 20th Street and Constitution Avenue, NW., Washington, DC 20551. However, because paper mail in the Washington area and at the Board of Governors is subject to delay, please consider submitting your comments by e-mail to

regs.comments@federalreserve.gov.

, or faxing them to the Office of the Secretary at (202) 452-3819 or (202) 452-3102. Members of the public may inspect comments in Room MP-500 of the Martin Building between 9 a.m. and 5 p.m. weekdays pursuant to § 261.12, except as provided by § 261.14, of the Board's Rules Regarding Availability of Information, 12 CFR 261.12 and 261.14.

FDIC:

Written comments should be addressed to Robert E. Feldman, Executive Secretary, Attention: Comments, Federal Deposit Insurance Corporation, 550 17th Street, NW., Washington, DC 20429. Commenters are encouraged to submit comments by facsimile transmission to (202) 898-3838 or by electronic mail to

Comments@FDIC.gov.

Comments also may be hand-delivered to the guard station at the rear of the 550 17th Street Building (located on F Street) on business days between 8:30 a.m. and 5 p.m. Comments may be inspected and photocopied at the FDIC's Public Information Center, Room 100, 801 17th Street, NW., Washington, DC between 9 a.m. and 4:30 p.m. on business days.

OTS:

Send comments to Regulation Comments, Chief Counsel's Office, Office of Thrift Supervision, 1700 G Street, NW., Washington, DC 20552, Attention: No. 2003-27. Delivery: Hand deliver comments to the Guard's desk, east lobby entrance, 1700 G Street, NW., from 9 a.m. to 4 p.m. on business days, Attention: Regulation Comments, Chief Counsel's Office, Attention: No. 2003-27. Facsimiles: Send facsimile transmissions to FAX Number (202) 906-6518, Attention: No. 2003-27. E-mail: Send e-mails to

regs.comments@ots.treas.gov

, Attention: No. 2003-27, and include your name and telephone number. Due to temporary disruptions in mail service in the Washington, DC area, commenters are encouraged to send comments by fax or e-mail, if possible.

FOR FURTHER INFORMATION CONTACT:

OCC:

Roger Tufts, Senior Economic Advisor (202-874-4925 or

roger.tufts@occ.treas.gov)

, Tanya Smith, Senior International Advisor (202-874-4735 or

tanya.smith@occ.treas.gov

), or Ron Shimabukuro, Counsel (202-874-5090 or

ron.shimabukuro@occ.treas.gov

).

Board:

Barbara Bouchard, Assistant Director (202/452-3072 or

barbara.bouchard@frb.gov

), David Adkins, Supervisory Financial Analyst (202/452-5259 or

david.adkins@frb.gov

), Division of Banking Supervision and Regulation, or Mark Van Der Weide, Counsel (202/452-2263 or

mark.vanderweide@frb.gov

), Legal Division. For users of Telecommunications Device for the Deaf (“TDD”) only, contact 202/263-4869.

FDIC:

Keith Ligon, Chief (202/898-3618 or

kligon@fdic.gov

), Jason Cave, Chief (202/898-3548 or

jcave@fdic.gov

), Division of Supervision and Consumer Protection, or Michael Phillips, Counsel (202/898-3581 or

mphillips@fdic.gov

).

OTS:

Michael D. Solomon, Senior Program Manager for Capital Policy (202/906-5654); David W. Riley, Project Manager (202/906-6669), Supervision Policy; or Teresa A. Scott, Counsel (Banking and Finance) (202/906-6478), Regulations and Legislation Division, Office of the Chief Counsel, Office of Thrift Supervision, 1700 G Street, NW., Washington, DC 20552.

SUPPLEMENTARY INFORMATION:

I. Executive Summary

A. Introduction

B. Overview of the New Accord

C. Overview of U.S. Implementation

The A-IRB Approach for Credit Risk

The AMA for Operational Risk

Other Considerations

D. Competitive Considerations

II. Application of the Advanced Approaches in the United States

A. Threshold Criteria for Mandatory Advanced Approach Organizations

Application of Advanced Approaches at Individual Bank/Thrift Levels

U.S. Banking Subsidiaries of Foreign Banking Organizations

B. Implementation for Advanced Approach Organizations

C. Other Considerations

General Banks

Majority-Owned or Controlled Subsidiaries

Transitional Arrangements

III. Advanced Internal Ratings-Based Approach (A-IRB)

A. Conceptual Overview

Expected Losses versus Unexpected Losses

B. A-IRB Capital Calculations

Wholesale Exposures: Definitions and Inputs

Wholesale Exposures: Formulas

Wholesale Exposures: Other Considerations

Retail Exposures: Definitions and Inputs

Retail Exposures: Formulas

A-IRB: Other Considerations

Purchased Receivables

Credit Risk Mitigation Techniques

Equity Exposures

C. Supervisory Assessment of A-IRB Framework

Overview of Supervisory Framework

U.S. Supervisory Review

IV. Securitization

A. General Framework

Operational Criteria

Differences Between the General A-IRB Framework and the A-IRB Approach for Securitization Exposures

B. Determining Capital Requirements

General Considerations

Capital Calculation Approaches

Other Considerations

V. AMA Framework for Operational Risk

A. AMA Capital Calculation

Overview of the Supervisory Criteria

B. Elements of an AMA Framework

VI. Disclosure

A. Overview

B. Disclosure Requirements

VII. Regulatory Analysis

A. Executive Order 12866

B. Regulatory Flexibility Act

C. Unfunded Mandates Reform Act of 1995

D. Paperwork Reduction Act

List of Acronyms

I. Executive Summary

A. Introduction

In the United States, banks, thrifts, and bank holding companies (banking organizations or institutions) are subject to minimum regulatory capital requirements. Specifically, U.S. banking organizations must maintain a minimum leverage ratio and two minimum risk-based ratios.

1

The current U.S. risk-based capital requirements are based on an internationally agreed framework for capital measurement that was developed by the Basel Committee on Banking Supervision (Basel Supervisors Committee or BSC) and endorsed by the G-10 Governors in 1988.

2

The international framework (1988 Accord) accomplished several important objectives. It strengthened capital levels at large, internationally active banks and fostered international consistency and coordination. The 1988 Accord also reduced disincentives for banks to hold liquid, low-risk assets. Moreover, by requiring banks to hold capital against off-balance-sheet exposures, the 1988 Accord represented a significant step forward for regulatory capital measurement.

1

The leverage ratio measures regulatory capital as a percentage of total on-balance-sheet assets as reported in accordance with generally accepted accounting principles (GAAP) (with certain adjustments). The risk-based ratios measure regulatory capital as a percentage of both on- and off-balance-sheet credit exposures with some gross differentiation based on perceived credit risk. The Agencies' capital rules may be found at 12 CFR Part 3 (OCC), 12 CFR Parts 208 and 225 (Board), 12 CFR Part 325 (FDIC), and 12 CFR Part 567 (OTS).

2

The BSC was established in 1974 by the central-bank governors of the Group of Ten (G-10) countries. Countries are represented on the BSC by their central bank and also by authorities with bank supervisory responsibilities. Current member countries are Belgium, Canada, France, Germany, Italy, Japan, Luxembourg, the Netherlands, Spain, Sweden, Switzerland, the United Kingdom, and the United States. The 1988 Accord is described in a document entitled “International Convergence of Capital Measurement and Capital Standards.” This document and other documents issued by the BSC are available through the Bank for International Settlements website at www.bis.org.

Although the 1988 Accord has been a stabilizing force for the international banking system, the world financial system has become increasingly more complex over the past fifteen years. The BSC has been working for several years to develop a new regulatory capital framework that recognizes new developments in financial products, incorporates advances in risk measurement and management practices, and more precisely assesses capital charges in relation to risk. On April 29, 2003, the BSC released for public consultation a document entitled “The New Basel Capital Accord” (New Accord) that sets forth proposed revisions to the 1988 Accord. The BSC will accept industry comment on the New Accord through July 31, 2003 and expects to issue a final revised Accord by the end of 2003. The BSC expects that the New Accord would have an effective date for implementation of December 31, 2006.

Accordingly, the Agencies are soliciting comment on all aspects of this ANPR, which is based on certain proposals in the New Accord. Comments will assist the Agencies in reaching a determination on a number of issues related to how the New Accord would be proposed to be implemented in the United States. In addition, in light of the public comments submitted on the ANPR, the Agencies will seek appropriate modifications to the New Accord.

B. Overview of the New Accord

The New Accord encompasses three pillars: minimum regulatory capital requirements, supervisory review, and market discipline. Under the first pillar, a banking organization must calculate capital requirements for exposure to both credit risk and operational risk (and market risk for institutions with significant trading activity). The New Accord does not change the definition of what qualifies as regulatory capital, the minimum risk-based capital ratio, or the methodology for determining capital charges for market risk. The New Accord provides several methodologies for determining capital requirements for both credit and operational risk. For credit risk there are two general approaches; the standardized approach (essentially a package of modifications to the 1988 Accord) and the internal ratings-based (IRB) approach (which uses an institution's internal estimates of key risk drivers to derive capital requirements). Within the IRB approach there is a foundation methodology, in which certain risk component inputs are provided by supervisors and others are supplied by the institutions, and an advanced methodology (A-IRB), where institutions themselves provide more risk inputs.

The New Accord provides three methodologies for determining capital requirements for operational risk; the basic indicator approach, the standardized approach, and the advanced measurement approaches (AMA). Under the first two methodologies, capital requirements for operational risk are fixed percentages of specified, objective risk measures (for example, gross income). The AMA provides the flexibility for an institution to develop its own individualized approach for measuring operational risk, subject to supervisory oversight.

The second pillar of the New Accord, supervisory review, highlights the need for banking organizations to assess their capital adequacy positions relative to overall risk (rather than solely to the minimum capital requirement), and the need for supervisors to review and take appropriate actions in response to those assessments. The third pillar of the New Accord imposes public disclosure requirements on institutions that are intended to allow market participants to

assess key information about an institution's risk profile and its associated level of capital.

The Agencies do not expect the implementation of the New Accord to result in a significant decrease in aggregate capital requirements for the U.S. banking system. Individual banking organizations may, however, face increases or decreases in their minimum risk-based capital requirements because the New Accord is more risk sensitive than the 1988 Accord and the Agencies' existing risk-based capital rules (general risk-based capital rules). The Agencies will continue to analyze the potential impact of the New Accord on both systemic and individual bank capital levels.

C. Overview of U.S. Implementation

The Agencies believe that the advanced risk and capital measurement methodologies of the New Accord are the most appropriate approaches for large, internationally active banking organizations. As a result, large, internationally active banking organizations in the United States would be required to use the A-IRB approach to credit risk and the AMA to operational risk. The Agencies are proposing to identify three types of banking organizations: institutions subject to the advanced approaches on a mandatory basis (core banks); institutions not subject to the advanced approaches on a mandatory basis, but that choose voluntarily to apply those approaches (opt-in banks); and institutions that are not mandatorily subject to and do not apply the advanced approaches (general banks). Core banks would be those with total banking (and thrift) assets of $250 billion or more or total on-balance-sheet foreign exposure of $10 billion or more. Both core banks and opt-in banks (advanced approach banks) would be required to meet certain infrastructure requirements (including complying with specified supervisory standards for credit risk and operational risk) and make specified public disclosures before being able to use the advanced approaches for risk-based regulatory capital calculation purposes.

3

3

The Agencies continue to reserve the right to require higher minimum capital levels for individual institutions, on a case-by-case basis, if necessary to address particular circumstances.

General banks would continue to apply the general risk-based capital rules. Because the general risk-based capital rules include a buffer for risks not easily quantified (for example, operational risk and concentration risk), general banks would not be subject to an additional direct capital charge for operational risk.

Under this proposal, some U.S. banking organizations would use the advanced approaches while others would apply the general risk-based capital rules. As a result, the United States would have a bifurcated regulatory capital framework. That is, U.S. capital rules would provide two distinct methodologies for institutions to calculate risk-weighted assets (the denominator of the risk-based capital ratios). Under the proposed framework, all U.S. institutions would continue to calculate regulatory capital, the numerator of the risk-based capital ratios, as they do now. Importantly, U.S. banking organizations would continue to be subject to a leverage ratio requirement under existing regulations, and Prompt Corrective Action (PCA) legislation and implementing regulations would remain in effect.

4

It is recognized that in some cases, under the proposed framework, the leverage ratio would serve as the most binding regulatory capital constraint.

4

Thus, for example, to be in the well-capitalized PCA category a bank must have at least a 10 percent total risk-based capital ratio, a 6 percent Tier I risk-based capital ratio, and a 5 percent leverage ratio. The other PCA categories also would not change.

Implementing the capital framework described in this ANPR would raise a number of significant practical and conceptual issues about the role of economic capital calculations relative to regulatory capital requirements. The capital formulas described in this ANPR, as well as the economic capital models used by banking organizations, assume the ability to assign precisely probabilities to future credit and operational losses that might occur. The term “economic capital” is often used to refer to the amount of capital that should be allocated to an activity according to the results of such an exercise. For example, a banking organization might compute the amount of income, reserves, and capital that it would need to cover the 99.9th percentile of possible credit losses associated with a given type of lending. The desired degree of certainty of covering losses is related to several factors including, for example, the banking organization's target credit rating. The higher the loss percentile the institution wishes to provide protection against, the less likely the capital held by the institution would be insufficient to cover losses, and the higher would be the institution's credit rating.

While the Agencies intend to move to a framework where regulatory capital is more closely aligned to economic capital, the Agencies do not intend to place sole reliance on the results of economic capital calculations for purposes of computing minimum regulatory capital requirements. Banking organizations face risks other than credit and operational risks, and the assumed loss distributions underlying banking organizations' economic capital calculations are subject to the risk of error. Consequently, the Agencies continue to view the leverage ratio tripwires contained in existing PCA and other regulations as important components of the regulatory capital framework.

The A-IRB Approach for Credit Risk

Under the A-IRB approach for credit risk, an institution's internal assessment of key risk drivers for a particular exposure (or pool of exposures) would serve as the primary inputs in the calculation of the institution's minimum risk-based capital requirements. Formulas, or risk weight functions, specified by supervisors would use the banking organization's estimated inputs to derive a specific dollar amount capital requirement for each exposure (or pool of exposures). This dollar capital requirement would be converted into a risk-weighted assets equivalent by multiplying the dollar amount of the capital requirement by 12.5—the reciprocal of the 8 percent minimum risk-based capital requirement. Generally, banking organizations using the A-IRB approach would assign assets and off-balance-sheet exposures into one of three portfolios: wholesale (corporate, interbank, and sovereign), retail (residential mortgage, qualifying revolving, and other), and equities. There also would be specific treatments for securitization exposures and purchased receivables. Certain assets that do not constitute a direct credit exposure (for example, premises, equipment, or mortgage servicing rights) would continue to be subject to the general risk-based capital rules and risk weighted at 100 percent. A brief overview of each A-IRB portfolio follows.

Wholesale (Corporate, Interbank, and Sovereign) Exposures

Wholesale credit exposures comprise three types of exposures: corporate, interbank, and sovereign. Generally, the meaning of interbank and sovereign would be consistent with the general risk-based capital rules. Corporate exposures are exposures to private-sector companies; interbank exposures are primarily exposures to banks and securities firms; and sovereign exposures are those to central governments, central banks, and certain

other public-sector entities (PSEs). Within the wholesale exposure category, in addition to the treatment for general corporate lending, there would be four sub-categories of specialized lending (SL). These are project finance (PF), object finance (OF), commodities finance (CF), and commercial real estate (CRE). CRE is further subdivided into low-asset-correlation CRE, and high-volatility CRE (HVCRE).

For each wholesale exposure, an institution would assign four quantitative risk drivers (inputs): (1) Probability of default (PD), which measures the likelihood that the borrower will default over a given time horizon; (2) loss given default (LGD), which measures the proportion of the exposure that will be lost if a default occurs; (3) exposure at default (EAD), which is the estimated amount owed to the institution at the time of default; and (4) maturity (M), which measures the remaining economic maturity of the exposure. Institutions generally would be able to take into account credit risk mitigation techniques (CRM), such as collateral and guarantees (subject to certain criteria), by adjusting their estimates for PD or LGD. The wholesale A-IRB risk weight function would use all four risk inputs to produce a specific capital requirement for each wholesale exposure. There would be a separate, more conservative risk weight function for certain acquisition, development, and construction loans (ADC) in the HVCRE category.

Retail Exposures

Within the retail category, three distinct risk weight functions are proposed for three product areas that exhibit different historical loss experiences and different asset correlations.

5

The three retail sub-categories would be: (1) Exposures secured by residential mortgages and related exposures; (2) qualifying revolving exposures (QRE); and (3) other retail exposures. QRE would include unsecured revolving credits (such as credit cards and overdraft lines), and other retail would include most other types of exposures to individuals, as well as certain exposures to small businesses. The key inputs to the three retail risk weight functions would be a banking organization's estimates of PD, LGD, and EAD. There would be no explicit M component to the retail A-IRB risk weight functions. Unlike wholesale exposures, for retail exposures, an institution would assign a common set of inputs (PD, LGD, and EAD) to predetermined pools of exposures, which are typically referred to as segments, rather than to individual exposures.

6

The inputs would be used in the risk weight functions to produce a capital charge for the associated pool of exposures.

5

Asset correlation is a measure of the tendency for the financial condition of a borrower in a banking organization's portfolio to improve or degrade at the same time as the financial condition of other borrowers in the portfolio improve or degrade.

6

When the PD, LGD, and EAD parameters are assigned separately to individual exposures, it may be referred to as a “bottom-up” approach. When those parameters are assigned to predetermined sets of exposures (pools or segments), it may be referred to as a “top-down” approach.

Equity Exposures

Banking organizations would use a market-based internal model for determining capital requirements for equity exposures in the banking book. The internal model approach would assess capital based on an estimate of loss under extreme market conditions. Some equity exposures, such as holdings in entities whose debt obligations qualify for a zero percent risk weight, would continue to receive a zero percent risk weight under the A-IRB approach to equities. Certain other equity exposures, such as those made through a small business investment company (SBIC) under the Small Business Investment Act or a community development corporation (CDC) or a community and economic development entity (CEDE), generally would be risk weighted at 100 percent under the A-IRB approach to equities. Banking organizations that are subject to the Agencies' market risk capital rules would continue to apply those rules to assess capital against equity positions held in the trading book.

7

Banking organizations that are not subject to the market risk capital rules would treat equity positions in the trading account as if they were in the banking book.

7

The market risk capital rules were implemented by the banking agencies in 1996. The market risk capital rules apply to any banking organization whose trading activity (on a consolidated worldwide basis) equals 10 percent or more of total assets, or $1 billion or more. The market risk capital rules are found at 12 CFR Part 3, Appendix B (OCC), 12 CFR Parts 208 and 225, Appendix E (Board), and 12 CFR Part 325, Appendix C (FDIC). The OTS, to date, has not adopted the market risk capital rules.

Securitization Exposures

Under the A-IRB treatment for securitization exposures, a banking organization that originates a securitization would first calculate the A-IRB capital charge that would have been assessed against the underlying exposures as if the exposures had not been securitized. This capital charge divided by the size of the exposure pool is referred to as KIRB. If an originating banking organization retains a position in a securitization that obligates the banking organization to absorb losses up to or less than KIRB, the banking organization would deduct the retained position from capital as is currently required under the general risk-based capital rules. The general risk-based capital rules, however, require a dollar-for-dollar risk-based capital deduction for certain residual interests retained by originating banking organizations in asset securitization transactions regardless of amount. The A-IRB framework would no longer require automatic deduction of such residual interests. The amount to be deducted would be capped at KIRB for most exposures. For a position in excess of the KIRB threshold, the originating banking organization would use an external-ratings-based approach (if the position has been rated by an external rating agency or a rating can be inferred) or a supervisory formula to determine the capital charge for the position.

Non-originating banking organizations that invest in a securitization exposure generally would use an external-ratings-based approach (if the exposure has been rated by an external rating agency or a rating can be inferred). For unrated liquidity facilities that banking organizations provide to securitizations, capital requirements would be based on several factors, including the asset quality of the underlying pool and the degree to which other credit enhancements are available. These factors would be used as inputs to a supervisory formula. Under the A-IRB approach to securitization exposures, banking organizations also would be required in some cases to hold regulatory capital against securitizations of revolving exposures that have early amortization features.

Purchased Receivables

Purchased receivables, that is, those that are purchased from another institution either through a one-off transaction or as part of an ongoing program, would be subject to a two-part capital charge: one part is for the credit risk arising from the underlying receivables and the second part is for dilution risk. Dilution risk refers to the possibility that contractual amounts payable by the underlying obligors on the receivables may be reduced through future cash payments or other credits to the accounts made by the seller of the receivables. The framework for determining the capital charge for credit risk permits a purchasing organization to use a top-down (pool) approach to estimating PDs and LGDs when the

purchasing organization is unable to assign an internal risk rating to each of the purchased accounts. The capital charge for dilution risk would be calculated using the wholesale risk weight function with some additional specified risk inputs.

The AMA for Operational Risk

Under the A-IRB approach, capital charges for credit risk would be directly calibrated solely for such risk and, thus, unlike the 1988 Accord, would not implicitly include a charge for operational risk. As a result, the Agencies are proposing that banking organizations operating under the A-IRB approach also would have to hold regulatory capital for exposure to operational risk. The Agencies are proposing to define operational risk as the risk of losses resulting from inadequate or failed internal processes, people, and systems, or external events. Under the AMA, each banking organization would be able to use its own methodology for assessing exposure to operational risk, provided the methodology is comprehensive and results in a capital charge that is reflective of the operational risk experience of the organization. The operational risk exposure would be multiplied by 12.5 to determine a risk-weighted assets equivalent, which would be added to the comparable amounts for credit and market risk in the denominator of the risk-based capital ratios. The Agencies will be working closely with institutions over the next few years as operational risk measurement and management techniques continue to evolve.

Other Considerations

Boundary Issues

With the introduction of an explicit regulatory capital charge for operational risk, an issue arises about the proper treatment of losses that can be attributed to more than one risk factor. For example, where a loan defaults and the banking organization discovers that the collateral for the loan was not properly secured, the banking organization's resulting losses would be attributable to both credit and operational risk. The Agencies recognize that these types of boundary issues are important and have significant implications for how banking organizations would compile loss data sets and compute regulatory capital charges.

The Agencies are proposing the following standard to govern the boundary between credit and operational risk: A loss event that has characteristics of credit risk would be incorporated into the credit risk calculations for regulatory capital (and would not be incorporated into operational risk capital calculations). This would include credit-related fraud losses. Thus, in the above example, the loss from the loan would be attributed to credit risk (not operational risk) for regulatory capital purposes. This separation between credit and operational risk is supported by current U.S. accounting standards for the treatment of credit risks.

With regard to the boundary between the trading book and the banking book, for institutions subject to the market risk rules, positions currently subject to those rules include all positions held in the trading account consistent with GAAP. The New Accord proposed additional criteria for positions includable in the trading book for purposes of market risk capital requirements. The Agencies encourage comment on these additional criteria and whether the Agencies should consider adopting such criteria (in addition to the GAAP criteria) in defining the trading book under the Agencies' market risk capital rules. The Agencies are seeking comment on the proposed treatment of the boundaries between credit, operational, and market risk.

Supervisory Considerations

The advanced approaches introduce greater complexity to the regulatory capital framework and would require a high level of sophistication in the banking organizations that implement the advanced approaches. As a result, the Agencies propose to require core and opt-in banks to meet certain infrastructure requirements and comply with specific supervisory standards for credit risk and for operational risk. In addition, banking organizations would have to satisfy a set of public disclosure requirements as a prerequisite for approval to using the advanced approaches. Supervisory guidance for each credit risk portfolio type, as well as for operational risk, is being developed to ensure a sufficient degree of consistency within the supervisory framework, while also recognizing that internal systems will differ between banking organizations. The goal is to establish a supervisory framework within which all institutions must develop their internal systems, leaving exact details to each institution. In the case of operational risk in particular, the Agencies recognize that measurement methodologies are still evolving and flexibility is needed.

It is important to note that supervisors would not look at compliance with requirements, or standards alone. Supervisors also would evaluate whether the components of an institution's advanced approaches are consistent with the overall objective of sound risk management and measurement. An institution would have to use appropriately the advanced approaches across all material business lines, portfolios, and geographic regions. Exposures in non-significant business units as well as asset classes that are immaterial in terms of size and perceived risk profile may be exempted from the advanced approaches with supervisory approval. These immaterial portfolios would be subject to the general risk-based capital rules.

Proposed supervisory guidance for corporate credit exposures and for operational risk is provided separately from this ANPR in today's

Federal Register

. The draft supervisory guidance for corporate credit exposures is entitled “Supervisory Guidance on Internal-Ratings-Based Systems for Corporate Credit.” The guidance includes specified supervisory standards that an institution's internal rating system for corporate exposures would have to satisfy for the institution to be eligible to use the A-IRB approach for credit risk. The draft operational risk guidance is entitled “Supervisory Guidance on Operational Risk Advanced Measurement Approaches for Regulatory Capital.” The operational risk guidance includes identified supervisory standards for an institution's AMA framework for operational risk. The Agencies encourage commenters to review and comment on the draft guidance pieces in conjunction with this ANPR. The Agencies intend to issue for public comment supervisory guidance on retail credit exposures, equity exposures, and securitization exposures over the next several months.

Supervisory Review

As mentioned above, the second pillar of the New Accord focuses on supervisory review to ensure that an institution holds sufficient capital given its overall risk profile. The concepts of Pillar 2 are not new to U.S. banking organizations. U.S. institutions already are required to hold capital sufficient to meet their risk profiles, and supervisors may require that an institution hold more capital if its current levels are deficient or some element of its business practices suggest the need for more capital. The Agencies also have the right to intervene when capital levels fall to an unacceptable level. Given these long-standing elements of the U.S. supervisory framework, the Agencies

are not proposing to introduce specific requirements or guidelines to implement Pillar 2. Instead, existing guidance, rules, and regulations would continue to be enforced and supplemented as necessary as part of this proposed new regulatory capital framework. However, all institutions operating under the advanced approaches would be expected by supervisors to address specific assumptions embedded in the advanced approaches (such as diversification in credit portfolios), and would be evaluated for their ability to account for deviations from the underlying assumptions in their own portfolios.

Disclosure

An integral part of the advanced approaches is enhanced public disclosure practices and improved transparency. Under the Agencies' proposal, specific disclosure requirements would be applicable to all institutions using the advanced approaches. These disclosure requirements would encompass capital, credit risk, equities, credit risk mitigation, securitization, market risk, operational risk, and interest rate risk in the banking book.

D. Competitive Considerations

It is essential that the Agencies gain a full appreciation of the possible competitive equity concerns that may be presented by the establishment of a new capital framework. The creation of a bifurcated capital framework in the United States—one set of capital standards applicable to large, internationally active banking organizations (and those that choose to apply the advanced approaches), and another set of standards applicable to all other institutions—has created concerns among some parties about the potential impact on competitive equity between the two sets of banking organizations. Similarly, differences in supervisory application of the advanced approaches (both within the United States and abroad) among large, internationally active institutions may pose competitive equity issues among such institutions.

The New Accord relies upon compliance with certain minimum operational and supervisory requirements to promote consistent interpretation and uniformity in application of the advanced approaches. Nevertheless, independent supervisory judgment will be applied on a case-by-case basis. These processes, albeit subject to detailed and explicit supervisory guidance, contain an inherent amount of subjectivity and must be assessed by supervisors on an ongoing basis. This supervisory assessment of the internal processes and controls leading to an institution's internal ratings and other estimates must maintain the high level of internal risk measurement and management processes contemplated in this ANPR.

The BSC's Accord Implementation Group (AIG), in which the Agencies play an active role, will seek to ensure that all jurisdictions uniformly apply the same high qualitative and quantitative standards to internationally active banking institutions. However, to the extent that different supervisory regimes implement these standards differently, there may be competitive dislocations. One concern is that the U.S. supervisory regime will impose greater scrutiny in its implementation standards, particularly given the extensive on-site presence of bank examiners in the United States.

Quite distinct from the need for a level playing field among internationally active institutions are the competitive concerns of those institutions that do not elect to adopt or may not qualify for the advanced approaches. Some banking organizations have expressed concerns that small or regional banks would become more likely to be acquired by larger organizations seeking to lever capital efficiencies. There also is a qualitative concern about the impact of being considered a “second tier” institution (one that does not implement the advanced approaches) by the market, rating agencies, or sophisticated customers such as government or municipal depositors and borrowers. Finally, there is the question of what, if any, competitive distortions might be introduced by differences in regulatory capital minimums between the advanced approaches and the general risk-based capital rules for loans or securities with otherwise similar risk characteristics, and the extent to which such distortions may be mitigated in an environment in which well-managed banking organizations continue to hold excess capital.

8

8

The Agencies note that under the general risk-based capital rules some institutions currently are able to hold less capital than others on some types of assets (for example, through innovative financing structures or use of credit risk mitigation techniques). In addition, some institutions may hold lower amounts of capital because the market perceives them as highly diversified, while others hold higher amounts of capital because of concentrations of credit risk or other factors.

Because the advanced framework described in this ANPR is more risk-sensitive than the 1988 Accord and the general risk-based capital rules, banking organizations under the advanced approaches would face increases in their minimum risk-based capital charges on some assets and decreases on others. The results of a Quantitative Impact Study (QIS3) the BSC conducted in late 2002 indicated the potential for the advanced approaches described in this document to produce significant changes in risk-based capital requirements for specific activities; the results also varied on an institution-by-institution basis. The results of QIS3 can be found at

http://www.bis.org

and various results of QIS3 are noted at pertinent places in this ANPR.

The Agencies do not believe the results of QIS3 are sufficiently reliable to form the basis of a competitive impact analysis, both because the inputs to the study were provided on a best-efforts basis and because the proposals in this ANPR are in some cases different than those that formed the basis of QIS3. The Agencies are nevertheless interested in views on how changes in regulatory capital (for the total of credit and operational risk) of the magnitude described in QIS3, if such changes were in fact realized, would affect the competitive landscape for domestic banking organizations.

The Agencies plan to conduct at least one more QIS, and potentially other economic impact analyses, to better understand the potential impact of the proposed framework on the capital requirements for individual U.S. banking organizations and U.S. banking organizations as a whole. This may affect the Agencies' further proposals through recalibrating the A-IRB risk weight formulas and making other modifications to the proposed approaches if the capital requirements do not seem consistent with the overall risk profiles of banking organizations or safe and sound banking practices.

If competitive effects of the New Accord are determined to be significant, the Agencies would need to consider potential ways to address those effects while continuing to seek to achieve the objectives of the current proposal. Alternatives could potentially include modifications to the proposed approaches, as well as fundamentally different approaches. The Agencies recognize that an optimal capital system must strike a balance between the objectives of simplicity and regulatory consistency across banking organizations on the one hand, and the degree of risk sensitivity of the regulation on the other. There are many criteria that must be evaluated in achieving this balance, including the resulting incentives for improving risk measurement and management practices, the ease of supervisory and regulatory enforcement, the degree to

which the overall level of regulatory capital in the banking system is broadly preserved, and the effects on domestic and international competition. The Agencies are interested in commenters' views on alternatives to the advanced approaches that could achieve this balance, and in particular on alternatives that could do so without a bifurcated approach.

9

9

In this regard, alternative approaches would take time to develop, but might present fewer implementation challenges. Additional work would be necessary to advance the goal of competitive equity among internationally active banking organizations. If consensus on alternative approaches could not be reached at the BSC, a departure from the Basel framework also could raise significant international and domestic issues.

The Agencies are committed to investigate the full scope of possible competitive impact and welcome all comments in this regard. Some questions are suggested below that may serve to focus commenters' general reactions. More specific questions also are suggested throughout this ANPR. These questions should not be viewed as limiting the Agencies' areas of interest or commenters' submissions on the proposals. The Agencies encourage commenters to provide supporting data and analysis, if available.

What are commenters' views on the relative pros and cons of a bifurcated regulatory capital framework versus a single regulatory capital framework? Would a bifurcated approach lead to an increase in industry consolidation? Why or why not? What are the competitive implications for community and mid-size regional banks? Would institutions outside of the core group be compelled for competitive reasons to opt-in to the advanced approaches? Under what circumstances might this occur and what are the implications? What are the competitive implications of continuing to operate under a regulatory capital framework that is not risk sensitive?

If regulatory minimum capital requirements declined under the advanced approaches, would the dollar amount of capital held by advanced approach banking organizations also be expected to decline? To the extent that advanced approach institutions have lower capital charges on certain assets, how probable and significant are concerns that those institutions would realize competitive benefits in terms of pricing credit, enhanced returns on equity, and potentially higher risk-based capital ratios? To what extent do similar effects already exist under the current general risk-based capital rules (for example, through securitization or other techniques that lower relative capital charges on particular assets for only some institutions)? If they do exist now, what is the evidence of competitive harm?

Apart from the approaches described in this ANPR, are there other regulatory capital approaches that are capable of ameliorating competitive concerns while at the same time achieving the goal of better matching regulatory capital to economic risks? Are there specific modifications to the proposed approaches or to the general risk-based capital rules that the Agencies should consider?

II. Application of the Advanced Approaches in the United States

By its terms, the 1988 Accord applied only to internationally active banks. Under the New Accord, the scope of application has been broadened also to encompass bank holding companies that are parents of internationally active “banking groups.”

A. Threshold Criteria for Mandatory Advanced Approach Organizations

The Agencies believe that for large, internationally active U.S. institutions only the advanced approaches are appropriate. Accordingly, the Agencies intend to identify three groups of banking organizations: (1) Large, internationally active banking organizations that would be subject to the A-IRB approach and AMA on a mandatory basis (core banks); (2) organizations not subject to the advanced approaches on a mandatory basis, but that voluntarily choose to adopt those approaches (opt-in banks); and all remaining organizations that are not mandatorily subject to and do not apply the advanced approaches (general banks).

For purposes of identifying core banks, the Agencies are proposing a set of objective criteria for industry consideration. Specifically, the Agencies are proposing to treat as a core bank any banking organization that has (1) total commercial bank (and thrift) assets of $250 billion or more, as reported on year-end regulatory reports (with banking assets of consolidated groups aggregated at the U.S. bank holding company level);

10

or (2) total on-balance-sheet foreign exposure of $10 billion or more, as reported on the year-end Country Exposure Report (FFIEC 009) (with foreign exposure of consolidated groups aggregated at the U.S. bank holding company level). These threshold criteria are independent; meeting either condition would mean an institution is a core bank.

10

For banks this means the December Consolidated Report of Condition and Income (Call Report). For thrifts this means the December Thrift Financial Report.

Once an institution becomes a core bank it would remain subject to the advanced approaches on a going forward basis. If, in subsequent years, such an institution were to drop below both threshold levels it would continue to be a core bank unless it could demonstrate to its primary Federal supervisor that it has substantially and permanently downsized and should no longer be a core bank. The Agencies are proposing an annual test for assessing banking organizations in reference to the threshold levels. However, as a banking organization approaches either of the threshold levels the Agencies would expect to have ongoing dialogue with that organization to ensure that appropriate practices are in place or are actively being developed to prepare the organization for implementation of the advanced approaches.

Institutions that by expansion or merger meet the threshold levels must qualify for use of the advanced approaches and would be subject to the same implementation plan requirements and minimum risk-based capital floors applicable to core and opt-in banks as described below. Institutions that seek to become opt-in banks would be expected to notify their primary Federal supervisors well in advance of the date by which they expect to qualify for the advanced approaches. Based on the aforementioned threshold levels, the Agencies anticipate at this time that approximately ten U.S. institutions would be core banks.

Application of Advanced Approaches at Individual Bank/Thrift Levels

The Agencies are aware that some institutions might, on a consolidated basis, exceed one of the threshold levels for mandatory application of the A-IRB approach and AMA and, yet, might be comprised of distinct bank and thrift charters whose respective sizes fall well below the thresholds. In those cases, the Agencies believe that all bank and thrift institutions that are members of a consolidated group that is itself a core bank or an opt-in bank should calculate and report their risk-based capital requirements under the advanced approaches. However, recognizing that separate bank and thrift charters may, to a large extent, be independently managed and have different systems and portfolios, the Agencies are interested in comment on the efficacy and burden of a framework that requires the advanced approaches to be implemented by (or pushed down to) each of the separate subsidiary banks and thrifts that make up the consolidated group.

U.S. Banking Subsidiaries of Foreign Banking Organizations

Any U.S. bank or thrift that is a subsidiary of a foreign bank would have to comply with the prevailing U.S. regulatory capital requirements applied to U.S. banks. Thus, if a U.S. bank or

thrift that is owned by a foreign bank meets the threshold levels for mandatory application of the advanced approaches, the U.S. bank or thrift would be a core bank. If it does not meet those thresholds, it would have the choice to opt-in to the advanced approaches (and be subject to the same supervisory framework as other U.S. banking organizations) or to remain a general bank. A top-tier U.S. bank holding company that is owned by a foreign bank also would be subject to the same threshold levels for core bank determination and would be subject to the applicable U.S. bank holding company capital rules. However, Federal Reserve SR Letter 01-1 (January 5, 2001) would remain in effect. Thus, subject to the conditions in SR Letter 01-1, a top-tier U.S. bank holding company that is owned or controlled by a foreign bank that is a qualifying financial holding company generally would not be required to comply with the Board's capital adequacy guidelines.

The Agencies are interested in comment on the extent to which alternative approaches to regulatory capital that are implemented across national boundaries might create burdensome implementation costs for the U.S. subsidiaries of foreign banks.

B. Implementation for Advanced Approach Organizations

As noted earlier, U.S. banking organizations that apply the advanced approaches would be required to comply with supervisory standards prior to use.

The BSC has targeted December 31, 2006 as the effective date for the international capital rules based on the New Accord. The Agencies are proposing an implementation date of January 1, 2007. The establishment of a final effective date in the United States, however, would be contingent on the issuance for public comment of a Notice of Proposed Rulemaking, and subsequent finalization of any changes in capital regulations that the Agencies ultimately decide to adopt.

Because of the need to pre-qualify for the advanced approaches, banking organizations would need to take a number of steps upon the finalization of any changes to the capital regulations. These steps would include developing detailed written implementation plans for the A-IRB approach and the AMA and keeping their primary supervisors advised of these implementation plans and schedules. Implementation plans would need to address all supervisory standards for the A-IRB approach and the AMA, include objectively measurable milestones, and demonstrate that adequate resources would be realistically budgeted and made available. An institution's board of directors would need to approve its implementation plans.

The Agencies expect core banks to make every effort to meet the supervisory standards as soon as practicable. In this regard, it is possible that some core banks would not qualify to use the advanced approaches in time to meet the effective date that is ultimately established. For those banking organizations, the implementation plan would need to identify when the supervisory standards would be met and when the institution would be ready for implementation. The Agencies note that developing an appropriate infrastructure to support the advanced approaches for regulatory capital that fully complies with supervisory conditions and expectations and the associated supervisory guidance will be challenging. The Agencies believe, however, that institutions would need to be fully prepared before moving to the advanced approaches.

Use of the advanced approaches would require the primary Federal supervisor's approval. Core banks unable to qualify for the advanced approaches in time to meet the effective date would remain subject to the general risk-based capital rules existing at that time. The Agencies would consider the effort and progress made to meet the qualifying standards and would consider whether, under the circumstances, supervisory action should be taken against or penalties imposed on individual core banks that have not adhered to the schedule outlined in the implementation plan they submitted to their primary Federal supervisor.

Opt-in banks meeting the supervisory standards could seek to qualify for the advanced approaches in time to meet the ultimate final effective date or any time thereafter. Institutions contemplating opting-in to the advanced approaches would need to provide notice to, and submit an implementation plan and schedule to be approved by, their primary Federal supervisor. As is true of core banks, opt-in banks would need to allow ample time for developing and executing implementation plans.

An institution's primary Federal supervisor would have responsibility for determining the institution's readiness for an advanced approach and would be ultimately responsible, after consultation with other relevant supervisors, for determining whether the institution satisfies the supervisory expectations for the advanced approaches. The Agencies recognize that a consistent and transparent process to oversee implementation of the advanced approaches would be crucial. The Agencies intend to develop interagency validation standards and procedures to help ensure consistency. The Agencies would consult with each other on significant issues raised during the validation process and ongoing implementation.

C. Other Considerations

General Banks

The Agencies expect that the vast majority of U.S. institutions would be neither core banks nor opt-in banks. Most institutions would remain subject to the general risk-based capital rules. However, as has been the case since the 1988 Accord was initially implemented in the United States, the Agencies will continue to make necessary modifications to the general risk-based capital rules as appropriate. In the event changes are warranted, the Agencies could implement revisions through notice and comment procedures prior to the proposed effective date of the advanced approaches in 2007.

The Agencies seek comment on whether changes should be made to the existing general risk-based capital rules to enhance their risk-sensitivity or to reflect changes in the business lines or activities of banking organizations without imposing undue regulatory burden or complication. In particular, the Agencies seek comment on whether any changes to the general risk-based capital rules are necessary or warranted to address any competitive equity concerns associated with the bifurcated framework.

Majority-Owned or Controlled Subsidiaries

The New Accord generally applies to internationally active banking organizations on a fully consolidated basis. Thus, consistent with the Agencies' general risk-based capital rules, subsidiaries that are consolidated under U.S. generally accepted accounting principles (GAAP) typically should be consolidated for regulatory capital calculation purposes under the advanced approaches as well.

11

With regard to investments in consolidated insurance underwriting subsidiaries, the New Accord notes that deconsolidation of assets and deduction of capital is an

appropriate approach. The Federal Reserve is actively considering several approaches to the capital treatment for investments by bank holding companies in insurance underwriting subsidiaries. For example, the Federal Reserve is currently assessing the merits and weaknesses of an approach that would consolidate an insurance underwriting subsidiary's assets at the holding company level and permit excess capital of the subsidiary to be included in the consolidated regulatory capital of the holding company. A deduction would be required for capital that is not readily available at the holding company level for general use throughout the organization.

11

One notable exception exists at the bank level where there is an investment in a financial subsidiary as defined in the Gramm-Leach-Bliley Act of 1999. For such a subsidiary, assets would continue to be deconsolidated from the bank's on-balance-sheet assets, and capital at the subsidiary level would be deducted from the bank's capital.

The Federal Reserve specifically seeks comment on the appropriate regulatory capital treatment for investments by bank holding companies in insurance underwriting subsidiaries as well as other nonbank subsidiaries that are subject to minimum regulatory capital requirements.

Transitional Arrangements

Core and opt-in banks would be required to calculate their capital ratios using the A-IRB and AMA methodologies, as well as the general risk-based capital rules, for one year prior to using the advanced approaches on a stand-alone basis. In order to begin this parallel-run year, however, the institution would have to demonstrate to its supervisor that it meets the supervisory standards. Therefore, banking organizations planning to meet the January 1, 2007 target effective date for implementation of the advanced approaches would have to receive approval from their primary Federal supervisor before year-end 2005. Banking organizations that later adopt the advanced approaches also would have a one-year dual calculation period prior to moving to stand-alone usage of the advanced approaches.

An institution would be subject to a minimum risk-based capital floor for two years following moving to the advanced approaches on a stand-alone basis. Specifically, in the first year of stand-alone usage of the advanced approaches, an institution's calculated risk-weighted assets could not be less than 90 percent of risk-weighted assets calculated under the general risk-based capital rules. In the following year, an institution's minimum calculated risk-weighted assets could not be less than 80 percent of risk-weighted assets calculated under the general risk-based capital rules.

12

12

The agencies note that the text above differs from the floor text in the New Accord, which is based on 90 and 80 percent of the minimum capital requirements under the 1988 Accord, rather than on risk-weighted assets. The Agencies expect that the final language of the New Accord would need to be consistent with this approach. The following example reflects how the floor in the first year would be applied by a U.S. banking organizaiton. If the banking organization's general risk-based capital calculation produced risk-weighted assets of $100 billion in its first year of implementation of the advanced approaches, then its risk weighted assets in that year could not be less than $90 billion. If the advanced approach calculation produced risk-weighted assets of $75 billion (a decrease of one quarter compared to the general risk-based capital rules), the organization would not calculate risk-based capital ratios on the basis of that $75 billion; rather, its risk-weighted assets would be $90 billion. Consequently, its minimum total risk-based capital charge would be $7.2 billion, and it would need $9 billion to satisfy PCA well-capitalized criteria.

As a consequence, advanced approach banking organizations would need to conduct two sets of capital calculations for at least three years. The pre-implementation calculation of A-IRB and AMA capital would not need to be made public, but the banking organization would be required to disclose risk-based capital ratios calculated under both advanced and general risk-based approaches during the two-year post-implementation period. The Agencies would not propose to eliminate the floors after the two-year transition period for any institution applying the advanced approaches until the Agencies are fully satisfied that the institution's systems are sound and accurately assess risk and that resulting capital levels are prudent.

These transitional arrangements and the floors established above relate only to risk-based capital ratios and do not affect the continued applicability to all advanced banking organizations of the leverage ratio and associated PCA regulations for banks and thrifts. Importantly, the minimum capital requirements and the PCA thresholds would not be changed. Furthermore, during the implementation period and before removal of the floors the Agencies intend to closely monitor the effect that the advanced approaches would have on capital levels at individual institutions and industry-wide capital levels. Once the results of this monitoring process are assessed, the Agencies may consider modifications to the advanced approaches to ensure that capital levels remain prudent.

Given the general principle that the advanced approaches are expected to be implemented at the same time across all material portfolios, business lines, and geographic regions, to what degree should the Agencies be concerned that, for example, data may not be available for key portfolios, business lines, or regions? Is there a need for further transitional arrangements? Please be specific, including suggested durations for such transitions.

Do the projected dates provide an adequate timeframe for core banks to be ready to implement the advanced approaches? What other options should the Agencies consider?

The Agencies seek comment on appropriate thresholds for determining whether a portfolio, business line, or geographic exposure would be material. Considerations should include relative asset size, percentages of capital, and associated levels of risk for a given portfolio, business line, or geographic region.

III. Advanced Internal Ratings-Based (A-IRB) Approach

This section describes the proposed A-IRB framework for the measurement of capital requirements for credit risk. Under this framework, banking organizations that meet the A-IRB infrastructure requirements and supervisory standards would incorporate internal estimates of risk inputs into supervisor-provided capital formulas for the various debt and equity portfolios to calculate the capital requirements for each portfolio. The discussion below provides background on the conceptual basis of the A-IRB approach and then describes the specific details of the capital formulas for two of the main exposure categories, wholesale and retail. Separate sections follow that describe the A-IRB treatments of loan loss reserves and partial charge-offs, the A-IRB treatment of purchased receivables, the A-IRB treatment of equity exposures, and the A-IRB treatment of securitization exposures. The A-IRB supervisory requirements and the A-IRB approach to credit risk mitigation techniques also are discussed in separate sections.

A. Conceptual Overview

The A-IRB framework has as its conceptual foundation the belief that any range of possible losses on a portfolio of credit exposures can be represented by a probability density function (PDF) of possible losses over a one-year time horizon. If known, the parameters of a PDF can be used to specify a particular level of capital that will lower the probability of the institution's insolvency due to adverse credit risk outcomes to a stated confidence level. With a known or estimated PDF, the probability of insolvency can be measured or estimated directly, based on the level of reserves and capital available to an institution.

The A-IRB framework builds off this concept and reflects an effort to develop a common set of risk-sensitive formulas for the calculation of required capital for credit risk. To a large extent, this framework resembles more systematic quantitative approaches to the

measurement of credit risk that many banking organizations have been developing. These approaches being developed by banking organizations generally rely on a statistical or probability-based assessment of credit risk and use inputs broadly similar to those required under the A-IRB approach. Like the value-at-risk (VaR) model that forms the basis for the market risk capital rules, the output of these statistical approaches to credit risk is typically an estimate of loss threshold on a credit exposure or pool of credit exposures that is highly unlikely to be exceeded by actual credit-related losses on the exposure or pool.

Many banking organizations now use such a credit VaR amount as the basis for an internal assessment of the economic capital necessary to cover credit risk. In this context, it is common for banking organizations' internal credit risk models to consider a one-year loss horizon, and to focus on a high loss threshold confidence level (that is, a loss threshold that has a small probability of being exceeded), such as the 99.95th percentile. This is because banking organizations typically seek to hold an amount of economic capital for credit risk whose probability of being exceeded is broadly consistent with the institution's external credit rating and its associated default probability. For example, the one-year historical probability of default for AA-rated firms is less than 5 basis points (0.05 percent).

There is a great deal of variation across banking organizations in the specifics of their credit risk measurement approaches. It is important to recognize that the A-IRB approach is not intended to allow banking organizations to use all aspects of their own models to estimate regulatory capital for credit risk. The A-IRB approach has been developed as a single, common methodology that all advanced approach banking organizations would use, and consists of a set of formulas (or functions) and a single set of assumptions regarding critical parameters for the formulas. The A-IRB approach draws on the same conceptual underpinnings as the credit VaR approaches that banking organizations have developed individually, but likely differs in many specifics from the approach used by any individual institution.

The specific A-IRB formulas require the banking organization first to estimate certain risk inputs, which the organization may do using a variety of techniques. The formulas themselves, into which the estimated risk inputs are inserted, are broadly consistent with the most common statistical approaches for measuring credit risk, but also are more straightforward to calculate than those typically employed by banking organizations (which often require computer simulations). In particular, an important property of the A-IRB formulas is portfolio invariance. That is, the A-IRB capital requirement for a particular exposure generally does not depend on the other exposures held by the banking organization; as with the general risk-based capital rules, the total credit risk capital requirement for a banking organization is simply the sum of the credit risk capital requirements on individual exposures or pools of exposures.

13

13

The theoretical underpinnings for obtaining portfolio-invariant capital charges within credit VaR models are provided in the paper “A Risk-Factor Model Foundation for Ratings-Based Bank Capital Rules,” by Michael Gordy, forthcoming in the

Journal of Financial Intermediation.

The A-IRB formulas are derived as an application of these results to a single-factor CreditMetrics-style mode. For mathematical details of this model, see M. Gordy, “A comparative Anatomy of Credit Risk Models.”

Journal of Banking and Finance,

January 2000, or H.R. Koyluogu and A. Hickman, “Reconcilable Differences.”

Risk,

October 1998.

As with the existing credit VaR models, the output of the A-IRB formulas is an estimate of the amount of credit losses over a one-year period that would only be exceeded a small percentage of the time. In the case of the A-IRB formulas, this nominal confidence level is set to 99.9 percent. This means that within the context of the A-IRB modeling assumptions a banking organization's overall credit portfolio capital requirement can be thought of as an estimate of the 99.9th percentile of potential losses on that portfolio over a one-year period. In practice, however, this 99.9 percent nominal target likely overstates the actual level of confidence because the A-IRB framework does not explicitly address portfolio concentration issues or the possibility of errors in estimating PDs, LGDs, or EADs. The choice of the 99.9th percentile reflects a desire on the part of the Agencies to align the regulatory capital standard with the default probabilities typically associated with maintaining low investment grade ratings (that is, BBB) even in periods of economic adversity and to ensure neither a substantial increase or decrease in overall required capital levels among A-IRB banking organizations compared with the capital levels that would be required under the general risk-based capital rules. It also recognizes that the risk-based capital rules count a broader range of instruments as eligible capital (for example, certain subordinated debt) than do internal economic capital methodologies.

Expected Losses Versus Unexpected Losses

The diagram below shows a hypothetical loss distribution for a portfolio of credit exposures over a one-year horizon. The loss distribution is represented by the curve, and is drawn in such a way that it depicts a higher proportion of losses falling below the mean value than falling above the mean. The average value of credit losses is referred to as expected loss (EL). The losses that exceed the expected level are labeled unexpected loss (UL). An overarching policy question concerns whether the proposed design of the A-IRB capital requirements should reflect an expectation that institutions would allocate capital to cover both EL and a substantial portion of the range of possible UL outcomes, or only the UL portion of the range of possible losses (that is, from the EL point out to the 99.9th percentile).

The Agencies recognize that some institutions, in their comment letters on earlier BSC proposals and in discussion with supervisory staffs, have highlighted the view that regulatory capital should not be allocated for EL. They emphasize that EL is normally incorporated into the interest rate and spreads charged on specific products, such that EL is covered by net interest margin and provisioning. The implication is that supervisors would review provisioning policies and the adequacy of reserves as part of a supervisory review, much as they do today, and would require additional reserves and/or regulatory capital for EL in cases where reserves were deemed insufficient. However, the Agencies are concerned that the accounting definition of general reserves differs significantly across countries, and that banking practices with respect to the recognition of impairment also are very different. Thus, the Agencies are proposing to include EL in the calibration of the risk weight functions.

The Agencies also note that the current regulatory definition of capital includes a portion of general reserves. That is, general reserves up to 1.25 percent of risk-weighted assets are included in the Tier 2 portion of total capital. If the risk weight functions were calibrated solely to UL, it could be argued that the definition of capital would also need to be revisited. In the United States, such a discussion would require a review of the provisioning practices of institutions under GAAP and of the distinctions drawn between specific and general provisions.

EP04AU03.000

The framework described in this ANPR calibrates the risk-based capital requirements to the sum of EL plus UL, which raises significant calibration issues. Those calibration issues would be treated differently if the calibration were based only on the estimate of UL. That is, decisions with respect to significant policy variables that are described below hinge crucially on the initial decision to base the calibration on EL plus UL, rather than UL only. These issues include, for example, the appropriate mechanism for incorporating any future margin income (FMI) that is associated with particular business lines, as well as the appropriate method for incorporating general and specific reserves into the risk-based capital ratios.

A final overarching assumption of the A-IRB framework is the role of asset correlations. Within the A-IRB capital formulas (as in the credit VaR models of many banking organizations), asset correlation parameters provide a measure of the extent to which changes in the economic value of separate exposures are presumed to move together. A higher asset correlation between a particular asset and other assets in the same portfolio implies a greater likelihood that the asset will decline in value at the same time as the portfolio as a whole declines in value. Because this means a greater chance that the asset will be a contributor to high loss scenarios, its capital requirement under the A-IRB framework also is higher.

Specifically, the A-IRB capital formulas described in detail below are based on the assumption that correlation in defaults across borrowers is attributable to their common dependence on one or more systematic risk factors. The basis for this assumption is the observation that a banking organization's borrowers are generally susceptible to adverse changes in the global economy. These systematic factors are distinct from the borrower-specific, or idiosyncratic, risk factors that determine the probability that a specific loan will be repaid. Like other risk-factor models, the A-IRB framework assumes that these borrower-specific factors represent idiosyncratic sources of risk, and thus (unlike the systematic risk-factors) are diversified in a large lending portfolio.

The A-IRB approach allows for much improved sensitivity to many of the loan-level determinants of economic capital (such as PD and LGD), but does not explicitly address how an exposure's economic capital might vary with the degree of concentration in the overall portfolio to specific industries or regions, or even to specific borrowers. That is, it neither rewards nor penalizes differences across banking organizations in diversification or concentration across industry, geography, and names. To introduce such rewards and penalties in an appropriate manner would necessarily entail far greater operational complexity for both regulatory and financial institutions.

In contrast, the portfolio models of credit risk employed by many banking organizations are quite sensitive to all forms of diversification. That is, the economic capital charge assigned to a loan within such a model will depend on the portfolio as a whole. In order to apply a portfolio model to the calibration of A-IRB capital charges, it would be necessary to identify the assumptions needed so that a portfolio model would yield economic capital charges that do not depend on portfolio characteristics. Recent advances in the finance literature demonstrate that economic capital charges are portfolio-invariant if (and only if) two assumptions are imposed.

14

First, the portfolio must be infinitely fine-grained. Second, there must be only a single systematic risk factor.

14

See forthcoming paper by M. Gordy referenced in footnot number 12 above.

Infinite granularity, while never literally attained, is satisfied in an approximate sense by the portfolios of large, internationally active banks. Analysis of data provided by such institutions shows that taking account of single-name concentrations in such portfolios would lead to only trivial changes in the total capital requirement. The single risk-factor assumption would appear, at first glance, more troublesome. As an empirical matter, there undoubtedly are distinct cyclical factors for different industries and different geographic regions. From a substantive perspective, however, the

relevant question is whether portfolios at large financial institutions are diversified across the various sub-sectors of the economy in a reasonably similar manner. If so, then the portfolio can be modeled as if there were only a single factor, namely, the credit cycle as a whole.

The Agencies seek comment on the conceptual basis of the A-IRB approach, including all of the aspects just described. What are the advantages and disadvantages of the A-IRB approach relative to alternatives, including those that would allow greater flexibility to use internal models and those that would be more cautious in incorporating statistical techniques (such as greater use of credit ratings by external rating agencies)? The Agencies also encourage comment on the extent to which the necessary conditions of the conceptual justification for the A-IRB approach are reasonably met, and if not, what adjustments or alternative approach would be warranted.

Should the A-IRB capital regime be based on a framework that allocates capital to EL plus UL, or to UL only? Which approach would more closely align the regulatory framework to the internal capital allocation techniques currently used by large institutions? If the framework were recalibrated solely to UL, modifications to the rest of the A-IRB framework would be required. The Agencies seek commenters' views on issues that would arise as a result of such recalibration.

B. A-IRB Capital Calculations

A common characteristic of the A-IRB capital formulas is that they calculate the actual dollar value of the minimum capital requirement associated with an exposure (or, in the case of retail exposures, a pool of exposures). This capital requirement must be converted to an equivalent amount of risk-weighted assets in order to be inserted into the denominator of a banking organization's risk-based capital ratios. Because the minimum risk-based capital ratio in the United States is 8 percent, the minimum capital requirement on any asset would be equal to 8 percent of the risk-weighted asset amount associated with that asset. Therefore, in order to determine the amount of risk-weighted assets to associate with a given minimum capital requirement, it would be necessary to multiply the dollar capital requirement generated by the A-IRB formulas by the reciprocal of 8 percent, or 12.5.

The following subsections of the ANPR detail the specific features of the A-IRB capital formulas for two principal categories of credit exposure: wholesale and retail. Both of these subsections include a proposed definition of the exposure category, a description of the banking organization-estimated inputs required to complete the capital calculations, a description of the specific calculations required to determine the A-IRB capital requirement, and tables depicting a range of representative results.

Wholesale Exposures: Definitions and Inputs

The Agencies propose that a single credit exposure category—wholesale exposures—would encompass most non-retail credit exposures in the A-IRB framework. The wholesale category would include the sub-categories of corporate, sovereign, and interbank exposures as well as all types of specialized lending exposures. Wholesale exposures would include debt obligations of corporations, partnerships, limited liability companies, proprietorships, and special-purpose entities (including those created specifically to finance and/or operate physical assets). Wholesale exposures also would include debt obligations of banks and securities firms (interbank exposures), and debt obligations of central governments, central banks, and certain public-sector entities (sovereign exposures). The wholesale exposure category would not include securitization exposures, or certain small-business exposures that are eligible to be treated as retail exposures.

The Agencies propose that advanced approach banking organizations would use the same A-IRB capital formula to compute capital requirements on all wholesale exposures with two exceptions. First, wholesale exposures to small- and medium-sized enterprises (SMEs) would use a downward adjustment to the wholesale A-IRB capital formula typically based on borrower size. Second, the A-IRB capital formula for HVCRE loans (generally encompassing certain speculative ADC loans) would use a higher asset correlation assumption than other wholesale exposures.

The proposed A-IRB capital framework for wholesale exposures would require banking organizations to assign four key risk inputs for each individual wholesale exposure: (1) Probability of default (PD); (2) loss given default (LGD); (3) exposure at default (EAD); and (4) effective remaining maturity (M). In addition, to use the proposed downward adjustment for wholesale SMEs described in more detail below, banking organizations would be required to provide an additional input for borrower size (S).

Probability of Default

The first principal input to the wholesale A-IRB calculation is the measure of PD. Under the A-IRB approach, a banking organization would assign an internal rating to each of its wholesale obligors (or in other words, assign each wholesale exposure to an internal rating grade applicable to the obligor). The internal rating would have to be produced by a rating system that meets the A-IRB infrastructure requirements and supervisory standards for wholesale exposures, which are intended to ensure (among other things) that the rating system results in a meaningful differentiation of risk among exposures. For each internal rating, the banking organization must associate a specific one-year PD value. Various approaches may be used to develop estimates of PDs; however, regardless of the specific approach, banking organizations would be expected to satisfy the supervisory standards. The minimum PD that may be assigned to most wholesale exposures is 3 basis points (0.03 percent). Certain wholesale exposures are exempt from this floor, including exposures to sovereign governments, their central banks, the BIS, IMF, European Central Bank, and high quality multilateral development banks (MDBs) with strong shareholder support.

The Agencies intend to apply standards to the PD quantification process that are consistent with the broad guidance outlined in the New Accord. More detailed discussion of those points is provided in the draft supervisory guidance on IRB approaches for corporate exposures published elsewhere in today's

Federal Register

.

Loss Given Default

The second principal input to the A-IRB capital formula for wholesale exposures is LGD. Under the A-IRB approach, banking organizations would estimate an LGD for each wholesale exposure. An LGD estimate for a wholesale exposure should provide an assessment of the expected loss in the event of default of the obligor, expressed as a percentage of the institution's estimated total exposure at default. The LGD for a defaulted exposure would be estimated as the expected economic loss rate on that exposure taking into account, where appropriate, recoveries, workout costs, and the time value of money. Banking organizations would estimate LGDs as the loss severities expected to prevail when default rates are high, unless they have information indicating that recoveries on a particular

class of exposure are unlikely to be affected to an appreciable extent by cyclical factors. As with estimates of other A-IRB inputs, banking organizations would be expected to be conservative in assigning LGDs.

Although estimated LGDs should be grounded in historical recovery rates, the A-IRB approach is structured to allow banking organizations to assess the differential impact of various factors, including, for example, the presence of collateral or differences in loan terms and covenants. The Agencies expect to impose limitations on the use of guarantees and credit derivatives in a banking organization's LGD estimates. These limitations are discussed in the separate section of this ANPR on the A-IRB treatment of credit risk mitigation techniques.

Exposure at Default

The third principal input to the wholesale A-IRB capital formula is EAD. The Agencies are proposing that banking organizations would provide their own estimate of EAD for each exposure. The EAD for an exposure would be defined as the amount legally owed to the banking organization (net of any charge-offs) in the event that the borrower defaults on the exposure. For on-balance-sheet items, banking organizations would estimate EAD as no less than the current drawn amount. For off-balance-sheet items, except over-the-counter (OTC) derivative transactions, banking organizations would assign an EAD equal to an estimate of the long-run default-weighted average EAD for similar facilities and borrowers or, if EADs are highly cyclical, the EAD expected to prevail when default rates are high. The EAD associated with OTC derivative transactions would continue to be estimated using the “add-on” approach contained in the general risk-based capital rules.

15

In addition, there would be a specific EAD calculation for the recognition of collateral in the context of repo-style transactions subject to a master netting agreement, the features of which are outlined below in the section on the A-IRB treatment of credit risk mitigation techniques.

16

15

Under the add-on approach, an institution would determine its EAD for an OTC derivative contract by adding the current value of the contract (zero if the current value is negative) and an estimate of potential future exposure (PFE) on the contract. The estimated PFE would be equal to the notional amount of the derivative multiplied by a supervisor-provided add-on factor that takes into account the type of instrument and its maturity.

16

Repo-style transactions include reverse repurchase agreements and repurchase agreements and securities lending and borrowing.

Definition of Default and Loss

A banking organization would estimate inputs relative to the following definition of default and loss. A default is considered to have occurred with respect to a particular borrower when either or both of the following two events has taken place: (1) The banking organization determines that the borrower is unlikely to pay its obligations to the organization in full, without recourse to actions by the organization such as the realization of collateral; or (2) the borrower is more than 90 days past due on principal or interest on

any

material obligation to the organization. The Agencies believe that the use of the concept of “unlikely to pay” is largely consistent with the practice of U.S. banking organizations in assessing whether a loan is on non-accrual status.

Maturity

The fourth principal input to the A-IRB capital formula is effective remaining maturity (M), measured in years. If a wholesale exposure is subject to a determinable cash flow schedule, the banking organization would calculate M as the weighted-average remaining maturity of the expected cash flows, using the amounts of the cash flows as the relevant weights. The banking organization also would be able to use the nominal remaining maturity of the exposure if the weighted-average remaining maturity of the exposure cannot be calculated. For OTC derivatives and repo-style transactions subject to master netting agreements, the institution would set M equal to the weighted-average remaining maturity of the individual transactions, using the notional amounts of the individual transactions as the relevant weights.

In all cases, M would be set no greater than five years and, with few exceptions, M would be set no lower than one year. The exceptions apply to certain transactions that are not part of a banking organization's ongoing financing of a borrower. For wholesale exposures that have an original maturity of less than three months—including repo-style transactions, money market transactions, trade finance-related transactions, and exposures arising from payment and settlement processes—M may be set as low as one day. For OTC derivatives and repo-style transactions subject to a master netting agreement, M would be set at no less than five days.

As with the assignment of PD estimates, the Agencies propose to apply supervisory standards for the estimation of LGD, EAD, and M that are consistent with the broad guidance contained in the New Accord. More detailed discussion of these issues is provided in the draft supervisory guidance on IRB approaches for corporate exposures published elsewhere in today's

Federal Register.

The Agencies seek comment on the proposed definition of wholesale exposures and on the proposed inputs to the wholesale A-IRB capital formulas. What are views on the proposed definitions of default, PD, LGD, EAD, and M? Are there specific issues with the standards for the quantification of PD, LGD, EAD, or M on which the Agencies should focus?

Wholesale Exposures: Formulas

The calculation of the A-IRB capital requirement for a particular wholesale exposure would be accomplished in three steps:

(1) Calculation of the relevant asset correlation parameter, which would be a function of PD (as well as borrower size (S) for SMEs);

(2) Calculation of a preliminary capital requirement assuming a maturity of one year, which would be a function of PD, LGD, EAD, and the asset correlation parameter calculated in the first step; and

(3) Application of a maturity adjustment for differences between the actual effective remaining maturity of the exposure and the one-year maturity assumption in the second step, where the adjustment would be a function of both PD and M.

These calculations result in the A-IRB capital requirement, expressed in dollars, for a particular wholesale exposure. As noted earlier, this amount would be converted to a risk-weighted assets equivalent by multiplying the amount by 12.5, and the risk-weighted assets equivalent would be included in the denominator of the risk-based capital ratios.

Asset Correlation

The first step in the calculation of the A-IRB capital requirement for a wholesale exposure is the calculation of the asset correlation parameter, which is denoted by the letter “R” in the formulas below. This asset correlation parameter is not a fixed amount; rather, the parameter varies as an inverse function of PD. For all wholesale exposures except HVCRE exposures, the asset correlation parameter approaches an upper bound value of 24 percent for very low PD values and approaches a lower bound value of 12 percent for very high PD values. This reflects the Agencies' view that borrowers with lower credit quality (that is, higher PDs) are likely to be more idiosyncratic in the factors affecting their likelihood of default than borrowers with higher credit quality (lower PDs). Therefore, the higher PD borrowers are proportionately less influenced by systematic (sector-wide or economy-wide) factors common to all borrowers.

17

17

See Jose Lopez, “The Empirical Relationship between Average Asset Correlation, Firm Probability of Default, and Asset Size.” Federal Reserve Bank of San Francisco Working Paper 02-05 (June 2002).

An important practical impact of having asset correlation decline with increases in PD

is to reduce the speed with which capital requirements increase as PDs increase, and to increase the speed with which EL dominates the total capital charge, thereby tending to reduce procyclicality in the application of the wholesale A-IRB capital formulas. The specific formula for determining the asset correlation parameter for all wholesale exposures except HVCRE exposures is as follows:

R = 0.12 * (1−EXP(−50 * PD)) + 0.24 * [1−(1−EXP(−50 * PD))]

Where:

R denotes asset correlation;

EXP(x) denotes the natural exponential function; and

PD denotes probability of default.

Capital Requirement With Assumed One-Year Maturity Adjustment

The second step in the calculation of the A-IRB capital requirement for a particular wholesale exposure is the calculation of the capital requirement that would apply to the exposure assuming a one-year effective remaining maturity. The specific formula to calculate this one-year-maturity capital requirement is as follows:

K

1

= EAD * LGD * N[(1−R)‸−0.5 * G(PD) + (R/(1− R))‸0.5 * G(0.999)]

Where:

K

1

denotes the one-year-maturity capital requirement;

EAD denotes exposure at default;

LGD denotes loss given default;

N(x) denotes the standard normal cumulative distribution function;

R denotes asset correlation;

G(x) denotes the inverse of the standard normal cumulative distribution function; and

18

18

The N(x) and G(x) functions are widely used in statistics and are commonly available in computer spreadsheet programs. A description of these functions may be found in the Help function of most spreadsheet programs or in basic statistical textbooks.

PD denotes probability of default.

There are several important aspects of this formula. First, it rises in a straight-line fashion with increases in EAD, meaning that a doubling of the exposure amount would result in a doubling of the capital requirement. It also rises in a straight-line fashion with increases in LGD, which similarly implies that a loan with an LGD estimate twice that of an otherwise identical loan would have twice the capital requirement of the other loan. This also implies that as LGD or EAD estimates approach zero, the capital requirement would likewise approach zero. The remainder of the formula is a function of PD, asset correlation (R), which is itself a function of PD, and the target loss percentile amount of 99.9 percent discussed earlier.

Maturity Adjustment

The third stage in the calculation of the A-IRB capital requirement for a particular wholesale exposure is the application of a maturity adjustment to reflect the exposure's actual effective remaining maturity (M). The A-IRB maturity adjustment multiplies the one-year-maturity capital requirement (K

1

) by a factor that depends on both M and PD. The fact that the A-IRB maturity adjustment depends on PD reflects the Agencies' view that there is a greater proportional need for maturity adjustments for high-quality exposures (those with low PDs) because there is a greater potential for such exposures to deteriorate in credit quality than for exposures whose credit quality is lower. The specific formula for applying the maturity adjustment and generating the A-IRB capital requirement is as follows:

K = K

1

* [1 + (M−2.5) * b]/[(1−1.5 * b)], where b = (0.08451−0.05898 * LN(PD))

2

and:

K denotes the A-IRB capital requirement;

K

1

denotes the one-year-maturity capital requirement;

M denotes effective remaining maturity;

LN(x) denotes the natural logarithm; and

PD denotes probability of default.

In this formula, the value “b” effectively determines the slope of the maturity adjustment and is itself a function of PD. Note that if M is set equal to one, the maturity adjustment also equals one and K will therefore equal K

1

.

To provide a more concrete sense of the range of capital requirements under the wholesale A-IRB framework, the following table presents the A-IRB capital requirements (K) for a range of values of both PD and M. In this table LGD is assumed to equal 45 percent. For comparison purposes, the general risk-based capital rules assign a capital requirement of 8 percent for most commercial loans.

Capital Requirements

[In percentage points]

PD

Effective remaining maturity (M)

1 month

1 year

3 years

5 years

0.05 percent

0.50

0.92

1.83

2.74

0.10 percent

1.00

1.54

2.71

3.88

0.25 percent

2.17

2.89

4.44

5.99

0.50 percent

3.57

4.40

6.21

8.03

1.00 percent

5.41

6.31

8.29

10.27

2.00 percent

7.65

8.56

10.56

12.56

5.00 percent

11.91

12.80

14.75

16.69

10.00 percent

17.67

18.56

20.50

22.45

20.00 percent

26.01

26.84

28.65

30.47

The impact of the A-IRB capital formulas on minimum risk-based capital requirements for wholesale exposures would, of course, depend on the actual values of PD, LGD, EAD, and M that banking organizations would use as inputs to the wholesale formulas. Subject to the caveats noted earlier, evidence from QIS3 suggested an average reduction in credit risk capital requirements for corporate exposures of about 26 percent for twenty large U.S. banking organizations.

SME Adjustment

For loans to SMEs not eligible for retail A-IRB treatment, the proposed calculation of the A-IRB capital requirement has one additional element—a downward adjustment based on borrower size (S). This adjustment would effectively lower the A-IRB capital requirement on wholesale exposures to SMEs with annual sales (or total assets) of less than $50 million. The Agencies believe the measure of borrower size should be based on annual sales (rather than total assets), unless the banking organization can demonstrate that it would be more appropriate for the banking organization to use the total assets of the borrower as its measure of borrower size. The borrower size adjustment would be made to the asset correlation parameter (R), as shown in the following formula:

R

SME

= R−0.04 * [1−(S− 5)/45]

Where

R

SME

denotes the size-adjusted asset correlation;

R denotes asset correlation; and

S denotes borrower size (expressed in millions of dollars).

The maximum reduction in the asset correlation parameter based on this formula is 4 percent, and is achieved when borrower size is $5 million. For all borrower sizes below $5 million, borrower size would be set equal to $5 million. The adjustment shrinks to zero as borrower size approaches $50 million. The broad rationale for this adjustment is the view that the credit condition of SMEs will be influenced relatively more by idiosyncratic factors than is the case for larger firms, and, thus, SMEs would be less likely to deteriorate simultaneously with other exposures. This greater susceptibility to idiosyncratic factors would imply lower asset correlation. The evidence in favor of this view is mixed, particularly after considering that the A-IRB framework already incorporates a negative relationship between asset correlation and PD. The following table illustrates the practical effect of the SME adjustment by depicting the capital requirements (K) across a range of PDs and borrower sizes. As in the previous table, LGD is assumed to equal 45 percent. For this table, M is assumed to be equal to three years. Note that the last column is identical to the three-year maturity column in the preceding table because the SME adjustment is phased out for borrowers of $50 million or more in size.

Capital Requirements

[In percentage points]

PD

Borrower size (S)

$5 million

$20 million

$35 million

≥ $50 million

0.05 percent

1.44

1.57

1.70

1.83

0.10 percent

2.14

2.33

2.51

2.71

0.25 percent

3.54

3.83

4.13

4.44

0.50 percent

4.97

5.37

5.79

6.21

1.00 percent

6.63

7.17

7.72

8.29

2.00 percent

8.40

9.11

9.83

10.56

5.00 percent

11.70

12.73

13.74

14.75

10.00 percent

16.76

18.05

19.30

20.50

20.00 percent

24.67

26.08

27.40

28.65

Subject to the caveats mentioned above, evidence from QIS3 suggested an average reduction in credit risk-based capital requirements for corporate SME exposures of about 39 percent for twenty large U.S. banking organizations.

If the Agencies include a SME adjustment, are the $50 million threshold and the proposed approach to measurement of borrower size appropriate? What standards should be applied to the borrower size measurement (for example, frequency of measurement, use of size buckets rather than precise measurements)?

Does the proposed borrower size adjustment add a meaningful element of risk sensitivity sufficient to balance the costs associated with its computation? The Agencies are interested in comments on whether it is necessary to include an SME adjustment in the A-IRB approach. Data supporting views is encouraged.

Wholesale Exposures: Other Considerations

Specialized Lending

The specialized lending (SL) asset class encompasses exposures for which the primary source of repayment is the income generated by the specific asset(s) being financed, rather than the financial capacity of a broader commercial enterprise. The SL category encompasses four broad exposure types:

•

Project finance

(PF) exposures finance large, complex, expensive installations that produce goods or services for sale, such as power plants, chemical processing plants, mines, or transportation infrastructure, where the source of repayment is primarily the revenues generated by sale of the goods or services by the installations.

•

Object finance

(OF) exposures finance the acquisition of (typically moveable) physical assets, such as ships or aircraft, where the source of repayment is primarily the revenues generated by the assets being financed, often through rental or lease contracts with third parties.

•

Commodities finance

(CF) exposures are structured short-term financings of reserves, inventories, or receivables of exchange-traded commodities, such as crude oil, metals, or agricultural commodities, where the source of repayment is the proceeds of the sale of the commodity.

•

Commercial real estate

(CRE) exposures finance the construction or acquisition of real estate (including land as well as improvements) where the prospects for repayment and recovery depend primarily on the cash flows generated by the lease, rental, or sale of the real estate.

19

The broad CRE category is further divided into two groups: low-asset-correlation CRE and HVCRE.

20

19

CRE exposures are typically non-recourse exposures, often to special purpose vehicles, and are distinguishable from corporate exposures that are collateralized by real estate for which the prospects for repayment and recovery depend primarily on the financial performance of the broader commercial enterprise that is the obligor.

20

To describe a loan portfolio as having a relatively high asset correlation means that any defaults that occur in that portfolio are relatively likely to occur at the same time, and for this reason the portfolio is likely to exhibit greater variability in aggregate default rates. For two portfolios with the same EL, the portfolio with more highly variable aggregate default rates warrants higher capital to cover UL (“bad-tail events”) with the same level of confidence. Describing a portfolio as having a relatively high asset correlation does

not

imply that loans in that portfolio have relatively high PD, LGD, or EL. In particular, loans in high asset correlation portfolios may well have very low PDs and LGDs and therefore ELs); conversely, loans in low asset correlation portfolios may have very high PDs and LGDs (and ELs). For any two loans from a portfolio with a given asset correlation (or from two different portfolios with the same asset correlation), the loan with the lower EL should be assigned a lower risk weight. For any two loans with the same EL, the loan from the portfolio with the lower asset correlation should incur a lower capital charge, because bad-tail events are less likely to occur in that portfolio.

Most of the issues raised below for comment are described in substantially greater detail, in the context of CRE exposures, in a white paper entitled “Loss Characteristics of CRE Loan Portfolios,” released by the Federal Reserve Board on June 10, 2003. Commenters are encouraged to read the white paper in conjunction with this section.

A defining characteristic of SL exposures (including CRE) is that the risk factors influencing actual default rates are likely to influence LGDs as well. This is because both the borrower's ability to repay an exposure

and

the banking organization's recovery on an exposure in the event of default are likely to depend on the same underlying factors, such as the net cash flows of the property being financed.

This suggests a positive correlation between observed default frequencies and observed loss rates on defaulted loans, with both declining during periods of favorable economic conditions and both increasing during unfavorable economic periods. While cyclicality in LGDs may be significant for a number of lending activities, the Agencies believe that cyclicality is likely to be the norm for SL portfolios, and that a banking organization's procedures for estimating LGD inputs for SL exposures should assess and quantify this cyclicality in a comprehensive and systematic fashion.

The Agencies invite comment on ways to deal with cyclicality in LGDs. How can risk sensitivity be achieved without creating undue burden?

For core and opt-in banks that may not be able to provide sufficiently reliable estimates of PD, LGD, and M for each SL exposure, the New Accord offers a Supervisory Slotting Criteria (SSC) approach. Under this approach, rather than estimating the loan-level risk parameters, banking organizations would use slotting criteria to map their internal risk rating grades to one of five supervisory rating grades: Strong, Good, Satisfactory, Weak, and Default. In addition, supervisory risk weights would be assigned to each of these supervisory rating grades. To assist banking organizations in implementing these supervisory rating grades, for reference purposes the New Accord associates each with an explicit range of external rating grades. If the SSC approach were allowed in the United States, the Agencies would have to develop slotting criteria that would take into account factors such as market conditions; financial ratios such as debt service coverage or loan-to-value ratios; cash flow predictability; strength of sponsor or developer; and other factors likely to affect the PD and/or LGD of each loan.

The Agencies invite comment on the merits of the SSC approach in the United States. The Agencies also invite comment on the specific slotting criteria and associated risk weights that should be used by organizations to map their internal rating grades to supervisory rating grades if the SSC approach were to be adopted in the United States.

Under the A-IRB approach, a banking organization would estimate the risk inputs for each SL exposure and then calculate the A-IRB capital charge for the exposure by substituting the estimated PD, LGD, EAD, and M into one of two risk weight functions. The first risk weight function is the wholesale risk weight function and applies to all PF, OF, and CF exposures, as well as to all low-asset-correlation CRE exposures (including in-place commercial properties). The second risk weight function applies to all HVCRE exposures. It also is the same as the wholesale risk weight function, except that it incorporates a higher asset correlation parameter. The asset correlation equation for HVCRE is as follows:

R = 0.12 × (1−EXP (−50 × PD)) + 0.30 × [EXP (−50 × PD)]

Where

R denotes asset correlation;

EXP denotes the natural exponential function; and

PD denotes probability of default.

The following table presents the A-IRB capital requirement (K) for a range of values of both PD and M. In this table, LGD is assumed to equal 45 percent. This LGD is used for consistency with the similar table above for wholesale exposures and should not be construed as an indication that 45 percent is a typical LGD for HVCRE exposures.

HVCRE Capital Requirements

[In percentage points]

PD

Effective remaining maturity

1 year

3 years

5 years

0.05 percent

1.24

2.46

3.68

0.10 percent

2.05

3.61

5.16

0.25 percent

3.74

5.76

7.77

0.50 percent

5.52

7.79

10.07

1.00 percent

7.53

9.89

12.25

2.00 percent

9.55

11.79

14.02

5.00 percent

13.12

15.12

17.11

10.00 percent

18.59

20.54

22.49

20.00 percent

26.84

28.65

30.47

All ADC loans would be treated as HVCRE exposures, unless the borrower has “substantial equity” at risk or the property is pre-sold or sufficiently pre-leased. In part, this reflects some empirical evidence suggesting that most ADC loans have relatively high asset correlations. It also, however, reflects a longstanding supervisory concern that CRE lending to finance speculative construction and development is vulnerable to, and may worsen, speculative swings in CRE markets, especially when there is little borrower equity at risk. Such lending was a major factor causing the stress experienced by many banks in the early 1990s, not only in the United States but in other countries as well.

Under the New Accord, SL loans financing the construction of one- to four-family residential properties (single or in subdivisions) are included with other ADC loans in the high asset correlation category. However, loans financing the construction of pre-sold one- to four-family residential properties would be eligible to be treated as low-asset-correlation CRE exposures. In some cases the loans may finance the construction of subdivisions or other groups of houses, some of which are pre-sold while others are not.

Under the New Accord, each national supervisory authority is directed to recognize and incorporate into its implementation of the New Accord the high asset correlation determinations of other national supervisory authorities for loans made in their respective jurisdictions. Thus, when the Agencies designate certain CRE properties as HVCRE, foreign banking organizations making extensions of credit to those properties also would be expected to treat them as HVCRE. Similarly, when non-U.S. supervisory authorities designate certain CRE as HVCRE, U.S. banking organizations that extend credit to those properties would be expected to treat them as HVCRE.

The Agencies invite the submission of empirical evidence regarding the (relative or absolute) asset correlations characterizing portfolios of ADC loans, as well as comments regarding the circumstances under which such loans would appropriately be categorized as HVCRE.

The Agencies also invite comment on the appropriateness of exempting from the high-asset-correlation category ADC loans with substantial equity or that are pre-sold or sufficiently pre-leased. The Agencies invite comment on what standard should be used in determining whether a property is sufficiently pre-leased when prevailing occupancy rates are unusually low.

The Agencies invite comment on whether high-asset-correlation treatment for one- to four-family residential construction loans is appropriate, or whether they should be included in the low-asset-correlation category. In cases where loans finance the construction of a subdivision or other group of houses, some of which are pre-sold while others are not, the Agencies invite comment regarding how the “pre-sold” exception should be interpreted.

The Agencies invite comment on the competitive impact of treating defined classes of CRE differently. What are commenters' views on an alternative approach where there is only one risk weight function for all CRE? If a single risk weight function for all CRE is considered, what would be the appropriate asset correlation to employ?

Lease Financings

Under the wholesale A-IRB framework, some lease financings require special consideration. A distinction is made for leases that expose the lessor to residual value risk, namely the risk of the fair value of the assets declining below the banking organization's estimate of residual risk at lease inception. If a banking organization has exposure to residual value risk, it would assign a 100 percent risk weight to the residual value amount and determine a risk-weighted asset equivalent for the lease's remaining net investment (net of residual value amount) using the same methodology as for any other wholesale exposure. The sum of these components would be the risk-weighted asset amount for a particular lease. Where a banking organization does not have exposure to residual value risk, the lease's net investment would be subject to a capital charge using the same methodology applied to any other wholesale exposure.

This approach would be used regardless of accounting classification as a direct finance, operating or leveraged lease. For leveraged leases, when the banking organization is the equity participant it would net the balance of the non-recourse debt against the discounted lease payment stream prior to applying the risk weight. If the banking organization is the debt participant, the exposure would be treated as any other wholesale exposure.

The Agencies are seeking comment on the wholesale A-IRB capital formulas and the resulting capital requirements. Would this approach provide a meaningful and appropriate increase in risk sensitivity in the sense that the results are consistent with alternative assessments of the credit risks associated with such exposures or the capital needed to support them? If not, where are there material inconsistencies?

Does the proposed A-IRB maturity adjustment appropriately address the risk differences between loans with differing maturities?

Retail Exposures: Definitions and Inputs

The second major exposure category in the A-IRB framework is the retail exposure category. This category encompasses the vast majority of credit exposures to individual consumers. The Agencies also are considering whether certain SME exposures should be eligible for retail A-IRB treatment. The retail exposure category has three distinct sub-categories: (1) Residential mortgages (and related exposures); (2) qualifying revolving exposures (QREs); and (3) other retail exposures. There are separate A-IRB capital formulas for each of these three sub-categories to reflect different levels of associated risk.

The Agencies propose that the residential mortgage exposure sub-category be defined to include loans secured by first or subsequent liens on one-to four-family residential properties, including term loans and revolving lines of credit secured by home equity. There would be no upper limit on the size of the exposure that could be included in the residential mortgage exposure sub-category, but the borrower would have to be an individual and the banking organization should generally manage the exposure as part of a pool of similar exposures. Residential mortgage exposures that are individually internally rated and managed similarly to commercial exposures, rather than managed and internally rated as pools, would be treated under the wholesale A-IRB framework.

The second sub-category of retail exposures is qualifying revolving exposures (QREs). The Agencies propose to define QREs as exposures to individuals that are revolving, unsecured, uncommitted, less than $100,000, and managed as part of a pool of similar exposures. In practice, QREs will include primarily exposures where customers' outstanding borrowings are permitted to fluctuate based on their own decisions to borrow and repay, up to a limit established by the banking organization. Most credit card exposures to individuals and overdraft lines on individual checking accounts would be QREs.

The third sub-category of retail exposures, other retail exposures, includes two types of exposures. First, it encompasses all exposures to individuals for non-business purposes that are generally managed as part of a pool of similar exposures and that do not meet the conditions for inclusion in the first two sub-categories of retail exposures. The Agencies are not proposing to establish a fixed upper limit on the size of exposures to individuals that are eligible for the other retail treatment. In addition, the Agencies are proposing that the other retail sub-category include certain SME exposures that are managed on a pool basis similar to retail exposures. These exposures could be to a company or to an individual. The Agencies are considering an individual borrower exposure threshold of $1 million for such exposures. For the purpose of assessing compliance with the individual borrower exposure threshold, the banking organization would aggregate all exposures to a particular borrower on a fully consolidated basis. Credit card accounts with balances between $100,000 and $1 million would be considered other retail exposures rather than QRE, even if the accounts are extended to or guaranteed by an individual and used exclusively for small business purposes.

The Agencies are interested in comment on whether the proposed $1 million threshold provides the appropriate dividing line between those SME exposures that banking organizations should be allowed to treat on a pooled basis under the retail A-IRB framework and those SME exposures that should be rated individually and treated under the wholesale A-IRB framework.

One of the most significant differences between the wholesale and retail A-IRB categories is that the risk inputs for retail exposures do not have to be assigned at the level of an individual exposure. The Agencies recognize that banking organizations typically manage retail exposures on a portfolio or pool basis, where each portfolio or pool contains exposures with similar risk characteristics. Therefore, a key characteristic of the retail A-IRB framework is that the risk inputs for retail exposures would be assigned to portfolios or pools of exposures rather than to individual exposures.

It is important to highlight that within each of the three sub-categories of retail

exposures, the retail A-IRB framework is intended to provide banking organizations with substantial flexibility to use the retail portfolio segmentation that they believe is most appropriate for their activities. In determining how to group their retail exposures within each sub-category into portfolio segments for the purpose of assigning A-IRB risk inputs, the Agencies believe that banking organizations should use a segmentation approach that is consistent with their approach for internal risk assessment purposes and that classifies exposures according to predominant risk characteristics.

As general principles for segmentation, banking organizations should group exposures in each of the three retail sub-categories into portfolios or pools according to the sub-category's principal risk drivers, and would have to be able to demonstrate that the resultant segmentation effectively differentiates and rank orders risk and provides reasonably accurate and consistent quantitative estimates of PD, LGD, and EAD. With the exceptions noted below, the Agencies are not proposing that institutions must consider any particular risk drivers or employ any minimum number of portfolios or pools in any of the three retail sub-categories. The only specific limitations that the Agencies would propose in regard to the portfolio segmentation of retail exposures are (1) banking organizations generally would not be permitted to combine retail exposures from multiple countries into the same portfolio segment (because of differences in national legal systems and bankruptcy regimes), and (2) banking organizations would need to separately segment delinquent retail exposures.

The inputs to the retail A-IRB capital formulas differ slightly from the inputs to the wholesale A-IRB capital formulas. Measures of PD, LGD, and EAD remain important elements, but there is no M input to the retail A-IRB capital formulas. Rather, the retail A-IRB capital formulas implicitly incorporate average maturity effects in general, such as in the residential mortgage sub-category.

Aside from the applicable definition of default, discussed below, the definitions of PD, LGD, and EAD for retail exposures are generally equivalent to those for wholesale exposures. One additional element of potential flexibility for banking organizations in the retail context needs to be highlighted. The Agencies recognize that certain banking organizations that may qualify for the advanced approaches segment their retail portfolios for management purposes by EL, rather than by separately measuring PD and LGD, as required under the A-IRB framework. Therefore, the Agencies propose that banking organizations be permitted substantial flexibility in translating measures of EL into the requisite PD and LGD inputs. For non-revolving portfolio segments, EL generally would equal the product of PD and LGD, so that if a banking organization has an estimate of EL and either PD or LGD, it would be able to infer an estimate of the other required input.

In addition, the Agencies are proposing that if one or the other of PD and LGD did not tend to vary significantly across portfolio segments, the banking organization would be permitted to apply a general estimate of that input to multiple segments and to use that general estimate, together with segment-specific estimates of EL, to infer segment-specific estimates of the other required input. The Agencies note, however, that this proposal offers substantial flexibility to institutions and may, in fact, be overly flexible (for example, because LGDs on residential mortgages tend to be quite cyclical). For these loans, the above method of inferring PDs or LGDs from a long-run average EL would not necessarily result in PD being estimated as a long-run average, and LGD would not necessarily reflect the loss rate expected to prevail when default rates are high. Banking organizations using an EL approach to retail portfolio segmentation would have to ensure that the A-IRB capital requirement under this method is at least as conservative as a PD/LGD method in order to minimize any potential divergences between capital requirements computed under the PD/LGD approach versus an EL approach.

As in the wholesale A-IRB framework, a floor of 3 basis points (0.03 percent) applies to the PD estimates for all retail exposures (that is, the minimum PD is 3 basis points). In addition, for residential mortgage exposures other than those guaranteed by a sovereign government, a floor of 10 percent on the LGD estimate would apply, based on the view that LGDs during periods of high default rates are unlikely to fall below this level if measured appropriately. Along with the overall monitoring of the implementation of the advanced approaches and the determination whether to generally relax the floors established during the initial implementation phases (that is, the 90 and 80 percent floors discussed above), the Agencies intend to review the need to retain PD and LGD floors for retail exposures following the first two years of implementation of the A-IRB framework.

The Agencies are proposing the following data requirements for retail A-IRB. Banking organizations would have to have a minimum of five years of data history for PD, LGD, and EAD, and preferably longer periods so as to include a complete economic cycle. Banking organizations would not have to give equal weight to all historical factors if they can demonstrate that the more recent data are better predictors of the risk inputs. Also, banking organizations would have to have a minimum of three years of experience with their portfolio segmentation and risk management systems.

Definition of Default and Loss

The retail definition of default and loss being proposed by the Agencies differs significantly from that proposed for the wholesale portfolio. Specifically, the Agencies propose to use the definitions of loss recognition embodied in the Federal Financial Institutions Examination Council (FFIEC) Uniform Retail Credit Classification and Account Management Policy.

21

All residential mortgages and all revolving credits would be charged off, or charged down to the value of the property, after a maximum of 180 days past due; other credits would be charged off after a maximum of 120 days past due.

21

The FFIEC Uniform Retail Credit Classification and Account Management Policy was issued on June 12, 2000. It is available on the FFIEC Web site at

www.FFIEC.gov.

In addition, the Agencies are proposing to define a retail default to include the occurrence of any one of the three following events if it occurs prior to the respective 120- or 180-day FFIEC policy trigger: (1) A full or partial charge-off resulting from a significant decline in credit quality of the exposure; (2) a distressed restructuring or workout involving forbearance and loan modification; or (3) a notification that the obligor has sought or been placed in bankruptcy. Finally, for retail exposures (as opposed to wholesale exposures) the definition of default may be applied to a particular facility, rather than to the obligor. That is, default on one obligation would not require a banking organization to treat all other obligations of the same obligor as defaulted.

Undrawn Lines

The treatment of undrawn lines of credit, in particular those associated with credit cards, merits specific discussion. Banking organizations would be permitted to incorporate undrawn retail lines in one of two ways. First, banking organizations could

incorporate them into their EAD estimates directly, by assessing the likelihood that undrawn balances would be drawn at the time of an event of default. Second, banking organizations could incorporate them into LGD estimates by assessing the size of potential losses in default (including those arising from both currently drawn and undrawn balances) as a proportion of the current drawn balance. In the latter case, it is possible that the relevant LGD estimates would exceed 100 percent. While the proposed EAD approach for undrawn wholesale and retail lines is the same, the Agencies are aware that the sheer volume of credit card undrawn lines and the ratio of undrawn lines to outstanding balances create issues for undrawn retail lines that differ from undrawn wholesale lines not only in degree but also in kind.

An additional issue arises in connection with the undrawn lines associated with credit card accounts whose drawn balances (but not undrawn balances) have been securitized. To the extent that banking organizations remain exposed to the risk that such undrawn lines will be drawn, but such undrawn lines are not themselves securitized, then there is a need for institutions to hold regulatory capital against such undrawn lines. The Agencies propose that a banking organization would be required to hold capital against the full amount of any undrawn lines regardless of whether drawn amounts are securitized. This presumes that the institution itself is exposed to the credit risk associated with future draws.

The Agencies are interested in comments and specific proposals concerning methods for incorporating undrawn credit card lines that are consistent with the risk characteristics and loss and default histories of this line of business.

The Agencies are interested in further information on market practices in this regard, in particular the extent to which banking organizations remain exposed to risks associated with such accounts. More broadly, the Agencies recognize that undrawn credit card lines are significant in both of the contexts discussed above, and are particularly interested in views on the appropriate retail A-IRB treatment of such exposures.

Future Margin Income

In the New Accord, the retail A-IRB treatment of QREs includes a unique additional input that arises because of the large amount of expected losses typically associated with QREs. As noted above, the A-IRB approach would require banking organizations to hold regulatory capital against both EL and UL. Banking organizations typically seek to cover expected losses through interest income and fees for all of their business activities, and the Agencies recognize that this practice is a particularly important aspect of the business model for QREs.

The Agencies are including in this proposal, for the QRE sub-category only, that future margin income (FMI) be permitted to offset a portion of the A-IRB retail capital charge relating to EL. For this purpose, the Agencies propose to define the amount of eligible FMI for the QRE sub-category as the amount of income anticipated to be generated by the relevant exposures over the next twelve months that can reasonably be assumed to be available to cover potential credit losses on the exposures after covering expected business expenses, and after subtracting a cushion to account for potential volatility in credit losses (UL). FMI would not be permitted to include anticipated income from new accounts and would have to incorporate assumptions about income from existing accounts that are in line with the banking organization's historical experience. The amount of the cushion to account for potential volatility in credit losses would be set equal to two standard deviations of the banking organization's annualized loss rate on the exposures. The Agencies would expect banking organizations to be able to support their estimates of eligible FMI on the basis of historical data and would disallow the use of FMI in the QRE capital formula if this is not the case. The step needed to recognize eligible FMI is discussed below.

Permitting a FMI offset to the A-IRB capital requirement for QREs could have a significant impact on the level of minimum regulatory capital at institutions adopting the advanced approaches. The Agencies would need to fully assess and analyze the impact of such an FMI offset on institutions' risk-based capital ratios prior to final implementation of the A-IRB approach. Furthermore, the Agencies anticipate the need to issue additional guidance setting out more specific expectations in this regard.

For the QRE sub-category of retail exposures only, the Agencies are seeking comment on whether or not to allow banking organizations to offset a portion of the A-IRB capital requirement relating to EL by demonstrating that their anticipated FMI for this sub-category is likely to more than sufficiently cover EL over the next year.

The Agencies are seeking comment on the proposed definitions of the retail A-IRB exposure category and sub-categories. Do the proposed categories provide a reasonable balance between the need for differential treatment to achieve risk-sensitivity and the desire to avoid excessive complexity in the retail A-IRB framework? What are views on the proposed approach to inclusion of SMEs in the other retail category?

The Agencies are also seeking views on the proposed approach to defining the risk inputs for the retail A-IRB framework. Is the proposed degree of flexibility in their calculation, including the application of specific floors, appropriate? What are views on the issues associated with undrawn retail lines of credit described here and on the proposed incorporation of FMI in the QRE capital determination process?

The Agencies are seeking comment on the minimum time requirements for data history and experience with portfolio segmentation and risk management systems: Are these time requirements appropriate during the transition period? Describe any reasons for not being able to meet the time requirements.

Retail Exposures: Formulas

The retail A-IRB capital formulas are very similar to the wholesale A-IRB formulas, and are based on the same underlying concepts. However, because there is no M adjustment associated with the retail A-IRB framework, the retail A-IRB capital calculations generally involve fewer steps than the wholesale A-IRB capital calculations. As with the wholesale A-IRB framework, the output of the retail A-IRB formulas is a minimum capital requirement, expressed in dollars, for the relevant pool of exposures. The capital requirement would be converted into an equivalent amount of risk-weighted assets by multiplying the capital requirement by 12.5. The two key steps in implementing the retail A-IRB capital formulas are (1) assessing the relevant asset correlation parameter, and (2) calculating the minimum capital requirement for the relevant pool of exposures.

Residential Mortgages and Related Exposures

For residential mortgage and related exposures, the retail A-IRB capital formula requires only one step. This is because the asset correlation parameter for such exposures is fixed at 15 percent, regardless of the PD of any particular pool of exposures. The fixed asset correlation parameter reflects the Agencies' view that the arguments for linking the asset correlation to PD, as occurs in the wholesale A-IRB framework and in the other two sub-categories of retail exposures, are not as relevant for residential mortgage-related exposures, whose performance is significantly influenced by broader trends in the housing market for borrowers of all credit qualities. The assumed asset correlation of 15 percent also seeks implicitly to reflect the higher average maturity associated with

residential mortgage exposures and is therefore higher than would likely be the case if a specific maturity adjustment were also included in the retail A-IRB framework. The proposed retail A-IRB capital formula for residential mortgage and related exposures is as follows:

K = EAD * LGD * N[1.08465 * G(PD) + 0.4201 * G(0.999)]

Where

K denotes the capital requirement;

EAD denotes exposure at default;

LGD denotes loss given default;

N(x) denotes the standard normal cumulative distribution function;

G(x) denotes the inverse of the standard normal cumulative distribution function; and

PD denotes probability of default.

The following table depicts a range of representative capital requirements (K) for residential mortgage and related exposures based on this formula. Three different illustrative LGD assumptions are shown: 15 percent, 35 percent, and 55 percent. For comparison purposes, the current capital requirement on most first mortgage loans is 4 percent and on most home equity loans is 8 percent.

Capital Requirements

[In percentage points]

PD

LGD

15 percent

35 percent

55 percent

0.05 percent

0.17

0.41

0.64

0.10 percent

0.30

0.70

1.10

0.25 percent

0.61

1.41

2.22

0.50 percent

1.01

2.36

3.70

1.00 percent

1.65

3.86

6.06

2.00 percent

2.64

6.17

9.70

5.00 percent

4.70

10.97

17.24

10.00 percent

6.95

16.22

25.49

20.00 percent

9.75

22.75

35.75

Subject to the caveats noted earlier, evidence from QIS3 suggested that advanced approach banking organizations would experience a reduction in credit risk capital requirements for residential mortgage exposures of about 56 percent.

Private Mortgage Insurance

The Agencies wish to highlight one issue associated with the A-IRB capital requirements for the residential mortgage sub-category relating to the treatment of private mortgage insurance (PMI). Most PMI arrangements effectively provide partial compensation to the banking organization in the event of a mortgage default. Accordingly, the Agencies consider that it may be appropriate for banking organizations to recognize such effects in the LGD estimates for individual mortgage portfolio segments, consistent with the historical loss experience on those segments during periods of high default rates. Such an approach would avoid requiring banking organizations to quantify specifically the effect of PMI on a loan-by-loan basis; rather, they could estimate the effect of PMI on an average basis for each segment. This approach effectively ignores the risk that the mortgage insurers themselves could default.

The Agencies seek comment on the competitive implications of allowing PMI recognition for banking organizations using the A-IRB approach but not allowing such recognition for general banks. In addition, the Agencies are interested in data on the relationship between PMI and LGD to help assess whether it may be appropriate to exclude residential mortgages covered by PMI from the proposed 10 percent LGD floor. The Agencies request comment on whether or the extent to which it might be appropriate to recognize PMI in LGD estimates.

More broadly, the Agencies are interested in information regarding the risks of each major type of residential mortgage exposure, including prime first mortgages, sub-prime mortgages, home equity term loans, and home equity lines of credit. The Agencies are aware of various views on the resulting capital requirements for several of these product areas, and wish to ensure that all appropriate evidence and views are considered in evaluating the A-IRB treatment of these important exposures.

The risk-based capital requirements for credit risk of prime mortgages could well be less than one percent of their face value under this proposal. The Agencies are interested in evidence on the capital required by private market participants to hold mortgages outside of the federally insured institution and GSE environment. The Agencies also are interested in views on whether the reductions in mortgage capital requirements on mortgage loans contemplated here would unduly extend the federal safety net and risk contributing to a credit-induced bubble in housing prices. In addition, the Agencies are also interested in views on whether there has been any shortage of mortgage credit under the general risk-based capital rules that would be alleviated by the proposed changes.

Qualifying Revolving Exposures

The second sub-category of retail exposures is QREs. The calculation of A-IRB capital requirements for QREs would require three steps: (1) Calculation of the relevant asset correlation parameter, (2) calculation of the minimum capital requirement assuming no offset for eligible FMI, and (3) application of the offset for eligible FMI. These steps would be performed for each QRE portfolio segment individually.

As in the case of wholesale exposures, it is assumed that the asset correlation for QREs declines as PD rises. This reflects the view that pools of borrowers with lower credit quality (higher PD) are less likely to experience simultaneous defaults than pools of higher credit quality (lower PD) borrowers, because with higher PD borrowers defaults are more likely to result from borrower-specific or idiosyncratic factors. In the case of QREs, the asset correlation approaches an upper bound value of 11 percent for very low PD values and approaches a lower bound value of 2 percent for very high PD values. The specific formula for determining the asset correlation parameter for QREs is as follows:

R = 0.02 * (1-EXP(−50 * PD)) + 0.11 * [1-(1-EXP(−50 * PD))]

Where

R denotes asset correlation;

EXP denotes the natural exponential function; and

PD denotes probability of default.

The second step in the A-IRB capital calculation for QREs would be the calculation of the capital requirement assuming no FMI offset. The specific formula to calculate this amount is as follows:

K

No FMI

= EAD * LGD * N[(1-R)‸−0.5 * G(PD) + (R/(1-R))‸0.5 * G(0.999)]

Where

K

No FMI

denotes the capital requirement assuming no FMI offset;

EAD denotes exposure at default;

LGD denotes loss given default;

N(x) denotes the standard normal cumulative distribution function;

R denotes asset correlation;

G(x) denotes the inverse of the standard normal cumulative distribution function; and

PD denotes probability of default.

Future Margin Income Adjustment

The result of this calculation effectively includes both an EL and a UL component. As already discussed, for QREs only, the Agencies are considering the possibility of allowing institutions to offset a portion of the EL portion of the capital requirement using eligible FMI. Up to 75 percent of the EL portion of the capital requirement could potentially be offset in this fashion. The specific calculation for determining the capital requirement (K) after application of the potential offset for eligible FMI is as follows.

K = K

No FMI

−eligible FMI offset

Where

K denotes the capital requirement after application of an offset for eligible FMI;

K

No FMI

denotes the capital requirement assuming no FMI offset;

Eligible FMI offset equals:

0.75 * EL if estimated FMI equals or exceeds the expected 12-month loss amount plus two standard deviations of the annualized loss rate, or zero otherwise;

EL denotes expected loss (EL = EAD * PD * LGD);

FMI denotes future margin income;

EAD denotes exposure at default;

PD denotes probability of default; and

LGD denotes loss given default.

If eligible FMI did not exceed the required minimum, then recognition of eligible FMI would be disallowed.

The Agencies are interested in views on whether partial recognition of FMI should be permitted in cases where the amount of eligible FMI fails to meet the required minimum. The Agencies also are interested in views on the level of portfolio segmentation at which it would be appropriate to perform the FMI calculation. Would a requirement that FMI eligibility calculations be performed separately for each portfolio segment effectively allow FMI to offset EL capital requirements for QREs?

The following table depicts a range of representative capital requirements (K) for QREs based on these formulas. In each case, it is assumed that the maximum offset for eligible FMI has been applied. The LGD is assumed to equal 90 percent, consistent with recovery rates for credit card portfolios. The table shows capital requirements with recognition of FMI and without recognition of FMI but using the same formula in other respects. As PDs increase, the proportion of total required capital held against EL after deducting the 75 percent offset rises at an increasing rate and the proportion held against UL declines at an increasing rate. Offsets from EL, as considered in this ANPR, would therefore have a proportionally greater impact on reducing required capital charges as default probabilities increase. For comparison purposes, the current capital requirement on drawn credit card exposures is 8 percent and is zero for undrawn credit lines.

Capital Requirement

[In percentage points]

PD

With FMI capital 8%

Without FMI capital 8%

0.05

0.68

0.72

0.10

1.17

1.23

0.25

2.24

2.41

0.50

3.44

3.78

1.00

4.87

5.55

2.00

6.21

7.56

5.00

7.89

11.27

10.0

11.12

17.87

20.0

17.23

30.73

Subject to the same qualifications mentioned earlier, the QIS3 results estimated an increase in credit risk capital requirements for QREs of about 16 percent.

Other Retail Exposures

The third and final sub-category of retail A-IRB exposures is other retail exposures. This sub-category encompasses a wide variety of different exposures including auto loans, student loans, consumer installment loans, and some SME loans. Two steps would be required to calculate the A-IRB capital requirement for other retail exposures: (1) Calculating the relevant asset correlation parameter, and (2) calculating the capital requirement. Both of these steps would be done separately for each portfolio segment included within the other retail sub-category.

As for wholesale exposures and QREs, the asset correlation parameter for other retail exposures declines as PD rises. In the case of other retail exposures, the asset correlation parameter approaches an upper bound value of 17 percent for very low PD values and approaches a lower bound value of 2 percent for very high PD values. The specific formula for determining the asset correlation for other retail exposures is as follows:

R = 0.02 * (1−EXP(−35 * PD)) + 0.17 * [1−(1−EXP(−35 * PD))]

Where

R denotes asset correlation;

EXP denotes the natural exponential function; and

PD denotes probability of default.

The second step in the A-IRB capital calculation for other retail exposures would be the calculation of the capital requirement (K). The specific formula to calculate this amount is as follows:

K = EAD * LGD * N[(1−R)‸−0.5 * G(PD) + (R / (1−R))‸0.5 * G(0.999)]

Where

K denotes the capital requirement;

EAD denotes exposure at default;

LGD denotes loss given default;

PD denotes probability of default;

N(x) denotes the standard normal cumulative distribution function;

G(x) denotes the inverse of the standard normal cumulative distribution function; and

R denotes asset correlation.

The following table depicts a range of representative capital requirements (K) for other retail exposures based on this formula. Three different LGD assumptions are shown—25 percent, 50 percent, and 75 percent—in order to depict a range of potential outcomes depending on the characteristics of the underlying retail exposure. For comparison purposes, the current capital requirement on most of the exposures likely to be included in the other retail sub-category is 8 percent.

Capital Requirements

[In percentage points]

PD

LGD

25 percent

50 percent

75 percent

0.05 percent

0.33

0.66

0.99

0.10 percent

0.56

1.11

1.67

0.25 percent

1.06

2.13

3.19

0.50 percent

1.64

3.28

4.92

1.00 percent

2.35

4.70

7.05

2.00 percent

3.08

6.15

9.23

5.00 percent

3.94

7.87

11.81

10.00 percent

5.24

10.48

15.73

20.00 percent

8.55

17.10

25.64

Subject to the qualifications described earlier, QIS3 estimated a 25 percent reduction in credit risk-based capital requirements for the other retail category.

The Agencies are seeking comment on the retail A-IRB capital formulas and the resulting capital requirements, including the specific issues mentioned. Are there particular retail product lines or retail activities for which the resulting A-IRB capital requirements would not be appropriate, either because of a misalignment with underlying risks or because of other potential consequences?

A-IRB: Other Considerations

As described earlier, the A-IRB capital requirement includes components to cover both EL and UL. Because banking organizations have resources other than capital to cover EL, the Agencies propose to recognize certain of these measures as potential offsets to the A-IRB capital requirement, subject to the limitations set forth below. The use of eligible FMI for QREs is one such potential mechanism that has already been discussed.

Loan Loss Reserves

A second important mechanism involves the allowance for loan and lease losses (ALLL), also referred to as general loan loss reserves. Under the general risk-based capital rules, an amount of the ALLL is eligible for inclusion as an element of Tier 2 capital, up to a limit equal to 1.25 percent of gross risk-weighted assets. Loan loss reserves above this limit are deducted from risk-weighted assets, on a dollar-for-dollar basis. The New Accord proposes to retain the 1.25 percent limit on the eligibility of loan loss reserves as an element of Tier 2 capital. However, the New Accord also contains, and the Agencies are proposing for comment, a feature that would allow the amount of the ALLL (net of associated deferred tax) above this 1.25 percent limit to be used to offset the EL portion of A-IRB capital requirements in certain circumstances.

The offset would be limited to that amount of EL that exceeds the 1.25 percent limit. For example, if the 1.25 percent limit equals $100, the ALLL equals $125, and the EL portion of the A-IRB capital requirement equals $110, then $10 of the capital requirement may be directly offset ($110−$100). The additional amount of the ALLL not included in Tier 2 capital and not included as a direct offset against the A-IRB capital requirement ($125−$110 = $15 in the example) would continue to be deducted from risk-weighted assets.

It is important to recognize that this treatment would likely result in a significantly more favorable treatment of such excess ALLL amounts than simply deducting them from risk-weighted assets. Under the proposal, banking organizations would be allowed to multiply the eligible excess ALLL by a factor of 12.5 because the minimum total capital requirement is 8 percent of risk-weighted assets. In effect, this treatment is 12.5 times more favorable than the treatment contained in the general risk-based capital rules, which allow only a deduction against risk-weighted assets on a dollar-for-dollar basis. In addition, it is important to note that a dollar-for-dollar offset against the A-IRB capital requirement is also more favorable than the inclusion of ALLL below the 1.25 percent limit in Tier 2 capital, because the latter has no impact on Tier 1 capital ratios, while the former does.

The Agencies recognize the existence of various issues in regard to the proposed treatment of ALLL amounts in excess of the 1.25 percent limit and are interested in views on these subjects, as well as related issues concerning the incorporation of expected losses in the A-IRB framework and the treatment of the ALLL generally. Specifically, the Agencies invite comment on the domestic competitive impact of the potential difference in the treatment of reserves described above.

Another issue the Agencies wish to highlight is the inclusion within the New Accord of the ability for banking organizations to make use of “general specific” provisions as a direct offset against EL capital requirements. Such provisions are not specific to particular exposures but are specific to particular categories of exposures and are not allowed as an element of Tier 2 capital. While several other countries make use of such provisions, the Agencies do not believe existing elements of the ALLL in the United States qualify for such treatment.

The Agencies seek views on this issue, including whether the proposed U.S. treatment has significant competitive implications. Feedback also is sought on whether there is an inconsistency in the treatment of general specific provisions (all of which may be used as an offset against the EL portion of the A-IRB capital requirement) in comparison to the treatment of the ALLL (for which only those amounts of general reserves exceeding the 1.25 percent limit may be used to offset the EL capital charge).

Charge-Offs

Another potential offset to the EL portion of the A-IRB capital requirements is the use of partial charge-offs, where a portion of an individual exposure is written off. Given the A-IRB definition of default, a partial charge-off would cause an exposure to be classified as a defaulted exposure (that is, PD=100%), in which case the A-IRB capital formulas ensure that the resulting capital requirement on the defaulted exposure is equal to EAD * LGD, where EAD is defined as the gross exposure amount prior to the partial charge-off. All of this capital requirement can be considered to be covering EL.

The New Accord proposes that for such partially charged-off exposures, the banking organization be allowed to use the amount of the partial charge-off to offset the EL component of that asset's capital charge on a dollar-for-dollar basis. In addition, to the extent that the

partial charge-off on a defaulted exposure exceeds the EL capital charge on that exposure, the amount of this surplus could be used to offset the EL capital charges on other defaulted assets in the same portfolio (for example, corporates, banks, residential mortgages,

etc.

), but not for any other purpose.

An implication of this aspect of the New Accord is that if a defaulted loan's charge-off were at least equal to its expected loss, no additional capital requirement would be incurred on that exposure. For example, consider a $100 defaulted exposure having an LGD of 40 percent, implying an expected loss of $40, equal to the IRB capital charge. If the charge-off were equal to $40, under the New Accord approach, there would be no additional capital required against the resultant $60 net position. The Agencies do not believe this is a prudent or acceptable outcome, since this position is not riskless and a banking organization could be forced to recognize additional charge-offs if the recoveries turn out to be less than expected.

To prevent this possibility, the Agencies propose that, for defaulted exposures, the A-IRB capital charge (inclusive of any EL offsets for charge-offs) be calculated as the sum of (a) EAD * LGD less any charge-offs and (b) 8 percent of the carrying value of the loan (that is, the gross exposure amount (EAD) less any charge-offs).

Also, the charged off amounts in excess of the EAD * LGD product would not be permitted to offset the EL capital requirements for other exposures. In effect, the proposed A-IRB capital charge on a defaulted exposure adds a buffer for defaulted assets and results in a floor equal to 8 percent of the remaining book value of the exposure if the banking organization has taken a charge-off equal to or greater than the EAD * LGD. Importantly, this treatment would not apply to a defaulted exposure that has been restructured and where the obligor has not yet defaulted on the restructured credit. Upon any restructuring, whether associated with a default or otherwise, the A-IRB capital charge would be based on the EAD, PD, LGD, and M applicable to the exposure after it has been restructured. The existence of any partial charge-offs associated with the pre-restructured credit would affect the A-IRB capital charge on the restructured exposure only through its impact on the post-restructured exposure's EAD, PD, and/or LGD.

Purchased Receivables

This section describes the A-IRB treatment for wholesale and retail credit exposures acquired from another institution (purchased receivables). The purchase of such receivables may expose the acquiring banking organization to potential losses from two sources: credit losses attributable to defaults by the underlying receivables obligors, and losses attributable to dilution of the underlying receivables.

22

The total A-IRB capital requirement for purchased receivables would be the sum of (a) a capital charge for credit risk, and (b) a separate capital charge for dilution risk, when dilution is a material factor.

22

Dilution refers to the possibility that the contractual amounts payable by the receivables obligors may be reduced through future cash or non-cash credits to the accounts of these obligors. Examples include offsets or allowances arising from returns of goods sold, disputes regarding product quality, possible debts of the originator/seller to a receivables obligor, and any payment or promotional discounts offered by the originator/seller (for example, a credit for cash payments within 30 days).

Capital Charge for Credit Risk

The New Accord's proposed treatment of purchased loans would treat a purchase discount as equivalent to a partial charge-off, and for this reason it could imply a zero capital charge against certain exposures. In general, a zero capital charge would emerge whenever the difference between a loan's face value and purchase price (the purchase discount) was greater than or equal to its LGD, as might be the case with a secondary market purchase of deeply distressed debt. Again, the Agencies believe that a zero capital charge in such a circumstance is unwarranted because the position is not riskless.

The Agencies propose that for a credit exposure that is purchased or acquired from another party, the A-IRB capital charge would be calculated as if the exposure were a direct loan to the underlying obligor in the amount of the loan's carrying value to the purchasing banking organization with other attributes of the loan agreement (for example, maturity, collateral, covenants) and, hence, LGD, remaining unchanged. This treatment would apply regardless of whether the carrying value to the purchasing banking organization was less than, equal to, or greater than the underlying instrument's face value. Thus, if a loan having a principal amount equal to $100 and associated PD and LGD of 10 percent and 40 percent was purchased for $80, the capital charge against the purchased loan would be calculated as if that loan had an EAD equal to $80, PD equal to 10 percent, and LGD equal to 40 percent.

In general, the same treatment would apply to

This text is long and has been trimmed here. Open the source document for the complete record.