Range of Practice in Banks' Internal Ratings Systems

Executive Summary

Purpose of this Report

In its recent Consultative Paper on a New Capital Adequacy Framework, the Basel Committee stated that development of an internal-ratings based approach ("IRB approach") to regulatory capital would be a key element in the multi-track effort to revise the Accord. In particular, the Committee noted that a capital regime based on internal ratings can prove to be more sensitive to the level of risk in a bank's portfolio, and can provide incentives for industry-wide improvements in risk management practices, consistent with the objectives set forward for the reform of the Accord. The Committee also noted that internal ratings may incorporate supplementary customer information which is usually out of the reach of an external credit assessment institutions; thus, in offering a parallel alternative to the standardised approach based on internal ratings, the Committee hopes that banks will further refine internal credit risk management and measurement techniques.

In spring 1999, the Committee's Models Task Force received a mandate to embark on a study of banks' internal rating systems and processes, and to evaluate the options for relating internal ratings to a regulatory scheme. (An internal rating refers to a summary indicator of the risk inherent in an individual credit. Ratings typically embody an assessment of the risk of loss due to failure by a given borrower to pay as promised, based on consideration of relevant counterparty and facility characteristics. A rating system includes the conceptual methodology, management processes, and systems that play a role in the assignment of a rating.) The Models Task Force is currently working towards this effort, consistent with the objectives noted above, by developing an evolutionary structure that moves rapidly toward basing credit risk capital requirements on a bank's internal ratings to the extent that current bank and supervisory practice will allow. The approach further allows for greater risk sensitivity across banks and over time through a series of incremental improvements - to be developed now or in the future - designed to reflect enhancements in banks' risk management practices. This report presents the preliminary findings of the Models Task Force's recent efforts in developing this evolutionary approach - an assessment of the current state of practice in rating systems and processes, and (of equal importance) the range of practices across institutions.

Since receiving its mandate, the Models Task Force has been actively engaged in gathering information about banks' internal rating systems, and assessing both the "best practice" and overall sound practice in this area. As part of its information gathering, in the spring the Models Task Force undertook a survey of around thirty institutions across the G-10, identified by respective national supervisors as having well-developed internal rating systems. The findings of this survey were supplemented by a series of in-depth presentations from banks and other industry practitioners in September and October, and continuing work by individual Models Task Force members.

These findings will clearly guide the Models Task Force's thinking on its further work regarding how an IRB approach to capital requirements might be structured, and the accompanying sound practice standards and guidelines banks will be expected to follow in order to qualify for the IRB approach. In addition, the Committee believes that these findings have an important bearing on sound practice in credit risk management and on the desirable course of evolution in this field.

The Committee has therefore decided to publish this document at this stage to seek the industry's comments on whether the range of practice identified in the report is truly representative of behaviour, across both banks and countries. In particular, the Committee would welcome feedback on:

• the extent to which the range of practice identified represents "best" or "sound" practice,
• whether important elements of a bank's rating process have been omitted or are given insufficient attention, and
• whether the Models Task Force's preliminary conclusions in respect of some of the elements - for example, the data constraints encountered by many banks in quantifying loss-given-default - are fair and reasonable.

Comments on the report should be sent to the Basel Committee Secretariat, Bank for International Settlements, Basel, Switzerland, CH 4002, by 31 March 2000.