Range of Practice in Banks' Internal Ratings Systems
A discussion paper by the Basel Committee on Banking Supervision
Summary of Findings: The efforts of the Models Task Force have highlighted both the similarities and differences in the structure, methodology, and application of internal rating systems at banking institutions. In particular, while it appears that there is presently no single standard for the design and operation of an internal rating system, a small number of alternative approaches emerged from the survey and presentations. In the report, we do not aim to judge the merits of each type of approach, but attempt to evaluate their key elements, and discuss how these processes may, over time, evolve into a well-defined range of sound practices.
The banks covered in our analysis tended to be large, diversified international banks; however, a small number of more specialised institutions was also included. This sample was selected with a view to uncovering the range of potential policy issues and considerations in constructing an IRB approach, such as the range of structures of rating systems, the various extents to which they have developed, the use to which the rating information is put, the type of portfolio rated, and the degree of reliance on expert judgment versus statistical models in assigning ratings. To a considerable extent, these decisions were guided by bank-specific rather than country-specific considerations.
In the following paragraphs, we provide a brief overview of our initial findings on the architecture of the rating systems at these banks:
- The survey highlighted a considerable number of common elements among rating systems. These include the following: (a) Banks appear to consider similar types of risk factors - such as the borrower's balance sheet, income statement, and cash flow performance - when assigning a rating. However, the relative importance of these factors, and the mix between quantitative and qualitative considerations, differed between the banks surveyed, and in some cases, between different borrower types within the same bank; (b) regardless of whether banks assign ratings to borrowers, facilities, or both, the vast majority of banks surveyed assign these ratings based on assessment of the counterparty. Those banks assigning grades to individual facilities also consider the degree of risk contributed by the specific characteristics of the transaction being rated, while some institutions assigning only borrower grades appear to be considering facility characteristics when allocating economic capital for credit risk; (c) the information gleaned from ratings is utilised (or expected to be utilised) in broadly similar processes at the banks surveyed, including management reporting, pricing and limit setting.
- While there does not appear to be a single standard for the structure and operation of internal rating systems, the survey highlighted a few alternative approaches. These can be viewed as points on a continuum with, at one extreme, systems focused on the judgment of expert personnel, and at the other, those based solely on statistical models.
- The survey has also highlighted a number of other areas where divergence in bank practice appeared more distinct. These include banks' methods for quantifying loss characteristics per grade. Data constraints also remain a challenge, both to banks' efforts to quantify risk, and, ultimately, to supervisors' efforts to validate banks' internal grades (the survey does suggest, however, that some banks are making progress in collecting and analysing internal data for certain market segments covering the past few years).
Download paper (1772K PDF) 46 pages