ABSTRACT
Basel II entails a three-pronged approach to bank capital regulation: (1) a comprehensive set of rules designed to measure the risks in banks’ portfolios and to produce minimum capital requirements, (2) a supervisory review process setting out the role of
bank supervisors in ensuring that the new framework is correctly implemented, and (3) disclosure requirements to induce banks to make available more information about the key risks in their books with the objective of improving bank accountability and market monitoring. Under the first ‘‘pillar,’’ banks can calculate credit-risk capital requirements by choosing one of the three approaches. This flexibility was introduced in answer to the criticism leveled at the ‘‘one-size-fits-all’’ approach of Basel I, the previous regulation. All three options fundamentally depart from Basel I in that they employ credit ratings as a way to assess individual exposures’ credit risk. The three options are called the standardized approach (SA), the foundation internal rating-based approach, and the advanced internal rating-based approach (collectively called IRBA hereafter). The SA is based on external ratings as assigned by recognized rating institutions whereas the other two approaches, as their names suggest, rely on ratings internally derived by the banks.* National supervisors allow banks to adopt the latter approaches only if they are satisfied that the internal rating assignment process is sufficiently accurate.y
The Basel Committee has conducted several ‘‘quantitative impact studies’’ to test the effect of the new rules on banks’ regulatory capital. However, the main focus of such studies was to compare the new and the old regulation and to determine whether the new regulation would yield substantially lower capital requirements. On the other hand, most of the available studies on the actual accuracy of Basel II refer to earlier drafts of the new framework which has been substantially modified since the release of the first consultative papers in 1999 and 2001, also to incorporate some of the points made in these studies. Examples are, Altman and Saunders (2001) who show that the risk weights associated with the risk categories in the SA do not accurately represent the credit risk of assets that fall in such categories and Sironi and Zazzara (2003) who measure the inconsistencies between the asset correlation assumptions in the IRBA and asset correlations implied from empirically observed default correlations. The final version of the New Accord (Basel Committee on Banking Supervision 2006a) indeed presents new risk weights for the SA and a revised treatment of asset correlation in the IRBA. However, following the latest changes, Resti and Sironi (2007a) take another look at the SA and conclude, in agreement with Linnell (2001), that the different treatment of banks and nonbank corporations in the New Accord is not justifiable. Their empirical evidence suggests that the credit risk of banks and nonbanks with the same rating is statistically indistinguishable. They also conclude that the current risk categories are too coarse and should be more granular to increase accuracy, and that the risk-weight curve across risk categories should be steeper.
