The table makes clear that for corporate, sovereign, and interbank exposures, all IRB banks must provide internal estimates of PD. In addition, advanced IRB banks must provide internal estimates of LGD and EAD, while foundation IRB banks will make use of supervisory values contained in CP3 that depend on the nature of the exposure. Advanced IRB banks will generally provide their own estimates of remaining maturity for these exposures, although there are some exceptions where supervisors can allow fixed maturity assumptions to be used instead. For foundation IRB banks, supervisors can choose on a national basis whether all such banks are to apply fixed maturity assumptions described in CP3 or to provide their own estimates of remaining maturity.

Another major element of the IRB framework pertains to the treatment of credit risk mitigants, namely, collateral, guarantees and credit derivatives. The IRB framework itself, particularly the LGD parameter, provides a great deal of flexibility to assess the potential value of credit risk mitigation techniques. For foundation IRB banks, therefore, the different supervisory LGD values provided in CP3 reflect the presence of different types of collateral. Advanced IRB banks have even greater flexibility to assess the value of different types of collateral. With respect to transactions involving financial collateral, the IRB approach seeks to ensure that banks are using a recognised approach to assessing the risk that such collateral could change in value, and thus a specific set of methods is provided, as in the standardised approach.

Retail Exposures

For retail exposures, there is only a single, advanced IRB approach and no foundation IRB alternative. The key inputs to the IRB retail formulas are PD, LGD and EAD, all of which are to be provided by the bank based on its internal estimates. In contrast to the IRB approach for corporate exposures, these values would not be estimated for individual exposures, but instead for pools of similar exposures.

In light of the fact that retail exposures address a broad range of products with each exhibiting different historical loss experiences, the framework divides retail exposures into three primary categories:

1. exposures secured by residential mortgages,

2. qualifying revolving retail exposures (QRRE), and

3. other non-mortgage exposures also known as 'other retail'.

Generally speaking, the QRRE category captures unsecured revolving credits that exhibit appropriate loss characteristics, which would include many credit card relationships. All other non-mortgage consumer lending including exposures to small businesses falls into the 'other retail' category. A separate risk-weight formula for each of the three categories is provided in CP3.

Specialised Lending

Basel II distinguishes several sub-categories of wholesale lending from other forms of corporate lending and refers to them as specialised lending. The term specialised lending is associated with the financing of individual projects where the repayment is highly dependent on the performance of the underlying pool or collateral. For all but one of the specialised lending sub-categories, if banks can meet the minimum criteria for the estimation of the relevant data inputs, they can simply use the corporate IRB framework to calculate the risk weights for these exposures. However, in recognition that the hurdles for meeting these criteria for this set of exposures may be more difficult in practice, CP3 also includes an additional option that only requires that a bank be able to classify such exposures into five distinct quality grades. CP3 provides a specific risk weight for each of these grades.

For one sub-category of specialised lending, 'high volatility commercial real estate' (HVCRE), IRB banks that can estimate the required data inputs will use a separate riskweight formula that is more conservative than the general corporate risk-weight formula in light of the risk characteristics of this type of lending. Banks that cannot estimate the required inputs will classify their HVCRE exposures into five grades, for which CP3 also provides specific risk weights.

Equity Exposures

IRB banks will be required to separately treat their equity exposures. Two distinct approaches are described in CP3. One approach builds on the PD/LGD approach for corporate exposures and requires banks to provide own PD estimates for the associated equity exposures. This approach, however, mandates the use of a 90% LGD value and also imposes various other limitations, including a minimum risk weight of 100% in many circumstances. The other approach is intended to provide banks with the opportunity to model the potential decrease in the market value of their equity holdings over a quarterly holding period. A simplified version of this approach with fixed risk weights for public and private equities is also included.

Implementation of IRB

By relying on internally generated inputs to the Basel II risk weight functions, there is bound to be some variation in the way in which the IRB approach is carried out. To ensure significant comparability across banks, the Committee has established minimum qualifying criteria for use of the IRB approaches that cover the comprehensiveness and integrity of banks' internal credit risk assessment capabilities. While banks using the advanced IRB approach will have greater flexibility relative to those relying on the foundation IRB approach, at the same time they must also satisfy a more stringent set of minimum standards.

The Committee believes that banks' internal rating systems should accurately and consistently differentiate between different degrees of risk. The challenge is for banks to define clearly and objectively the criteria for their rating categories in order to provide meaningful assessments of both individual credit exposures and ultimately an overall risk profile. A strong control environment is another important factor for ensuring that banks' rating systems perform as intended and the resulting ratings are accurate. An independent ratings process, internal review and transparency are control concepts addressed in the minimum IRB standards.

Clearly, an internal rating system is only as good as its inputs. Accordingly, banks using the IRB approach will need to be able to measure the key statistical drivers of credit risk. The minimum Basel II standards provide banks with the flexibility to rely on data derived from their own experience, or from external sources as long as the bank can demonstrate the relevance of such data to its own exposures. In practical terms, banks will be expected to have in place a process that enables them to collect, to store and to utilise loss statistics over time in a reliable manner.