Internal rating systems
Basel II offers banks the opportunity to adopt more sophisticated methods for the measurement of regulatory capital credit risk.
The development of the inferential apparatus needed to calculate this figure, in accordance with the AIRB (Advanced Internal Rating Based approach), requires the construction of appropriate statistical models for the prediction of three indicators for each portfolio presentation (i-th):
- PD (probability of default)
- LGD (loss given default)
- EAD (exposure at default)
The Basel Committee supplies for each presentation (businesses, governments, banks, retailers) the formulas for the calculation of the assets requirements needed to cover expected and unexpectedlosses associated with each of the presentations. These formulas have exactly the above-mentioned indicators as the basic elements. The sum of these requirements gives us the size of the Surveillance Asset.
For the estimate of each of these three components, we proceed to the construction of one or more scoring models on the basis of the homogeneity of the default rates of the presentations which make up the portfolio.
In the construction of an internal rating system, the principal features that must be dealt with relate to:
- Default definition
- Definition of the components of the loss associated with the collection applications and the calculation modes of the loss for each presentation
- Sample extraction for model development
- Historical depth analyses of the internal behavioural data
- Quality of available data and selection of useful information
- Assessment of data acquisition from external data banks.
The internal rating systems which are worked out to calculate capital requirements in accordance with the Basel indications, are also an efficient tool to monitor the risk on the current applications in portfolio and can consequently influence both the collection procedures, and as a result of this, also the credit policy in the application phase.