When estimating risk parameters, the institution faces a number of methodological challenges - ITCB's experts can address them! 

Does the institution have a historical database of sufficient size? 

We recommend a central tendency-based, default frequency-based rating/scoring calibration. This method can predict PD values for all categories based on a given long-term expectation using traditional scorecard-based estimates.

Although frequency calibration is not possible, there are many other methods available. For a non-frequency calibration, we calibrate the existing scores to a secondary rating. This involves transactions for which there is a secondary internal rating or for which the long-term default rate (central tendency) associated with the secondary external rating is known. In this case the calibration is a two-step process:

1. The calibration between the internal score and the external rating has to take place.
2. The calibration between the external rating and the central tendency as PD must take place.
3. The default rates for the external rating can be assigned to the internal category point averages, and thus the above frequency-type calibration steps should be performed between the internal point averages and the default rates.

The secondary rating can thus also be a categorisation consistently defined by experts (normally an external rating can be mapped to the portfolio in this way).

After calibration, determining the PD values for each category based on the known default rate time series is a straightforward task. Knowing the central tendency for each definit category, the rating category values can be translated into average PD values, i.e. a PD-score correlation can be established. Knowing this correlation, the PD value of any portfolio component or even client can be determined, and the calibration results can be mapped to the master scale used.