FAQs on the PD & Rating platform
Q: For what purposes can I use the data from the PD & Rating platform?
A: By making use of the database, participating member banks have the possibility to further enhance the backtesting, calibration and benchmarking of their internal PD models. Although banks are required to use their own internal default history to build and calibrate PD models, in case of low default portfolios additional data information can be necessary to ensure a smooth calibration or backtesting. In addition, members banks can directly compare their level of calibration and model accuracy with peers and therefore get further insights into the overall performance of their models.
Some concrete use cases are:
- Benchmark your PD masterscale (comparison PDs and default rates per rating class)
- Benchmark your system’s discriminatory power
- Identify macro-economic dependencies in default and migration data: Extract a “systemic factor” from rating migrations or default rates
- Benchmark your asset correlations and long term default rates
- Benchmark your stage allocation / SICR buckets (thresholds for “life-time PD” movements) under IFRS 9
- Reduce uncertainty add-ons for lack of data
Q: How does GCD take into account that the participating member banks use different methods on how to calculate default rates?
A: Many market practitioners, such as portfolio owners (banks and investment companies), external rating agencies, auditors/ regulators and academics perform and publish calculations on default and transition rates. For estimating transition and default probabilities of rating grades, the so-called “cohort approach” is still considered to be the most common one. The committees of GCD have decided to implement that approach for its “PD & Rating platform” in order to get uniform results and results which are the most comparable with other calculations from market participants.
Members banks can either
a) deliver details of their entire portfolio at (anonymized) borrower level and allow GCD to calculate default rates and migration matrices on the defined methodology or
b) receive a tool (the so-called “desktop tool”) to use inside the member’s office, which calculates default rates and migration matrixes and creates aggregation files to be delivered to GCD.
The defined methodology takes into account how to deal with “double defaults / multiple defaults”, exits from the portfolio, new borrowers etc. and the aggregated output is therefore uniform for all participating banks. Nevertheless, further analysis is possible for all participating banks on the effect of those elements, using the additional markers delivered.
Q: How does GCD take into account that the participating member banks use different internal rating scales?
A: The requirement is that banks map their internal rating system categories to the S&P rating scale. Therefore, the delivery to GCD does not include the raw internal ratings but the “internal rating mapped to the nearest external rating category”. Banks need to do this mapping for other purposes already, such as external disclosure, regulatory reports and even for using rating agency historical default data and therefore have the knowledge in house to perform such a mapping before submitting data to GCD. Our advice is that they refer to the rating agency’s guidance on definitions of rating grades as well as take into account the published long-term default rates of the rating agencies.
Q: Is there a possibility to participate if my bank is not willing to share raw portfolio information?
A: Yes, our desktop tool is exactly built for those cases. Some member banks have expressed in the past their concerns that they do not want / cannot share borrower-level data / portfolio snapshots directly with GCD, even if the data is anonymized. A participating bank can decide whether they want to submit borrower-level input data or aggregated information. “Aggregated information” means that the information on borrower level will be aggregated inside the desktop tool into averages / counts per asset class, country, rating grade, etc. The submission files contain this aggregated information as well as the calculated migrations and defaults. Members can check the submission files to ensure that they are truly aggregated information, before manually submitting them through the portal.