Response to discussion Paper on the future of the IRB Approach
Go back
Once IRB scope is prioritized at first, we agree with the further prioritization as suggested.
Therefore we suggest to increase the time period in such a way that no backward data collection would be necessary. This would mean 5 years of data collection and another two years for calibration and development before filing an application for material changes.
For point e), we agree with the suggested one year horizon, in case the changes do not trigger material changes in parameter estimations, especially in LGD. In that case additional time is necessary.
We suggest that EBA defines minimum standards for the supervisory tolerance of historical data collections and related data quality expectations either in the meaning of defining what could be not acceptable or in the way of defining acceptable methods. For example, whether it would be acceptable to monitor differences between current and proposed definitions for one year and adjust the data history backward accordingly. The clarity in the supervisory assessment would reduce the costs that the industry faces in a significant way, especially related to the uncertainty in the approval process and would also contribute to the reduction of the variability of models.
As for the treatment of LDPs, we believe that the identified difficulties of internal modelling is a very important topic, and as mentioned in the first question we suggest to prioritize as first step in the sequence of regulatory changes in order to save costs in the implementation of the expected regulatory changes. We support the use of the STA approach, as the standards are well known and implemented in all banks. Creating a new regulatory model would further increase the complexity of credit risk approaches, and the definitions of standards, the implementation and a potential approval would be again long lasting and burdensome.
2. What would you consider the areas of priorities?
Due to the costs and resources related to data collection, model redevelopment, calibration, implementation and the related supervisory approval process, we suggest taking into account aspects of how the costs could be minimized when prioritization is set up. Our suggestion in this regard would be to prioritize the ‘scope of IRB’ as first topic and revise PPU scope, especially the inclusion of LDPs. In case a decision is made that PPU scope is potentially applicable for LDPs, or other boundaries of the current PPU would be removed, all the costs related to IRB changes could be saved for such portfolios that could qualify for PPU. It would not make sense to carry on the whole exercise of data collection, model redevelopment and a full scale regulatory approval process for portfolios that a year later would qualify for PPU.Once IRB scope is prioritized at first, we agree with the further prioritization as suggested.
3. Do you consider the proposed timeframe reasonable? In particular do you consider reasonable the proposed timeline for the implementation of the changes in the area of: a. definition of default; b. LGD and conversion factor estimation; c. PD estimation; d. treatment of defaulted assets; e. CRM?
For the topics a)-d) we do not consider the proposed timeline realistic at all. In case backward data collection is necessary, we cannot imagine that a 2 year time line would manageable including not just data collection, but also redevelopment, calibration, and change approval by supervisors. Even if data could be collected backwards, the quality of that data would be questioned by the supervisors in a conservative way, where again the exercise would lead to increased conservativism and less risk-sensitivity of models. (This we see especially unmanageable for internationally active banks, where historical data is often only available in diverse geographic locations.)Therefore we suggest to increase the time period in such a way that no backward data collection would be necessary. This would mean 5 years of data collection and another two years for calibration and development before filing an application for material changes.
For point e), we agree with the suggested one year horizon, in case the changes do not trigger material changes in parameter estimations, especially in LGD. In that case additional time is necessary.
5. Do you have experience with adjustments of historical data? What are the methods that you used to adjust historical data, including both internal and external data?
We believe that this question is a very important one, and when EBA publishes the related guidelines, the supervisory standards for evaluating the historical data collections should be also specified. There should be no big deviation in the assessment standards, otherwise the variability of the IRB models will be just reinforced for as long as the backward data is influencing the models. (The standards are especially critical for retail and small businesses, where no external data collection is possible.)We suggest that EBA defines minimum standards for the supervisory tolerance of historical data collections and related data quality expectations either in the meaning of defining what could be not acceptable or in the way of defining acceptable methods. For example, whether it would be acceptable to monitor differences between current and proposed definitions for one year and adjust the data history backward accordingly. The clarity in the supervisory assessment would reduce the costs that the industry faces in a significant way, especially related to the uncertainty in the approval process and would also contribute to the reduction of the variability of models.
6. To what extent is it possible to adjust your historical data to the proposed concept of materiality threshold for the purpose of calibration of risk estimates?
We do not see currently an opportunity to collect default data backwards in such a quality that it would comply with existing standards. Low quality or ‘estimated’ default data is especially not acceptable, when the change has significant effect on RWAs or expected losses, not mentioning the extreme work of redeveloping dozens of models and re-estimating all related parameters. On the other hand, when the default data would be based only on a two year horizon, the results would be very cyclical requiring frequent adjustments/ redevelopments of models. In addition, we would expect that the conservatism required by supervisors would be punishing due to the uncertainty of the short time horizon.8. Do you consider the direction of the proposed changes adequate to address the weaknesses and divergences in the models across institutions?
In our opinion, it has to be carefully chosen in what detail the changes will be specified in the RTS or in the Guidelines. Too many details would mean material changes for most institutions, on the other hand, if a description is not specific enough, then the envisioned convergence of models will not be achieved.9. Are there any other aspects related with the estimation of risk parameters that should be clarified in the EBA guidelines?
We would appreciate, if EBA could also make more specific criteria to the definition of “long run average” or “economic cycle”, especially with regard to how many minimum or maximum years are to be taken into account or what criteria could define beginning and end of a cycle in such a way that it is acceptable to supervisors. In addition, more specific information would be needed on such related questions as whether the estimation based on long run averages also requires validation on long run data. (And also how supervisors should assess when due to the long run ‘average’ there could be years when rates/estimates measured on one year data are above the average.)13. What are the impacts for the institutions that should be considered when specifying the conditions for PPU and roll-out?
In our opinion it would be important to define the base measure for PPU and TPU. Especially in case of PPU it is unclear, whether it should be measured only for credit institutions and investment firms (i.e. “institutions”) within a consolidated group, or all small legal entities, where naturally/legally (i.e. not covered by CRR on stand-alone basis) only PPU can be used (this would count for PPU scope). (As currently PPU is applied for small legal entities, doing ancillary activities, the ratio of PPU is considerably higher on consolidated level as on single institution level.) Clarity in such respect would be important for consistent treatment within the EU.14. Do you expect that your organisational structure and/or allocation of responsibilities will have to be changed as a result of the rules described in section 4.3.5?
Yes. The independence of the validation function, including separate reporting lines, may require considerable re-organisation by banks. It could also lead to staffing issues, both in terms of hiring and retaining qualified staff. Creating a completely separate validation function may cause difficulties such as high turnover of qualified staff as the work can be seen as repetitive. Availability of experienced staff to conduct validation functions is limited and expensive. It goes without saying that such an important function should not be covered by inexperienced staff.20. What would you consider an appropriate solution with regard to the definition and treatment (modelling restrictions) of the low default portfolios?
We support that the definition of LDPs moves away from the ‘exposure segment based’ approach to the qualitative and quantitative definition taking into account the number of defaults identified and the availability of external data.As for the treatment of LDPs, we believe that the identified difficulties of internal modelling is a very important topic, and as mentioned in the first question we suggest to prioritize as first step in the sequence of regulatory changes in order to save costs in the implementation of the expected regulatory changes. We support the use of the STA approach, as the standards are well known and implemented in all banks. Creating a new regulatory model would further increase the complexity of credit risk approaches, and the definitions of standards, the implementation and a potential approval would be again long lasting and burdensome.
22. Do you see merit in moving towards the harmonisation of the exposure classes for the purpose of the IRB and the Standardised Approach?
Based on the provided description it is not clear enough what such harmonization would mean and why it would restrict modelling choices under the IRB approach.23. Would the requirement to use TTC approach in the rating systems lead to significant divergences with the internal risk management practices?
In our opinion, TTC approach does not lead to divergences with internal risk management practices.Upload files
STN EBA Future of the IRB Approach.docx
(44.47 KB)