Primary tabs

German Banking Industry Committee - GBIC

The RTS with regard to the IRBA requirements is considered of the highest priority. As these regulations will change many fundamental aspects (e.g. the separation of the development and the validation function, consistent rating within the group) and this RTS will have significant implications on the institutes, sufficient and considerable implementation are inevitable.
The RTS and guidelines (sometimes still at draft stage) which have become known so far strongly suggest that the specifications will necessitate a plethora of substantial model changes and even organisational restructuring. Moreover, the various changes indeed have strong interdependencies. Against this backdrop, it should definitely be made sure that no contradicting regulations will be formulated by the standards and guidelines and that a concerted approach will be used to implement the model changes. With respect to the implementation of the changes, pragmatic ways of handling, also on the part of the supervisory authorities, and appropriate implementation periods should be found to make translation of the changes into the models manageable in the first place.
In our opinion, the implementation steps should, therefore, definitely be based on a wider plan. Among other things, it should be avoided that initially excessive requirements will be made on the IRB approach for individual portfolios (including also low default portfolios, cf. our comments on question Q20) and massive investments will become necessary to then give up the model approach in the medium term.
Moreover, it should be made sure that it is clearly argued in the respective standards and guidelines to what extent the changes proposed serve the purpose of making the equity capital requirements for comparable risks comparable across the banks. In the process, the implementation effort should always be seen in relation to the benefit expected to arise from achieving this goal. In the RTS and guidelines presented so far, it is not always clearly argued to what extent the comparability of the RWAs can be improved by the respective measure, cf. e.g. the organisational requirements in Art. 4(3) and Art. 10 of the RTS on assessment methodology for IRB approach" and our comments on this.
According to the discussion paper, it is intended to initially implement in phase 1 the RTS "Assessment methodology of the IRB Approach" which genuinely does not address the banks but the supervisory authorities. The supervisory authorities will use these RTS for the future IRB reviews. To avoid findings by the supervisory authorities, the banks need to adjust their systems and analyses beforehand. As a consequence, the banks will need to adjust their systems and perform extensive analyses already in phase 1. Therefore, the relief planned by the EBA as regards implementation by the banks fails to take effect. In phases 2 and 3, additional adjustments will then probably be made, due to additional and more concrete RTS or guidelines, with the requirement that phases 2 and 3 will have to be implemented simultaneously by the end of 2018. We believe it would make much sense to develop a final regulation, so that all phases can be implemented together and over a longer period of time (see our reply to question Q3).
The implementation periods can be estimated concretely once the precise design is available. Moreover, we would welcome it if the EBA supplemented the implementation of the discussion paper by impact studies.
We, therefore, believe it is indispensable that the EBA make another adjustment regarding realistic implementation periods as soon as the finalisation of the individual standards and guidelines as regards their contents is due."
Besides finalizing the assessment methodology on the IRB approach we consider the definition of default, PD estimation and Treatment of defaulted assets as priorities in which a revision and standardization of interpretations and approaches will provide the highest value added.
The necessary implementation periods very much depend on the concrete design of the respective requirements. With increasing expenditure, the implementation period required by the banks will become longer. As the requirements have not been specified in detail so far, a blanket statement regarding the implementation periods cannot be made. The EBA will have to consider that in the case of substantial changes the supervisory review process will have to be carried out prior to implementation (cf. RTS on the conditions for assessing the materiality of extensions and changes of internal approaches […]"), which based on our experience the supervisory authorities as well cannot easily do.
We assume that retroactive implementation of the definition of default is impossible. This assumption is based on the fact that data acquired hitherto can no longer be used and new time series need to be obtained. Under the requirements of the CRR, the banks need to make PD estimations on the basis of long-term means of the annual default rates. According to Art. 180 Paragraph 1h and 2e, the underlying observation period must be five years. Implementation would at any rate be delayed by that period.
Within the timeframe proposed, it is realistically possible only to acquire data for two or three years. We assume that such short observation periods are not intended to be used. Moreover, in the case of implementation by 2018, portfolios might be created in which in the year 2018 just one or no year of "new" history exists, in particular because the measurement of default is automated in most banks and sometimes takes place in computer centres. Such a constellation must not result in existing IRBA approvals being revoked.
Furthermore, the existing time series are directly linked to the models. Possibly it is not enough to only replace the time series. In addition, extensive adjustments need to be made to the models which entail substantial expenditure. These changes have to be approved by the supervisory authorities before application. Experience in the past has shown that these approvals mean a major challenge to both the banks and the supervisory authorities, in particular if all methods were to be reviewed simultaneously. Moreover, the question also arises as to the effects on the use and experience test.
Should the EBA consider to impose safety margins by way of precaution during the implementation period, we expressly object to this. The present changes are changes to the statutory basis for the purpose of harmonisation. However, since the methods in question have already been approved by the supervisory authorities and satisfy their requirements, we expect that the banks will be given appropriate time to adjust to the new requirements.
Should, contrary to our expectations, the proposed timeframe apply, the banks would in actual fact have to complete implementation of the requirements in 2017 in order to obtain approval in 2018. The banks cannot achieve this. Either the implementation period must be extended or the expenditure for implementation caused by the changes must be reduced.
The RTS and guidelines (sometimes still at draft stage) which have become known so far strongly suggest that the specifications will necessitate a plethora of substantial model changes and even organisational restructuring. Moreover, the various changes indeed have strong interdependencies. Against this backdrop, it should definitely be made sure that no contradicting regulations will be formulated by the standards and guidelines and that a concerted approach will be used to implement the model changes. With respect to the implementation of the changes, pragmatic ways of handling, also on the part of the supervisory authorities, and appropriate implementation periods should be found to make translation of the changes into the models manageable in the first place.

Q3 (cont’d): In particular do you consider reasonable the proposed timeline for the implementation of the changes in the area of:
a. definition of default;
The proposed timeframe for implementation is inappropriate and by far too short. A changed definition of default will lead to significant adjustments of the data history and with regard to model adjustments. These adjustments will require a materially higher implementation period of at least five years. Besides the institutes using the IRB approach, this affects the institutes applying the standard approach as well.
We would like to point out that the proposed timeframe of 2.5 years to implement the materiality threshold will not only be shortened by determined legislative process itself. The 2.5 years start when EBA's final RTS draft is handed over to the European Commission. Until publication in the EU Official Journal a longer period than three month have to be taken in account. The implementation period is not realistic also against the background that publication in the EU Official Journal initially only means that the national authorities need to adjust their country-specific materiality thresholds. Specification of the materiality thresholds at the national level will likewise take time, so that we assume that the implementation period will be even substantially shorter than 2.5 years.
We find that the adjustment of historical default and loss time series which is strived for cannot be achieved within the envisaged time period of 2.5 years. If adjustment is possible at all, it will require an extremely high amount of manual work which would need appropriate time. However, even then it would not be possible to take account of the banks' internal reactions to worsening situations of borrowers and this would result in an incorrect data basis. Should the EBA wish to alter the requirements regarding credit rating-related restructuring, adjustment of the historical data series would not be possible at all because this would require to also trace back, among other things, all past statutory changes of the insolvency and enforcement law in the individual countries.
At any rate, it should be avoided that a follow up adjustment of the definition of default will need to be done twice, i.e. in 2015 due to the EBA RTS "On the specification of the assessment methodology for competent authorities …" and in 2016 due to the planned "RTS under Article 178(6) – on the materiality threshold" and the "GL under Article 178(7) – on the application of the definition of default". We clearly favour simultaneous implementation of all the standards and guidelines that influence each other.
As said in the GBIC's comments on the "Draft RTS on materiality threshold of credit obligation past due under Article 178 of Regulation (EU) 575/2013", the proposed change of the materiality threshold for 90-days past due would have a substantial effect on the models used in Germany. The validation and model adjustment require the replication of a default detection. Moreover, a large number of data is needed for which the new materiality threshold has to be calculated. This process is very complex and requires close monitoring because the definition of default is highly relevant to all models. The existing historical time series which have been used hitherto for validating and calibrating would need to be adjusted completely. All models for calculating PD, LGD or EAD would need to be redeveloped due to the adjustment of the historical data, but at least recalibrated. Recalibration is possible only if the changes to the definition of default have the same effect on the relevant portfolios, whereas redevelopment of the models is associated with considerable efforts and high costs. Hence, it is of enormous importance that a reasonable handling time be provided for these planned adjustments. To be able to use a sound data basis, we believe it is necessary to provide a transitional period of seven years. Otherwise – if a shorter data period is used – the data will be less meaningful. Moreover, a new definition of default should be converted to at a specific key date because calculation in parallel with the existing definition means high expenditure and can hardly be achieved by the banks.
In our opinion, each bank should state a concrete date at which it will change its calculations from the old materiality threshold to the new one, after the national authorities have defined the respective absolute and relative thresholds.
We would like to point out that a transitional period of seven years will be necessary even if the EBA considered use of the present definition in parallel with the new definition. The differences between the data pools can be analysed and reasonable recalibration factors determined only after an appropriate data set has been created after a few years. Another two or three years are required for the appropriate validation of the parameters because the new validation pools may have a different composition and rating functions may have to be newly developed. For complex models, it has to be assumed that meaningful validation will hardly be possible within a timeframe of less than three years.
However, if it can be assumed that the impact of the expected standards will be rather minor, the new time slices might be added to the existing data history as from the implementation key date and in this way a new data set might be created over time.
We assume that a transitional period would be fairly shorter than in case of major contentual changes of definition of default. Moreover, it would accelerate implementation proceedings substantially, if competent authorities would allow a treatment outside the scope and complexity of the model change policy pursuant to Delegated Regulation (EU) No 529/2014 for these kind of changes which directly arise from changed legal requirements.
At present, an estimation cannot yet be given regarding the qualitative criteria of the definition of default because we need to wait and see how it will be designed in the planned guideline. The same goes for the "return to non-defaulted status". Impact and planned implementation depend on the guidelines still to be provided by the EBA. However, we would like to point out already at this time that the necessary lending processes might not be in line with the models also for a longer period of time because the processes can be converted faster than the models. This discrepancy caused by the change of the statutory requirements must not be to the detriment of the banks.
b. LGD and conversion factor estimation;
We refer to our reply to question Q2.
As said in DK's statement on the "Draft RTS on assessment methodology for IRB approach", we are very critical of the regulation to base the LGD estimation only on the number-weighted average. Of course, the decision needs to be comprehensible based on data, and consistent consideration of all defaults in the PD and LGD estimation and in the EL estimation must be guaranteed.
Banks having chosen the volume-based approach to calculate the LGD will need to make a substantial effort to convert to the number-weighted approach. Due to the necessary analyses, we believe that an implementation period of two years is impossible to comply with.
To give a precise estimation of the implementation period, we need to wait for the final RTS.
c. PD estimation;
n/a
d. treatment of defaulted assets;
In our opinion, it is presently not possible to give an estimate as to the implementation period for the PD estimation and defaulted assets because we need to wait and see how the planned standard will be designed.
e. CRM?
Should the planned RTS be available only at the end of 2017, implementation by the end of 2018 will hardly be realistic, unless the changes resulting from the RTS will be just marginal."
We assess the areas touched upon the definition of default as comprehensive and would like to encourage the EBA to provide practical guidelines resulting in a harmonized default detections, e.g. with regard to counting days past due and allocation of payments. Furthermore, the guidelines should also consider the institutes´ current practices with regard to the definition of default in the IRB approach.
Our experience has shown that precise adjustment usually is not possible and, instead, pragmatic approaches need to be found based on situation and segments. For example, information can in many cases not be obtained subsequently at all or be obtained just with disproportionate effort because no data based on historical cases is available at all regarding the criteria which are relevant now. These approaches include the qualitative review of the definitions of default for compliance with Basel II, the use of regression models for the extrapolation of the time series to the past and the use of scaling factors (e.g. when using benchmark ratings provided by external agencies).
The adjustment of historical data will be, given our experience, extremely challenging and time consuming which holds true on the level of individual banks but also for banking organizations. One concern with regard to this is related to data availability in general. For instance, data warehouses typically store outstanding amounts only at certain reporting dates (e.g. monthly, quarterly). In this case it is not possible (or at least very difficult) to exactly reconstruct historical default dates with a changed materiality threshold. Certain assumptions with regard to the evolution of the outstanding amount between the reporting dates have to be made. Thus, we recommend to the EBA to allow enough time for i) adjustments of historical data, and ii) impact studies on credit risk parameters and RWA / capital effects.
For us this clearly depends strongly on the chosen option in Q1 of the RTS under Article 178(6) – on the materiality threshold". In case of option 1 that changes the definition of default in Germany, existing default time series used so far for validation and calibration purposes of all PD, LGD and EAD rating systems must be more than slightly adapted in German banks. And a large set of additional credit obligations has to be analysed. As this cannot be done in an exact and sufficient manner retrospectively due to non-availability of data, it will take about 1 – 2 economic cycles (about 5 – 10 years) until this may be completely remedied. Alternatively, default time series have to be shortened significantly or the quality of the data will be limited. This is likely to affect the models negatively.
Depending on the chosen options in the final RTS, the adjustment of the historic time series is only possible with extreme costs, or practically impossible. Due to this reason for each institution there should be the possibility in the RTS to avoid the adjustment and make an alternative evaluation of the effect of the change of the materiality threshold on the performance and the calibration of all the affected components of the rating system (PD, LGD and EAD).
In our view, banks should not only be allowed to continue the application of the current definition of default. In fact, this definition of default should be even expanded to other controlling circles (e.g. FINREP, IRFS 9) to reduce complexity. If the effect of a change in the materiality threshold is small, banks should be allowed to apply a simple adjustment factor to central tendencies, especially in case of F-IRB banks."
The RTS and guidelines (sometimes still at draft stage) which have become known so far strongly suggest that the specifications will necessitate a plethora of substantial model changes and even organisational restructuring. Moreover, the various changes indeed have strong interdependencies. Against this backdrop, it should definitely be made sure that no contradicting regulations will be formulated by the standards and guidelines and that a concerted approach will be used to implement the model changes. With respect to the implementation of the changes, pragmatic ways of handling, also on the part of the supervisory authorities, and appropriate implementation periods should be found to make translation of the changes into the models manageable in the first place.
In our opinion, the implementation steps should, therefore, definitely be based on a wider plan. Among other things, it should be avoided that initially excessive requirements will be made on the IRB approach for individual portfolios (including also low default portfolios, cf. our comments on question Q20) and massive investments will become necessary to then give up the model approach in the medium term.
Moreover, it should be made sure that it is clearly argued in the respective standards and guidelines to what extent the changes proposed serve the higher purpose of making the equity capital requirements for comparable risks comparable across the banks. This is not always the case in the standards and guidelines presented so far, cf. e.g. the organisational requirements in Art. 4(3) and Art. 10 of the RTS on assessment methodology for IRB approach" and our comments on this.
Sometimes, the individual changes and their impact on the IRB models are difficult to assess because a number of regulations are still outstanding and their contents have so far only been hinted at. At the moment, we are expecting that the changes presented by the EBA are substantial changes that require the prior approval of the competent authorities. With respect to this, the necessary handling times ("point of no return") for IT implementation must necessarily be provided for in the time schedules. In the event of implementation by the end of 2018, examination for approval would have to take place already in 2017 because otherwise IT implementation and roll-out of the revised methods cannot be effected by the end of 2018. We are very critical of this narrow timeframe.
i) Treatment of default: The treatment of multiple defaults is of particular relevance to retail subsidiaries, because the proposed changes would make them face significant problems. We do expect major impacts on the magnitude of risk estimates based on the proposed similar treatment of multiple defaults, final expectation is subject of an impact study. Models will likely have to be redeveloped and calibrated.
ii) Default rate: We do not expect major impacts since our applied methodology is already similar with the suggested approach.
iii) PD estimation: Our PD models follow a central tendency approach considering default histories across an economic cycle. Thus, we expect the impact to be minor.
iv) LGD estimation: The proposed changes to the LGD estimation will have significant implications for the less relevant due to standardized / F-IRB status of banks inside the mutual banking organization.
v) Downturn adjustment: Less relevant because of standardized / F-IRB status of banks inside the mutual banking organization."
In principle, all institutions have a common interest in a uniform approach to the issues mentioned and that a level playing field is created with respect to this. However, the individual measures can be reasonably assessed only once concrete proposals are available. We have submitted appropriate comments on the RTS and guidelines already published.
In addition, we refer to our reply to question Q7.
Basically, we welcome the EBA's attempt to reinforce the robustness of internal models and the confidence in them. We also understand that the IRB approach is much more than the application of models for calculating equity capital securitisation. However, in some of the proposals regarding the RTS, ITS and guidelines, it does not become clear what the connection is, or how the proposal is to contribute to the stability and comparability of the internal models. This goes in particular for the organisational regulations added to the RTS on assessment methodology for IRB approach", cf. also the DK's comments on this.
The proposed changes do not compensate the loss of degrees of freedom for the modelling and the consideration of portfolio characteristics. The supervision should focus on the transparency of the approaches (e.g. by benchmarking and disclosure) in order to identify and reduce unjustified differences between institutes."
We recommend to address also the area of margins for conservatism in modeling / calibration, e.g. to overcome representative status issues / data issues etc., and suggest to provide some guidelines for application and approval of those margins.
Some banks have LGD models for exposures in default. For these banks, part of the requirements that have become known so far are covered but full coverage of the requirements certainly still requires minor and major adjustments.
On an abstract level we consider the direction of the proposed changes as appropriate.
n/a
Implementation efforts (which depend on data availability, size and age of the portfolio) and strategic importance of a certain portfolio. Besides that we propose to reconsider the balance between IRB thresholds and suitability of certain portfolios for empirical modelling, e.g. LDP, Specialized Lending.
The prohibition in Art. 4(3) in the RTS on assessment methodology for IRB approach" of cooperation between model developers and model validators under outsourcing arrangements renders pool solutions common in Germany practically infeasible, especially as it is not clear how an institution can satisfy the requirement of CRR Art. 190(2)(f) CRR, if it has outsourced the development of the rating system as a key task of the credit risk control unit and Art. 4(3) RTS draft does not permit the outsourcing unit to be involved in the validation unit's activities. The provisions in CRR Art. 179(2) and Art. 190(3)(a)-(e) are de facto rendered null and void by the RTS draft provisions and the large number of advantages for institutions and supervisory bodies alike can no longer be used, see our response to this RTS.
Having to forego pool-rating procedures would have a huge impact on the institutions affected: massive adjustments to the IRBA systems and for each individual institution's organisational structures would be necessary (especially installing resources for permanent local model development and validation). It remains unclear whether the validity of an individual institution's models can be shown in the long run without recourse to the shared data pool (as the starting point of shared modelling).
The independence demanded for large banks up to the senior management level is in principle not workable. It would necessarily entail, however, not only a massive adjustment of the organisation structures but also a duplication of activities, without an iota of additional objectivity. Here, we believe there should be no overshooting of the goal, particularly as the conflicts of interest in conjunction with model development and its validation are clearly smaller than, for example, for the organisational separation between front and back offices, see also our response to the "RTS on assessment methodology for IRB approach".
In any case, such a rigid organisational separation between validation and model development would clearly delay the model optimisation implementation processes.
It is not obvious to us to what extent the comparability of the RWAs can be improved by the organisational requirements set out in Art. 4(3) and Art. 10 of the "RTS on assessment methodology for IRB approach", which is very disproportionate to the enormous costs required to implement this requirement."
We agree on this.
n/a
Generally we agree that the area of disclosure shall be strengthened. However, it should be clarified why publishing at the EBA website is considered necessary and why the current Pillar 3 disclosure channels should not be sufficient.
We do support the EBA guidelines on disclosure requirements. However, the timelines and also the content shall take into account other regulatory initiatives having impacts on data, models and reports, e.g. BCBS 239 efforts but also IFRS 9.
n/a
In general, we find that is indispensable to have available also for low default portfolios (LDP) risk measurement methods which guarantee risk-adequate modelling of the underlying credit risks in segments relevant to the banks. Very general model approaches (such as the modelling of the standardised approach in partial use) do not do justice to the risk content of the respective financing and may give wrong incentives to the banks to do business. Observing the principle of proportionality, the modelling of important portfolio segments of a bank usually is more sophisticated than for less important portfolios.
The usability of the respective models also for the purposes of regulatory equity capital securitisation should, accordingly, also not be regulated in a wholesale manner for specific portfolio areas but be based on which data basis or experience in the relevant field of business is available to the bank. The data basis in particular includes own experience with default and loss, the availability of well-founded economic expert opinions and available external data sources (external ratings, insolvency statistics, market data etc.) which can be used as benchmarks. Based on our experience, economically plausible and statistically valid models can be constructed also for LDPs. Remaining modelling uncertainties can be handled by appropriately more conservative parameters and the limitation of differentiation. The supervisory authorities have the option to narrow the scopes of action by making more strict requirements and in this way force a more uniform procedure.
We believe that the pool models widely used in Germany are one possibility to handle the limited availability of data in LDPs because on a larger data basis the model parameters can be estimated with higher reliability.
The key idea of pool models is to pool data from several institutions, based on which a central outsourced unit (the pool service provider") develops, maintains, reviews, enhances, and operates collective rating systems. The institutions involved participate in the development and enhancement of the systems. All decisions, for example about changes to the rating systems, are made by joint bodies according to clearly defined rules. Based on the pooled data the quality of the pool model is regularly analysed by the pool service provider and evaluated by the joint bodies. In addition to the pool analyses the central unit provides statistics on each institution's individual portfolio, based on which institutions will perform independent internal validations and examine the representativeness of the pooled data and the validity of the pool results for their own portfolios (see Article 179, section 2, item b of the CRR).
Such pool solutions present a number of advantages especially for low default portfolios:
-There is much more data available for estimating the models. As a result models are significantly more differentiated and accurate in terms of discriminatory power than individual models developed by each institution.
-Jointly developed rating systems reduce the variability of the estimated risk parameters between the institutions involved.
-Combining the institutions' expertise helps improve the rating systems in general.
-The joint processes applied by the institutions and the central unit lead to more objective decisions about the models; individual interests of the parties involved become transparent.
Experience shows that there are also advantages for supervisory authorities: they need to audit the models and all changes to them only once and can communicate efficiently with the pool service provider's central points of contact.
Since the standardised approach allows set-off of collaterals to a smaller extent than, in particular, the advanced IRBA (e.g. assigned book accounts), using the standardised approach for low default portfolios should entail higher equity capital requirements. Instead of going back to the standardised approach, we believe it makes more sense to further specify the regulatory requirements for the internal rating methods.
Moreover, we feel that rating certain claim classes (e.g. states and institutions) generally as low default portfolios is not appropriate."
Generally we propose to define harmonized standards for some selected portfolios like sovereigns or specialized lending portfolios for which any kind of internal models will be at their statistical borderline. Further and more general limitations of the use of the IRB approach are not appropriate. Generally, we propose an enhanced transparency level by means of benchmarking and disclosure.
The general idea of harmonization might be convincing. However, banks have their own definitions of portfolios that are not 100% in line with IRB or the Standardized Approach (SA) often. Thus, the advantages of harmonizing IRB and the SA are assessed to be limited, especially when looking at cost benefit considerations.
As you write yourself (subsection 177): In practice the rating systems are in most cases a hybrid of PIT and TTC approach". The current paper does not make it clear why the TTC approach should be more suitable than the PIT approach. This still goes if reference is made to alleged pro-cyclical effects. In its studies, even the EBA comes to the result that there are no pro-cyclical effects worth mentioning due to a corresponding behaviour by the banks (EBA, 2013: Report on the pro-cyclicality of capital requirements under the Internal Ratings Based Approach).
At the moment, we see no reason for adjustments to this general rating philosophy towards pure TTC models because, from our point of view, the portfolio specifics and the economic environment have to be taken into account in the modelling of the methods. Not least, the ratings are not only used for the purpose of regulatory equity capital securitisation but also control many processes at banks and, among other things, go into the pricing of credit commitments. Due to the non-consideration of existing systematic risks, pure TTC approaches may result in misallocation of the resources, especially in crisis situations (low internal price vs. high spread on the market).
Unfortunately, your discussion paper also does not provide a consistent definition of the term TTC but rather just a vague idea. From our point of view, the cyclicality of a method can be judged only in connection with the cyclicality of the portfolio segment looked at and, in particular, with the default rates observed there. So, if we assume a very cyclical portfolio segment, modelling without taking account of the economic environment would inevitably lead to a loss of forecasting quality (the default rates on an annual basis are predicted less precisely and selectivity, seen across time slices, decreases). We generally doubt that such isolation of the non-cyclical risk content can meaningfully succeed, except for very homogeneous portfolios, because the economic environment must necessarily be reflected in the indices used for risk assessment.
From an economic point of view as well, it must be asked whether the equity capital requirements should completely decouple from the fluctuating risks of the credit portfolio. Although the cyclical risks are taken account of in the derivation of the equity capital formula within the meaning of unexpected loss, on the one hand, this is true only under the modelling assumptions (including granular portfolios) and, on the other, especially the financial, shipping, housing and debt crises have shown that it is a particular challenge to capture an underlying cycle. Methods that react to the changed environment only after the parameters have been changed contribute to delayed detection of the underlying risks.
Depending on the portfolio and the risk management application the use of TTC approach in the rating systems will lead to significant divergences. Examples:
-Consumer lending requires a horizon of less than one year frequently. A TTC approach is difficult because the population of the portfolio changes quickly (e.g. continuously changing lending conditions etc.)
-Pricing and provisioning purposes require often a Point in Time view which is reinforced by recent publications, e.g. with regard to IFRS 9 and the guidance on accounting for expected credit losses published by the BCBS
At present, our institutes appropriately make use of mixed approaches. In order to ensure a consistent PD for the majority of controlling circles (e.g. own funds requirements, risk bearing ability, accounting), the supervision should allow institutes to continue proceeding like this in the future as well. Besides a reduction of complexity, this allows for an unambiguous controlling."
In general the idea of removing the permission for the data waiver is in-line with the views stipulated across the entire discussion paper of the future of the IRB portfolio and we agree on this. However, this also means that new products, portfolios etc. must be covered by the standardized approach as long as the data history will not exceed the five years threshold. This might harm the utilization of internal models in general or lead to an attitude of overstretching model chance policies respectively tweaking existing models to new products and portfolios.
We have concerns about the deletion of the data waiver", in particular against the backdrop of the fundamental changes which are planned with respect to the rating methods and the data to be used.
However, the discussion paper does not make it clear why the option to use a two-year period in exceptional cases shall no longer be available in the future. Based on the current descriptions, it is not possible to judge this."
n/a
Silvio Andrae
+49 30 20225 5437