We understand that guidelines set out in section 4.2 apply to all EU credit institutions regardless of the applicable accounting framework, whereas the guidelines set out in section 4.3 apply to credit institutions which prepare their financial statements in conformity with IFRS standards.

The aim of section 4.2. is to define streamlined and harmonised credit risk management practices related to an ECL model but regardless of the accounting framework, whether it is an ECL accounting framework or not, as stated in §10. We believe that accounting should not become distanced from risk management practises. Thus, there is no rationale to build a sound credit risk management disconnected with the application of the impairment accounting model. Besides, the implementation of an ECL risk management model depends on the availability of the necessary historical and statistical and the capacity to collect a sufficient level of data at portfolio levels or entity levels. Therefore, when defining the scope of application of the guidelines, the EBA draft guidelines should recognise the possibility of a proportionated application of an ECL model and the consistency between the impairment accounting framework and credit risk management practises.
As the IFRS 9 standard is implemented on 1st January 2018 subject to its endorsement in the EU, the date of application of the guidelines of 1 January 2018 is appropriate.
We appreciate that the Draft Guidelines recognize the benefit of a proportionate approach “in a manner that is appropriate to their size, internal organization and the nature, scope and complexity of their activities and, more generally, all other relevant facts and circumstances of the credit institution and the group (if any) to which it belongs” (§17).

We acknowledge that banks should endeavour to develop systems and processes in order to achieve a high quality implementation of the ECL model.

However, we believe that consideration should be given to cost effectiveness. Indeed, the development of a high quality model and processes rely notably on sound statistical data sets.

When developing adequate models and processes, smaller banks will often face the issues of identifying relevant and reasonable information and having historical data that are reliable and sufficiently deep. Particularly as models to score customer behaviour will improve with long-term experience and as new models will be developed based on past experience by the time of IFRS 9 implementation. Larger and more complex banks face similar issues in developing appropriate models as less complex banks. They may face a lack of sound and deep statistical data for their smaller, less complex or immaterial portfolios.

Therefore, we believe that the principle of proportionality should also consider that more complex banks should be allowed to apply a similar approach for their smaller or less complex portfolios. It should apply to credit institutions of different size or complexity, as well as to subsidiaries or branches and portfolios of banking groups.
Hence, we advocate that banks should take into account a cost - benefit analysis in accordance with the principle of proportionality and that the proportionality principle should not be limited to less complex banks only, rather it should be applicable to credit portfolios based on their size, nature and complexity.

As far as practical expedients are concerned, the Draft Guidelines limits the use of practical expedients to “both smaller and less complex banks” (§129). We believe that in order to be consistent with a proportionate approach, the practical expedients should be used by larger banks at portfolio levels for immaterial or less complex credit exposures as they were intended to be available to banks to ease implementation of IFRS 9.
We appreciate that the EBA Draft Guidelines intend to be in line with the BCBS guidance.

However we would like to raise the following concerns:

Forward-looking information.

The paragraph 23 of the Draft Guidelines states that robust documentation should be provided when information is not retained as not reasonable and supportable creating for banks additional operational burden. We believe that it overrides IFRS 9 disposals as it narrows the application of the “undue cost and effort” notion. Thus, we would suggest to delete the last sentence of the paragraph.

Usage of scenarios.

We question the reference to back testing scenarios. Banks may leverage stress-testing methodologies to apply appropriate information adjustments. However, the use of information resulting from the stress testing or sensitivity analysis could be possible, but only on the basis of expected macroeconomic information and not on stressed macroeconomic factors. Therefore, the latter could be not relevant as a measurement basis of ECL allowances, as stated in §33o).
Clarification should be provided to paragraph 38d) whether, in a context of multiple scenarios, the relevant scenarios should be back tested or whether the relevant criteria should be considered to be back tested.

The paragraph 41 requires that the probation period as defined in Commission Implementing Regulation (EU) 2015/227 should apply to the accounting impairment. Yet, the Regulation is intended to apply for regularity reporting purposes only. For accounting matters, the reference to the Regulation (EU) 2015/227 should not be compulsory and therefore its use should be left at the discretion of the bank if relevant.

ECL model validation.

In some instances, the wording of the EBA Guidelines is stricter than the wording of the BCBS Guidance. Therefore, we would suggest retaining the wording of the BCBS Guidance in the following cases.

To avoid that requirements related to sound ECL methodologies would be seen as checklists rather than indicators, “should” should be replaced by “could” (§36, §37) or by “generally will” (§33). Indicators or fact patterns should be presented as examples and not as requirements (§40).

While banks must have robust validation methodologies of their credit-risk ratings systems and ECL models, some events may be not included in the ECL models because they cannot be modelled or because they are too recent to be included in the models. The expert judgment will then be used, notably when incorporating relevant forward looking information, to determine the ECL allowances. Thus, we suggest to amend the wording of §65 of the EBA Guidelines and to retain the wording of the §60 of the BCBS Guidance (“As the development and use of ECL assessment and measurement models involve extensive judgment, effective model validation policies and procedures are crucial”).

Disclosures requirements.

We question the relevance of the disclosure requirements within the Draft Guidelines, notably in paragraphs 80 and 81. IFRS 7 already requires extensive qualitative and quantitative disclosures in order to explain risk management practices and assumptions related to ECL estimates and forward looking information. The Pillar 3 framework aims to improve the consistency and comparability of institutions’ disclosures for users.

Guidelines specific to credit institutions applying IFRS 9.

We believe that the use of practical expedients envisaged by IFRS 9 should not be forbidden for banks as long as their use does not affect the timing or the amount of ECL recognised in these situations. We understand that for some portfolios (e.g. trade receivables that a conglomerate bank might have on its statement of financial position) the use of practical expedients could be appropriate.

We believe that the paragraphs §127 - § 130 should clarify that the use of practical expedients should also be considered at the portfolio levels and not only at the entity level.

The EBA limits the use of the 30-days-past-due indicator (§137 - § 102).The EBA should acknowledge that there may be limited circumstances in which the more-than-30-days-past-due rebuttable presumption may be a relevant factor and be considered as an acceptable indicator.

It would notably be the case for some retail portfolios where the 30-days-past-due indicator would better reflect consumer behaviour and where macroeconomic factors and forward-looking information would not be necessarily relevant. The 30-days-past-due indicator is appropriate for these nature of retail portfolio and consistent with high quality implementation of an ECL model.

Loss allowance at an amount equal to 12-month ECL.

In paragraph 86, the EBA seems to consider that the ECL should be assessed at an individual level, even for high quality portfolios, whilst IFRS 9 is not prescriptive regarding the assessment of expected losses on an individual basis or a portfolio basis. We believe that clarification should be provided in the draft guidelines not only for loss allowances at an amount equal to 12 month ECL but also for lifetime loss allowances.

Assessment of significant increases in credit risk.

We understand that according to §113 of the EBA Draft Guidelines, there is a systematic link between degradation of a rating or more intensive monitoring of a borrower and significant increase in credit risk. We disagree that a downgrade of a rating should automatically be considered as a significant increase in credit risk and, thus, lead to a transfer from stage 1 to stage 2. A rating may be downgraded without a significant increase in credit risk having occurred. Out of an internal rating grid composed of 8 grades for instance, the downgrade of a counterparty from the grade 1 to the grade 2 could not be translated in automatic “significant” decrease of credit risk. Indeed, the associated probability of default of such counterparty would still be extremely low. Degradation of a rating or more intensive monitoring of a borrower is an indicator but it should be considered as an indicator among others to assess the level of significance of the credit risk deterioration. No automaticity should prevail when determining if there is a significant increase of credit risk.
We have no comments.
We have no additional comments.
Isabelle Huard