Response to consultation on draft Guidelines on resubmission of historical data
Go back
We would also highlight that even though we appreciate the EBA’s effort to limit historical resubmissions, a cutoff date set at one year would still not make it easy for several reasons to resubmit certain data points. This is due to various factors, including: frequent changes in taxonomy, difficulties in archiving data and/or retrieving data under a previous data point model - DPM (e.g. tools may not allow it), IT costs that this can generate, significant work overload both for the business teams and for the subsidiaries when it comes to transactional data that must be collected over several months beforehand. Even with a cutoff date of less than one year, it is very demanding at the institution level and at the group level to provide a data set that meets all the technical reporting requirements.
According to the consultation paper, the requirements for the resubmission of historical data should be applied at the individual, sub-consolidated and consolidated level. Particularly in the case of large groups of institutions, this requirement would lead to an almost unfeasible effort. Error processing or data inconsistencies cannot be avoided by an individual members of a group of institutions despite all the quality assurance processes. If the respective members were now required to immediately forward errors to the consolidating company and if the latter were required to immediately submit a resubmission at group level (without waiting for deliveries from other group members), this could lead to a huge amount of resubmissions. We are of the view that this cannot be the intended outcome, which could be more precisely clarified in the Guidelines.
Indeed, the requirement to resubmit by low thresholds will limit the capacity of banks and financial institutions to invest resources in improving their reporting processes. In our opinion, the focus of resubmission of data should therefore be on the resubmission on material errors. The present draft guidelines do not include materiality thresholds and should be supplemented by a differentiation as to which data are incorrect. Currently, there is no differentiation as to whether central regulatory data or informational data are incorrect. Regarding a materiality threshold, a differentiation should be considered, for example, as to whether material changes occur, for example, in the own funds, or so-called memo items are incorrect, which ultimately do not affect minimum key ratios relevant to supervision.
The trigger of a resubmission listed in the consultation paper is described in the example (TZ 18 a) on the basis of a "monetary value" of €100,000 and a "precision" of '-3' assumed for the form item. The resulting tolerance amount (+/- 500 €) is then no longer set in relation to the amount of 100,000 €, i.e. this tolerance amount would also apply to amounts in other form positions with the same "precision" and would have to be used independently of the amount of the respective form position. If, for example, the relevant form cell is filled with a value of €100,000,000, a general tolerance amount of +/- €500 would not be practicable and simply not feasible on the part of the institutions.
If the supervisor sticks to a 100% percent precision requirement, banks will not have the technical and operational capability to fully comply with the proposed guidelines and supervisors will be confronted with an overload of resubmissions. The EBA should rethink the objective and premise of the guidelines (i.e. to what extent a resubmission fulfils the purpose of prudential reporting).
We believe instead, that a simple but more differentiated approach, as pursued in the ECB-SSM project on significant resubmissions would be more suitable and deliver more consistent and relevant information to authorities. It would also prevent that institutions have to comply with two separate and non-comparable resubmission requirements, although both have the same objective to improve the reporting by institutions to authorities.
The ECB approach, defining “classes” of information with specific thresholds, would avoid generating an overwhelming number of resubmissions. The ECB provides the following distinctions, for the “Red” data class a 2% relative threshold, for the "Amber” data class a 5% threshold, for the “Green” data class a 10% threshold. This is accompanied by an absolute threshold of 1p.p. (and 5 p.p. for the LCR and NSFR).
We would also highlight that institutions have the possibility to report more conservatively than the minimum requirements stipulated by the supervisory regime. From a supervisory perspective, there is neither a loss of information nor a decrease of knowledge because there is no error in the derived reported data. A resubmission should only be required when an institution has presented an overly optimistic picture of the risk situation, especially with regard to equity and liquidity.
The supervisory authority should clarify what is meant by an “error” or an “inconsistency”. A data inconsistency discovered after a report has been submitted, the resubmission of which would lead to an improvement in the regulatory ratios, should not be described as an error but rather as a positive change and should be excluded from a requirement for a resubmission.
Other examples that could lead to resubmissions but does not stem from an “error”/ non-compliance of EBA validation rules, could be the following: i) wrong figure due to issues/lack of clarity in templates or reporting instructions – a resubmission could be requested to override problems coming from issues in the EBA templates; ii) non-official EGDQ – supervisors sometimes request to resubmit data of several reporting periods due to non-official EGDQs. As these are not shared with institutions before implementing them the institutions cannot be held accountable for issues resulting from them. This highlights the need for the proposed approach to be flexible and not include such scenarios as errors on historical data.
Please see also the General Comments section in the attached document
The new requirements of the EBA Guidelines, without reconsidering: 1) a materiality threshold in close cooperation with the ECB and its significant resubmissions project and 2) a solution not to resubmit data based on a prior DPM, would therefore lead to a considerable additional effort for institutions and supervisors and would not essentially increase the quality and effectiveness of the supervisory tools that the supervisory authorities already have. It will in short not lead to improved supervision.
Prudential reporting infrastructures are highly complex and given the different objectives of the reports to fulfill the needs of regulators (i.e. EBA, SRB, ESMA, IEOPA, ECB), they require institutions to retrieve data from many different sources. Due to this pre-requisite, but also the way that reporting infrastructures are built, backward compatibility to prior DPMs is not feasible without very costly, inefficient (labor intensive) and error prone manual intervention by staff within the institutions. Therefore, it is of key importance that a (material) error shall only be reported in line with the requirements set in the guideline for monthly, quarterly or annual reports with regard to the DPM that is effective and operational the moment the error is reported (i.e. resubmission is sent).
The reporting tools and IT solutions currently in place are not capable of storing the historical reporting parameters in order to resubmit the historical data when there are changes in taxonomy or regulations.
On another note, institutions will soon need to implement the changes (including to reporting systems) stemming from the implementation of Basel III in CRR-CRD along with the new guidelines for the resubmission of historical data. The developments related to the Basel implementation (and therefore IT resources) are prioritized due to the regulatory significance. In light of this, we would recommend introducing a transition period during which data resubmissions would be limited to the current reference period (and not reference dates from at least one calendar year back) .
The guidelines would demand a substantial investment in IT infrastructure in addition to that needed to implement the new regulatory framework. We believe that the overall timeline for the resubmissions’ framework should therefore be revised.
Instead of the "one-size fits all" approach of focusing on a single absolute value, further materiality considerations should be added. Resubmissions should only have to be made if they are material. This concept is applied for example by the IASB to take the idea of "cost-benefit" into account. Analogous to the IASB approach, the assessment of materiality with regard to resubmission of data should be based on whether a resubmission can influence a decision by the data addressee.
Ratios inherently also take into account the size of the institutions which form the basis for decisions by the data addressee. In this respect, it would be advisable to implement the possibility to define proportionality and materiality on the basis of ratios and their potential change through resubmission. A suitable set of indicators would be, for example, the EBA Risk Indicator, which is used analogously by the ECB initiative on "significant resubmissions" currently in its pilot phase.
EBA risk indicators are a suitable way to enable materiality without increasing complexity. Specific materiality thresholds could be based on the potential change in a ratio.
1. What are your general views on the proposed approach to the resubmission of historical data?
EACB members understand very well the importance of high data quality, and are therefore constantly investing in improving the prudential reporting. This does not mean that errors and inconsistencies in day-to-day business no longer occur, especially given the complex technical and professional interdependencies and despite all the control mechanisms. Also, the sometimes complex reporting instructions included in the ITS and non-official EGDQ do not always help to prevent this. In this regard, it is a pity that so many Q&As are necessary to clarify them. Thus, in our opinion, the proposed tolerance limits set out in the consultation paper – i.e. establishing a direct link to the filing rules – are disproportionate to the goal of improving supervisory authorities knowledge by the resubmission of data.We would also highlight that even though we appreciate the EBA’s effort to limit historical resubmissions, a cutoff date set at one year would still not make it easy for several reasons to resubmit certain data points. This is due to various factors, including: frequent changes in taxonomy, difficulties in archiving data and/or retrieving data under a previous data point model - DPM (e.g. tools may not allow it), IT costs that this can generate, significant work overload both for the business teams and for the subsidiaries when it comes to transactional data that must be collected over several months beforehand. Even with a cutoff date of less than one year, it is very demanding at the institution level and at the group level to provide a data set that meets all the technical reporting requirements.
According to the consultation paper, the requirements for the resubmission of historical data should be applied at the individual, sub-consolidated and consolidated level. Particularly in the case of large groups of institutions, this requirement would lead to an almost unfeasible effort. Error processing or data inconsistencies cannot be avoided by an individual members of a group of institutions despite all the quality assurance processes. If the respective members were now required to immediately forward errors to the consolidating company and if the latter were required to immediately submit a resubmission at group level (without waiting for deliveries from other group members), this could lead to a huge amount of resubmissions. We are of the view that this cannot be the intended outcome, which could be more precisely clarified in the Guidelines.
Indeed, the requirement to resubmit by low thresholds will limit the capacity of banks and financial institutions to invest resources in improving their reporting processes. In our opinion, the focus of resubmission of data should therefore be on the resubmission on material errors. The present draft guidelines do not include materiality thresholds and should be supplemented by a differentiation as to which data are incorrect. Currently, there is no differentiation as to whether central regulatory data or informational data are incorrect. Regarding a materiality threshold, a differentiation should be considered, for example, as to whether material changes occur, for example, in the own funds, or so-called memo items are incorrect, which ultimately do not affect minimum key ratios relevant to supervision.
The trigger of a resubmission listed in the consultation paper is described in the example (TZ 18 a) on the basis of a "monetary value" of €100,000 and a "precision" of '-3' assumed for the form item. The resulting tolerance amount (+/- 500 €) is then no longer set in relation to the amount of 100,000 €, i.e. this tolerance amount would also apply to amounts in other form positions with the same "precision" and would have to be used independently of the amount of the respective form position. If, for example, the relevant form cell is filled with a value of €100,000,000, a general tolerance amount of +/- €500 would not be practicable and simply not feasible on the part of the institutions.
If the supervisor sticks to a 100% percent precision requirement, banks will not have the technical and operational capability to fully comply with the proposed guidelines and supervisors will be confronted with an overload of resubmissions. The EBA should rethink the objective and premise of the guidelines (i.e. to what extent a resubmission fulfils the purpose of prudential reporting).
We believe instead, that a simple but more differentiated approach, as pursued in the ECB-SSM project on significant resubmissions would be more suitable and deliver more consistent and relevant information to authorities. It would also prevent that institutions have to comply with two separate and non-comparable resubmission requirements, although both have the same objective to improve the reporting by institutions to authorities.
The ECB approach, defining “classes” of information with specific thresholds, would avoid generating an overwhelming number of resubmissions. The ECB provides the following distinctions, for the “Red” data class a 2% relative threshold, for the "Amber” data class a 5% threshold, for the “Green” data class a 10% threshold. This is accompanied by an absolute threshold of 1p.p. (and 5 p.p. for the LCR and NSFR).
We would also highlight that institutions have the possibility to report more conservatively than the minimum requirements stipulated by the supervisory regime. From a supervisory perspective, there is neither a loss of information nor a decrease of knowledge because there is no error in the derived reported data. A resubmission should only be required when an institution has presented an overly optimistic picture of the risk situation, especially with regard to equity and liquidity.
The supervisory authority should clarify what is meant by an “error” or an “inconsistency”. A data inconsistency discovered after a report has been submitted, the resubmission of which would lead to an improvement in the regulatory ratios, should not be described as an error but rather as a positive change and should be excluded from a requirement for a resubmission.
Other examples that could lead to resubmissions but does not stem from an “error”/ non-compliance of EBA validation rules, could be the following: i) wrong figure due to issues/lack of clarity in templates or reporting instructions – a resubmission could be requested to override problems coming from issues in the EBA templates; ii) non-official EGDQ – supervisors sometimes request to resubmit data of several reporting periods due to non-official EGDQs. As these are not shared with institutions before implementing them the institutions cannot be held accountable for issues resulting from them. This highlights the need for the proposed approach to be flexible and not include such scenarios as errors on historical data.
Please see also the General Comments section in the attached document
2. How do you see the proposed approach in relation to your existing resubmission policies set out in your institutions, agreed with internal audit and control functions?
Extensive control mechanisms have already been established by institutions to ensure the highest possible data quality. In this context, processes have been introduced which, taken together, mean that the data delivered to the supervisory authority on an aggregated basis already have a high data quality and at the same time reflect the risk profile of the institution.The new requirements of the EBA Guidelines, without reconsidering: 1) a materiality threshold in close cooperation with the ECB and its significant resubmissions project and 2) a solution not to resubmit data based on a prior DPM, would therefore lead to a considerable additional effort for institutions and supervisors and would not essentially increase the quality and effectiveness of the supervisory tools that the supervisory authorities already have. It will in short not lead to improved supervision.
3. How do you see the proposed approach in relation to actual practices for the resubmission of data also considering the legal requirements set out in existing legislation (e.g. Article 3(5) of Commission Implementing Regulation (EU) 2021/451)?
When looking at Article 3(5) of (EU) 2021/451 we understand that “other corrections” to be submitted are those which are meaningful for the banking supervisor to understand a bank’ situation – that is, not reporting or correcting them could give a different view about the robustness of the institution with regard to the prudential adequacy and financial stability (i.e. capital requirements, leverage ratio, liquidity ratios, NPL ratios, composition of the balance sheet, and so on). But this would not imply that errors, which do not materially change the view of supervisors, have to be reported. Given that accounting materiality also has an effect on certain parts of the prudential reporting (i.e. amount of own funds, loan loss provisions, valuation of financial instruments) we suggest to also consider the materiality concepts used by the accounting standards setters (e.g. IASB) in order to get a conceptually fully aligned approach with regard to resubmissions.4. Would the proposed approach be feasible from the technology perspective considering the current reporting solutions?
No, totally not feasible.Prudential reporting infrastructures are highly complex and given the different objectives of the reports to fulfill the needs of regulators (i.e. EBA, SRB, ESMA, IEOPA, ECB), they require institutions to retrieve data from many different sources. Due to this pre-requisite, but also the way that reporting infrastructures are built, backward compatibility to prior DPMs is not feasible without very costly, inefficient (labor intensive) and error prone manual intervention by staff within the institutions. Therefore, it is of key importance that a (material) error shall only be reported in line with the requirements set in the guideline for monthly, quarterly or annual reports with regard to the DPM that is effective and operational the moment the error is reported (i.e. resubmission is sent).
The reporting tools and IT solutions currently in place are not capable of storing the historical reporting parameters in order to resubmit the historical data when there are changes in taxonomy or regulations.
On another note, institutions will soon need to implement the changes (including to reporting systems) stemming from the implementation of Basel III in CRR-CRD along with the new guidelines for the resubmission of historical data. The developments related to the Basel implementation (and therefore IT resources) are prioritized due to the regulatory significance. In light of this, we would recommend introducing a transition period during which data resubmissions would be limited to the current reference period (and not reference dates from at least one calendar year back) .
The guidelines would demand a substantial investment in IT infrastructure in addition to that needed to implement the new regulatory framework. We believe that the overall timeline for the resubmissions’ framework should therefore be revised.
a. If not, please provide concrete and realistic proposals for improving the proportionality element that can be efficiently implemented in the reporting systems without unreasonable costs or increasing the overall complexity.
Our view is that the proposed "one-size fits all” approach to the resubmissions that EBA has included in the draft guidelines is from a technical and operational perspective unworkable for institutions. When the Guidelines become effective, institutions would not be able to comply even with the best intent/effort, meaning that the starting point of the new resubmissions’ framework is not fit for purpose.Instead of the "one-size fits all" approach of focusing on a single absolute value, further materiality considerations should be added. Resubmissions should only have to be made if they are material. This concept is applied for example by the IASB to take the idea of "cost-benefit" into account. Analogous to the IASB approach, the assessment of materiality with regard to resubmission of data should be based on whether a resubmission can influence a decision by the data addressee.
Ratios inherently also take into account the size of the institutions which form the basis for decisions by the data addressee. In this respect, it would be advisable to implement the possibility to define proportionality and materiality on the basis of ratios and their potential change through resubmission. A suitable set of indicators would be, for example, the EBA Risk Indicator, which is used analogously by the ECB initiative on "significant resubmissions" currently in its pilot phase.
EBA risk indicators are a suitable way to enable materiality without increasing complexity. Specific materiality thresholds could be based on the potential change in a ratio.