UniCredit appreciated that the European Banking Authority has involved the industry in the consultative process and is grateful of this opportunity to raise its position on the proposed changes to the reporting requirements of operational risk.
Overall UniCredit strongly believes that it is crucial to develop uniform reporting requirements to guarantee the comparability among banks’ loss profiles, avoiding different reporting implementations. In UniCredit opinion, to meet this target, it is fundamental to clarify as much as possible the reporting requirements through detailed explanations and basic examples.
It is worth mentioning that the proposal to uniform reporting requirements is very welcome by UniCredit. Even if the reporting requirements are defined in the Regulation EU No 680/2014 (successively amended by Regulation 2015/227), there is still room for different interpretations and then for not homogeneous implementations of the reporting among institutions. As a consequence, there could be a lack of comparability among banks’ loss profiles, and then among banks’ risk profiles.
Moreover, regarding OpRisk losses, UniCredit and its peers are currently subject to several Supervisory Reporting requests, in particular:
• COREP C 17.00, i.e. the object of this EBA Consultation Paper;
• SSM Short-Term Exercise (STE);
• BCBS Quantitative Impact Study (QIS);
• EBA EU-wide Stress Test.
All these Supervisory Reporting requests require similar loss data statistics, but they include different peculiarities. Some requirements could be difficult to understand, leading the banks to different interpretations. In particular, it is not always clear how to assign number of events and loss amounts to different years and different ranges.
The current situation leads to human resources efforts and implementation costs dedicated to:
• answer redundant reporting requests and
• data reconciliation to justify any different result among reporting requests.
In light of mentioned rationales, regarding OpRisk losses reporting requirements, UniCredit would strongly support the following actions:
1. Unify the Supervisory Reporting requests into COREP
2. Define uniform and, possibly, stable reporting requirements, including detailed explanations and basic examples, to avoid different interpretations/implementations
UniCredit recognises that the content of this Consultation Paper represents a significant step forward in the expected direction, i.e. the development of uniform reporting requirements for OpRisk losses. In UniCredit opinion, this initiative will lead to costs/efforts reductions and to an increased consistency among loss data reported by different institutions.
In case of AMA partial use institution, it would be useful to specify the scope of the report. In our understanding, the scope includes AMA, TSA and BIA segments of the institution.
We would also raise the issue of simplicity, since this is one the main rationales of the Standardized Measurement Approach (SMA), introduced by the related BCBS Consultation Document as of March 2016. The new templates C 17.01 and C 17.02, compared to the current C 17.00 represents an increase of complexity, although SMA is meant to reduce it. Generally, every dimension in reporting enhances the complexity further. In our opinion, some “dimensions” of information are not necessary and could be abolished in favour of simplification and comparability across the industry. For example, the reporting by Basel Business Line could be no longer necessary, in case the Standardized Approach (TSA), which is based on Basel BL, is replaced by the SMA.
UniCredit really appreciates the following clarifications related to the C 17.01 template (substituting C 17.00):
• “number of events” and “gross loss amount” consider only values relating to OpRisk events “accounted for the first time” within the reporting period;
• “maximum single loss” and the “sum of the five largest losses” do not include any recoveries;
• boundary credit related OpRisk events should not be reported in the template;
• the differentiation between direct recoveries and recoveries from insurance and other risk transfer mechanisms, and collection of this information independently from when the original event occurred.
It is also appreciated the distinction between OpRisk events with first impact in the reference period and loss adjustments of events with first impact in previous periods to better understand the current level of losses of the institution. This distinction includes the following useful clarifications:
• the inclusion of new rows, i.e. “Number of events subject to loss adjustments” and “Loss adjustments relating to previous reporting periods”, to collect (positive or negative) loss adjustments “accounted for the first time” within the reporting period but relating to OpRisk events “accounted for the first time” in previous reporting periods;
• the fact that the rows “Number of events”, “Gross loss amount”, “Number of events subject to loss adjustments”, “Loss adjustments relating to previous reporting periods”, “Maximum single loss”, “Sum of the five largest losses” do not include any recoveries.
• the inclusion of new rows to collect the breakdown of loss adjustments by size of the loss (not by size of the adjustment). Values can be positive or negative as they show the adjustment, not the overall loss amount.
Nonetheless, it could be useful to specify which economic components are included in the gross loss definition. In our understanding, the following ones are included:
• losses, including direct charges and settlements;
• provisions write-backs.
The sentence in the point 128 (b) of the Annex 2 “The Maximum single loss is the larger of either the largest gross loss amount for events reported for the first time within the reporting period or the largest positive loss adjustment for events reported for the first time within a previous reporting period” should probably be referred to a distinct point, i.e. 129.
In our opinion, it is important to clearly define how to treat the amounts of provisions, losses, provisions write-backs and recoveries related to non-Euro currencies. Usually, the non-Euro amounts are converted in Euro on the basis of the FX rate related to the accounting date. For example, if we have an event with a provision and a provision write-back having the same amount in US Dollar, accounted in two different dates within the same year, then we will not observe any loss (or loss adjustment) in US Dollar, but we could observe a positive or negative loss (or loss adjustment) in Euro, on the basis of the different FX rates in the two different accounting dates.
It is also important to clearly define how to treat the amounts of provisions, losses, provisions write-backs and recoveries related to proportionally consolidated legal entities. Usually, the amounts are multiplied by the consolidation percentage of the related legal entity. For example, if a legal entity is proportionally consolidated at 40%, then all the related amounts will be multiplied by 40%.
The Annex 2, for rows 930-934, reports that “Based on the rules for the determination of the number of loss events subject to loss adjustments per range, the total number of loss events subject to loss adjustments (row 930) may be lower than the sum of the numbers of loss events subject to loss adjustments per range (rows 931 – 934)”. It is worth mentioning that the total number of loss events subject to loss adjustments (row 930) may be also higher than the sum of the numbers of loss events subject to loss adjustments per range (rows 931 – 934), because the internal data collection threshold can be lower than 10,000 €.
In UniCredit opinion, some doubts can be raised on the assignment of loss adjustments to ranges as defined for rows 940 to 944. It is specified that: “If a loss moved to a different size class, the negative (previous) amount should be reported in the size class before the adjustment and the positive (new) amount in the size class after the adjustment.” .
Moreover, the Annex 2, for rows 930-934, specifies that:
(c) If the loss adjustment entails a change in the range, institutions shall reclassify the loss by excluding the original loss amount from the range applicable before the adjustment (report the amount with a negative sign in the respective row) and assigning the adjusted loss amount to the range applicable after the adjustment (report the amount with a positive sign).
(d) If, due to a negative loss adjustment, the adjusted loss amount of an event falls below the internal data collection threshold of the institution, the institution shall exclude the original loss amount from the range applicable before the adjustment (report the amount with a negative sign in the respective row), but not include the adjusted amount in another range.
The interpretation of these statements could be not so trivial. Therefore, some basic examples should be included to avoid different interpretations and inconsistent implementations among institutions. Some possible examples are included in the attached Excel file “UniCredit Response to EBA-CP-2016-20 - OpRisk Examples”. For each example, several solutions, corresponding to different interpretations of ITS, are reported. We suggest to include some examples with the correct solutions.
Yes, they are. UniCredit has no further comments to add on cost impact.
The inclusion of the new template C 17.02, related to the largest OpRisk loss incidents in the last year, is welcomed, under the condition that such type of template will be no longer included in the SSM Short-Term Exercise. Some further clarifications are needed to fully clarify the content of the new template.
The point 139 in the Annex 2 describes the selection of the largest loss event. Where it is stated “Further events shall be reported if the gross loss amount of those events is ≥ 10,000,000 €”, regarding the loss adjustments of an operational risk event previously reported, it should be specified if the gross loss amount is based on:
• the overall loss amount related to the operational risk event, or
• the amount of the loss adjustment.
The following clarifications should be provided for the content of required fields:
• Event ID: the column “Event ID” has the following definition “The event ID is a row identifier and shall be unique for each row in the table. It shall follow the numerical order 1, 2, 3, etc”. Apart from unicity, it is not fully clear which logics should be used to assign the ID. To avoid misunderstandings, it could be useful to specify that the Event ID should be the one used the OpRisk institution’s database.
• Date of accounting: some format should be specified for date of accounting and other dates, e.g. dd/mm/yyyy, or mm/dd/yyyy, etc.
• The same question raised above for the gross loss of events ≥ 10,000,000 € could be made for fields “Gross loss” and “Gross loss net of direct recoveries”, in all cases of loss adjustments of operational risk events previously reported.
• It should be clarified if the ranking of losses in the new template C 17.02, considering also the loss adjustments of operational risk event previously reported, is based on:
o the overall gross loss amount related to the operational risk event, or
o the overall gross loss amount net of direct recoveries related to the operational risk event, or
o the gross loss adjustment, or
o the gross loss adjustment net of direct recoveries.
• Legal Entity ID: it is specified that “LEI code of the legal entity as reported in column 025 of C 06.02 where the loss – or the greatest share of the loss, if several entities were affected – occurred”. It is not clear what if two or more legal entities have the same share of the loss.
• Business Unit: it is not specified if the name, or the ID, or both is/are required.
• Description: it should be specified if there is a max number of characters.
It is not clear if the point 132 in Annex 2, referred to C 17.01 template: “The figures reported in June of the respective year are interim figures, while the final figures are reported in December. Therefore the figures in June have a six-month reference period (i.e. from 1 January to 30 June of the calendar year) while the figures in December have a twelve-month reference period (i.e. from 1 January to 31 December of the calendar year).” can be confirmed also for C 17.02 template.
Regarding the three presented options (a), (b) and (c), in UniCredit opinion there is not so much difference in terms of cost and complexity, provided that the clarification requests above are answered.