Response to consultation on ITS amending Implementing Regulation (EU) No 680/2014 with regard to operational risk and sovereign exposures

Go back

Question 1: Could you please quantify both the implementation costs and recurring production costs (expressed in man days) that would arise when implementing the changed reporting requirements on OpRisk as part the regular reporting framework? How would these recurring production costs compare to a situation in which institutions were required to comply with ad-hoc data requests that are required to comply with current competent authorities’ requests on institutions’ OpRisk losses (e.g. SSM short-term exercise)? [see page 16]

The ITS apparently merge FINREP data and COREP data. In this context, it should be clarified whether primarily FINREP or COREP data requests are involved. Maintaining/linking accounting and reporting data would enlarge the resulting mix.
• Implementation calls for modification of reporting software. For institutions that use such software, implementation will therefore depend mainly on how quickly the responsible software firm can set up implementation.
• Generally speaking, the costs depend on how the reporting requirements are finalised and the size of the institution or whether it is a group, for example. Implementation costs: The data vary between 40 and 100 man days.
• Recurring production costs: The data vary between 1 and at least 10 man days per delivery/report.

Question 2: Could you please quantify the implementation costs (expressed in man days) that would arise when implementing the new reporting requirements on sovereign exposures as part the regular reporting framework? How would these implementation costs compare to a situation in which institutions were required to comply with ad-hoc data requests that are required (i) to comply with the EBA’s transparency exercises and (ii) to comply with competent authorities’ requests on institutions’ sovereign exposures (e.g. SSM short-term exercise)? [see page 17]

n/a

Question 3: The threshold defined in Article 5 (b) 3 (a) exempts institutions that fall short of the threshold from the new requirements. Do you think that this threshold is appropriate so that (i) institutions with material sovereign exposures are required to report (and hence supervisors will have the relevant information for their assessments) while (ii) smaller and less complex institutions are more likely to be exempt from the new reporting requirements? [see page 17]

• The threshold is appropriate.
• A further threshold geared to the materiality of non-domestic positions could be added. The ECB’s low-interest-rate policy means that, to enhance returns, banks are being forced to invest abroad. While a broad diversification of these positions increases the relative share of the sovereign portfolio, it punishes banks through the reporting burden. There should only be a reporting requirement for every single country in the geographical breakdown where positions exceed, for example, 1% of the sovereign portfolio.

Question 4: Is there a noteworthy difference in terms of costs between point (b) which requires a full country breakdown and point (c) which limits the breakdown to a total and domestic country? If there is a noteworthy difference, please try to quantify the cost difference and put it into context with the overall implementation costs that you expect with the new reporting requirements on sovereign exposures. [see page 17]

There is an additional burden particularly for reporting persons who have to complete and validate every report. Technically speaking, the information will always have to be collected to calculate the thresholds.

Question 5: Are the reporting templates related to sovereign exposures (C 33.01 and C 33.02) as set out in Annex I and related instructions in Annex II sufficiently clear? In case of uncertainties on what needs to be reported, please provide clear references to the respective columns/rows of a given template as well as specific examples that highlight the need for further clarifications. [see page 18]

• The reporting templates should show clearly which columns are to be completed by IFRS users and which by nGAAP users.
• Regarding the usage of FINREP values, e.g. in column 010, we assume that the values should be reported prior to substitution, whereas in column 320 RWA values are to be reported after substitution. From this perspective, this seems to be inconsistent.
• Regarding the disclosure of RWAs in column 320, it is not quite clear whether or not transfer risk should be included.
• We should like to point out the need for clarification with regard to Accumulated negative changes in fair value due to credit risk (C. 33.01, from row 160, columns 170-190) on the following points:
o Is the number relative or absolute (either accrued effects from the outset or accrued effects since, for example, the beginning of the year)?
o Is accrued PL of items that have been sold in the current year to be reported? In our view, the result depends on the model and should not be determined trivially, since, besides the credit spread effect, cash values could also be created by lapse of time, coupon payments and changes in FX and interest rates.
• General (C. 33.01, rows 170 & 220):
o A split into a standard and internal model approach does not appear appropriate to us, as application of partial use may result in multiple entries. This may, for example, be the case for interest rate risk if a bond is subject to both general and specific interest rate risk and the related own funds requirements are calculated using different approaches.

Question 6: Are the reporting templates related to OpRisk losses (C 17.01 and C 17.02) as set out in Annex I and the related instructions in Annex II sufficiently clear? In case of uncertainties on what needs to be reported, please provide clear references to the respective columns/rows of a given template as well as specific examples that highlight the need for further clarifications. [see page 19]

Currently, the reporting period is always the previous twelve months, so if the submission is in January 2016, the data would be from Q4 2014 to Q4 2015; in July 2016, the data would be from Q2 2015 to Q2 2016. Would the same kind of reporting period be used for the updated template? And, indeed, what is considered to be the “previous reporting period”? Is that referring to the previous submission?

In general, we should appreciate it if the following differentiation with regard to the terms “reporting period” and “previous reporting period” in both types of reporting were made for both templates in Annex II:

Type of COREP reporting Reporting period Previous reporting period
half-year report current half-year previous year-end
year-end report current full year previous year-end

Some specifications are still missing at detailed level and hence should be clarified with regard to the following aspects:
• Number of events (e.g. row 010): Unlike the old instructions, the number contains not only events accounted for the first time within the reporting period but all events for which gross losses were accounted within the reporting period. In this context, it is not clear whether these figures include the “Number of events subject to loss adjustments”, which is delivered in a separate row (e.g. row 030).
• The same goes for rows 910-914, where it is not clear whether these figures include the events subject to loss adjustments and whether, for those events, the same determinations as in rows 931-934 should be applied.
• In cases where the business line of an event changes and the loss amount remains unchanged, it is not totally clear whether this adjustment should be reported at all (e.g. the business line of an existing loss changes from CORPORATE FINANCE to TRADING AND SALES; the loss amount of EUR 1m remains unchanged).
• The explanatory text for consultation purposes says that the largest losses from the last year should be reported. This information is also to be included in chapter 4.2.3.1 of Annex II. We would propose the following approach: events that are either “accounted for the first time” within the reporting period or “accounted for the first time” within a previous reporting period should be included if the event has not been included in any previous supervisory report. We are firmly convinced that it is in order to deliver the description of the event in German.
• Legal entity name/ ID (template C 17.02, columns 170 and 180): As we understand it, the legal entity name and ID refer to the legal entity of the bank where the loss occurred and not to the counterparty of the bank. Moreover, it is not clear which cases have to be included. Are these to be the nine largest cases in the reporting period or also cases from previous periods that become relevant for the nine largest cases through a gross loss increase? There is also the question of whether these are then to be taken into account by way of the aggregate gross loss or only by way of the “loss adjustment”?
• In 17.02, what would be considered as a sufficient description? Why is this required? Sometimes the descriptions in IT systems are very brief, delivering only limited added value.
• Is there a threshold for loss adjustments? – i.e. if a loss has been adjusted by EUR 100, would this loss need to be reported? Or is there a more significant threshold that would need to be considered?
• Please confirm that we should not be reporting any gains or losses below EUR 10,000.

Further details of the individual points for the changes in Template C 17.01, together with examples, would help to provide an even better understanding of the supervisory approach outlined.

The requirements as to which events need to be included in the rows “total direct loss recovery” and “total recoveries from insurance and other risk transfer mechanisms” regarding the incorporation of new or changed recoveries are unclear in Annex II. The consultation paper says that this information is collected independently from when the original event occurred. In our opinion, the amount of recoveries should refer to operational risk events accounted for the first time within the reporting period and accounted for the first time within a previous reporting period if the event has not been included in any previous supervisory report. Additionally, adjustments (positive or negative) should also be incorporated (negative values are possible in this case).

When “risk transfer mechanisms” are referred to in template 17.01, what does this mean?

The definition of the “maximum single loss” is included in paragraph 128 (chapter 4.2.1 of Annex II). We assume that a new paragraph should be opened up.

In addition, further questions arise in connection with the “date of accounting” in Annex II, paragraph 123. Is this to be understood as the cut-off date for accruals in reporting? This raises the question of how historical cases that have not yet been assigned any “date of accounting” are to be treated. This may, for example, be the case for migration of data from subsidiaries that have not recorded any booking date. As subsequently recording the date of accounting is highly burdensome, we suggest using the “date of discovery” instead. In this context, we also suggest changing the “date of occurrence” to the “date of first occurrence”, since a case may occur several times before it is noticed, e.g. time clock fraud.

We should also welcome it if examples were added to the explanations so that the approach with regard to special cases becomes clearer, e.g. for losses that do not trigger any booking such as funding losses or losses recorded in the trading P&L account, the approach in the case of differences between book value and attributable value (e.g. where buildings and works of art are involved), the approach to legal cases (aggregate loss revocation) where legal action is filed by many individual clients over a period of several years, and the handling of reserves for legal cases that become known in the first half of the year but are only booked at the end of the year.

Question 7: Are the rules for the assignment of loss adjustments to ranges as defined for rows 940 to 944 sufficiently clear? In case of uncertainties, please provide suggestions to improve the clarity and/or effectiveness of the reporting instructions for loss adjustments. [see Annex II, page 7]

The rules for the assignment of loss adjustments to ranges as defined are sufficiently clear.

Question 8: Are the new rules for the determination of the number of loss events subject to loss adjustments for certain ranges of gross loss amounts as defined for rows 931 – 934 and the rules for the assignment of loss adjustments to ranges as defined for rows 940 to 944 appropriate in terms of cost/benefit? Please try to quantify the cost impact and put it into context with the overall implementation costs that you expect with the changed reporting requirements on OpRisk. [see Annex II, page 7]

The requirements as described today regarding rows 931-934 and 941-944 require manual adjustments in every report because of the double counting of events if the loss amount falls into another range. The effort required for manual adjustment is far too great compared to the outcome.
The rules regarding the number of loss events subject to loss adjustments are quite complicated in cases where the size range changes. A less complex rule would be sufficient, e.g. to report the event only in the range before the adjustment instead of reporting this event in both the old and the new range.
Alternative:
In terms of cost/benefit, rows 930 – 934; if there is still a justified need to count the adjustments in the different ranges, we suggest the following approach, with (a) and (b) remaining unchanged (change log visible):
• (c) If the loss adjustment entails a change in the range, institutions should report this as one event in the row dedicated to the range applicable after the adjustment.
• (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 or below EUR 10,000, the institution should report this as one event in row 930.

The same goes for rows 940-944: “loss adjustments relating to previous reporting periods”.
If there is still a justified need to assign adjustments to ranges, we would suggest the following approach, with (a) and (b) remaining unchanged (change log visible):
• (c) If the loss adjustment (negative or positive) entails a change in the range, institutions should report the adjustment in the range applicable after the adjustment. The fields may contain negative numbers.
• (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 or below EUR 10,000, the institution should report the adjustment in row 940.

The COREP templates are expanded relatively frequently, so that we would welcome it if a reduction in their size were also considered. For example, we regard the breakdown of OpRisk losses by business line as unnecessary. Both the Basel Committee and European supervisors have concluded that these say nothing whatsoever about loss frequency. The treatment of OpRisk business lines is based on the standardised approach and the AMA (own funds requirements for each business line); both are to be abolished, so that business lines will no longer play any role in future. The breakdown of losses by business line is in any case often arbitrary or differs according to an institution’s organisational structure.

We should also welcome deletion of “GROSS LOSS BY BUSINESS LINE“ row, since, for one thing, losses often cannot be assigned accurately and, for another, the information collected does not allow any really meaningful conclusions.

Question 9: Which option as regards the threshold for OpRisk loss events is the least complex or least costly in terms of implementation? [see Annex II, page 8]

Of the considered options, we recommend proposal b).

Name of organisation

Die Deutsche Kreditwirtschaft (DK) // The German Banking Industry Committee