Response to consultation on ITS amending Commission Implementing Regulation EU 2016-2070 on Benchmarking

Go back

Do you agree that the combined split of rating and country in template C103 can generally be replaced by a simpler rating split per model (i.e., rating distribution) in template 105, which will cover all models in the scope of the benchmarking exercise (HDP and LDP) without losing explanatory information on the variability of benchmarking parameters? Is there any data point collected in the new template 105.04 that involve significant IT costs or burden and should be dropped?

We agree that the combined split of rating and country can be dropped without any loss of information. This applies especially to institutions that leave a lot of portfolios blank.

Although the costs associated with data collection can be considered manageable, the introduction of new templates always generates significant IT costs.

Do you agree that SLE portfolios should be reported in a separate exposure class? Do you agree that the proposed level-2 breakdown on (a) the proposed sectors of counterparties and (b) the proposed types of exposures (i.e. categories of specialized lending) might be relevant components to explain the variability of risk parameters? Which option do you prefer with respect to the rating split under the slotting approach?

We agree that SLE portfolios should be reported in a separate exposure class.

We also agree that the proposed level-2 breakdown might be a relevant component to explain the variability of risk parameters.

Do you expect that the LDP sub-portfolio characterized by eligible covered bods will cover a material share of exposure? Do you expect that the separation of these exposures can contribute to explain RWA variability?

In some institutions, covered bonds already cover a material share, and that would explain why, for example, LGD raises questions especially at consolidated level.

Do you think the alternative portfolio split would provide for a higher explanatory power as regards RWA variability induced by differences in CRM usage?

We think a homogeneous split in terms of collateralisation has an additional explanatory power as regards RWA variability.

Do you expect that the proposed NACE Code breakdown for HDP sub-portfolios will provide more explanation for RWA variability for a material share of exposure? Do you expect that the separation of these exposures can contribute to explain RWA variability in the according HDP portfolios or do you consider the current split using only NACE code F sufficient? Does the selection of a subset of NACE codes significantly reduce the burden of the data collection (compared to a comprehensive collection of all NACE codes)?

Generally speaking, the burden of data collection for a subset of NACE codes is lower than for a comprehensive collection of all NACE codes. A subset of NACE codes would reduce the amount of data to be collected, though in previous exercises institutions had to report all other NACE codes other than F under 'Not applicable', so a real reduction can only be achieved if the NACE codes are specified generally.

In all respects, the burden of reporting all NACE codes separately would be nearly the same as at present.

Some institutions have a proper distribution in many NACE codes, so more information could be gained from a comprehensive collection of all NACE codes.

Do you expect that the proposed ILTV buckets for HDP sub-portfolios secured by immovable property will provide more explanation for RWA variability for a material share of exposure? Do you expect that the separation of these exposures can contribute to explain RWA variability in the according HDP portfolios?

We do not expect better explanation of RWA variability for a material share of exposures (i.e. real estate portfolio transactions: LTVs are calculated at the property level, but the impact on the RWA of the portfolio transaction depends on their share of the total exposure).

Do you agree with the Additional pricing information requested? Please, provided detailed explanation for your answer.

The additional pricing information requested in the consultation paper is inappropriate, in our view, in the form proposed by the EBA. According to the consultation paper, the sensitivities of all risk factors used and, in some cases, even the second derivative of the risk factors used are to be determined and reported. Collecting sensitivities along with IMV for trades subject to internal model treatment may not achieve the stated objective of helping competent authorities verify that the instruments were correctly interpreted (cf. paragraph 33 on page 17 of the consultation paper). The main reasons for this assessment are:

 The market data setups and, in particular, the selection of specific risk factors differ significantly at individual institution level. For example, differences in the number and choice of yield curve grid points or significant volatility surface differences are likely. There will, moreover, be further differences (e.g. definition in the base curve setup). Even where the actual risk factor choices agree, naming conventions will not.
 Institutions may currently not calculate all the requested sensitivities for internal purposes (e.g. when utilizing a full revaluation framework in the internal model or when excluding aging effects from it), and generating these sensitivities ad hoc for hypothetical trades within one week after the booking date appears to be very challenging operationally. The current proposal is likely to direct significant resources from ensuring correct trade booking to verification of numbers (i.e. sensitivities) that may not feed the internal model in the end.
Hence, competent authorities will likely find it challenging to consolidate sensitivities collected from different institutions. This will mean that the sensitivity data provided by institutions will not be comparable. Because of this lack of comparability, we believe that the request for such information would be of little use.
There are several more general options that are more likely to achieve the supervisory goal of ensuring a better understanding of trade specifications. Some or all of them may be combined:
 Option 1: Keep the trade universe stable over time and remove particularly complex trades (e.g. trades 18 and 23) completely. That will ensure that most of the misunderstandings will be fixed once a specific trade has been part of at least two benchmarking cycles. In this regard, the industry welcomes the approach to not significantly change the trade universe for the 2020 exercise.
 Option 2: Specify the trades for the hypothetical portfolio exercise in a standardised (e.g. ISDA compliant) term sheet format, potentially with the support of industry experts, down to a level of detail required for a legally binding definition of trades.
 Option 3: Dropping the second derivative and, instead of risk factor sensitivities, capturing more general sensitivities. For example, a 1 BP parallel shift in the entire yield curve or 1% shift in the entire volatility surface could be reported (values could be reported by currency).
 Option 4: Collect sensitivities specified as per Standardised Approach under FRTB instead of sensitivities calculated for the internal model and standardise the format. This would address the challenge of absence of standardisation to a large extent but, given regulatory timelines, this must lead to a later collection of sensitivities, not before the capital requirement for the FRTB standardized approach is being introduced. As the standardisation itself will require significant effort (ISDA Common Risk Interchange Format may be a starting point).
 Option 5: Increase time between IMV reference date and IMV remittance date to at least four weeks to allow for quality assurance of the significantly increased number of values to be verified.
Of these, the industry considers a combination of Option 1 and Option 4 most advisable. Option 3 itself does not address the core issue but would at least allow avoiding additional complexity due to operational errors.

Do you agree with the simplification introduced in the time setting of the references date for the instruments?

Yes.

Do you have any concerns on the clarity of the instructions?

Some of the instrument-specific instructions can be made less ambiguous. Detailed suggestions are given in our response to Question 12. Specification of term sheets for all trades would most likely remove all ambiguities.
We note that instruction (kk) in Annex V appears to apply to instruments “52 to 67 and 69” instead of instruments “52 to 73”.

Can you please provided detailed explanation of the instruments that are not clear and a way to clarify the description?

Instruments 9-16: For options expiring in December the expiry date is the end of December, whereas, in line with the market standard, the expiry date for options expiring in June is the third Friday of the month: this is inconsistent. Suggestion: align treatment of December and June options and change expiry date from “End of December Year T” to “December Year T”.
Instrument 38/39: “Short 6-month EUR/USD (or EUR/GBP respectively) forward contract” is misleading. Direction of a forward contract should be defined by the currency exchange rate. Suggestion: remove words “long/short” for forward contracts.
Instrument 40: “Long 1 MLN USD at the EUR/USD ECB reference spot rate” is misleading. Suggestion: change description to “Long 1 MLN USD Cash”.

Upload files

Name of organisation

Association of German Banks