We are concerned that some of the proposals in the consultation documents go beyond what would be reasonably required for “an understanding of where the accounting fair value sits within the notional range of plausible values at an aggregate level”. In particular, we are not in favour of the proposed calculation of upside uncertainty (column 120 of C 32.02) as this would in many cases at least double the workload undertaken by institutions for the purposes of estimating uncertainty. At the same time, upside uncertainty is not relevant to Prudent Valuation, which is precisely a “prudent”, downside measure of uncertainty. Furthermore, in practical terms, estimating upside uncertainty requires a significant amount of inputs and methodologies which would not otherwise be used in the context of prudent valuation. We therefore disagree with upside uncertainty being “essential context” for the downside of the plausible range, and the administrative costs associated with this option do not seem to be outweighed by possible benefits.
As mentioned above, in our response to Question 1 of the consultation, we do not believe that any inclusion of upside uncertainty would pass the cost-benefit analysis. We note that such a calculation is not required by the prudent valuation RTS. We would therefore propose, on the basis of the administrative costs that the requirement would create, the exclusion of the reference of upside uncertainty from the revised EBA reporting templates for prudent valuation. It should also be noted that resources utilised in calculating an upside uncertainty with the same rigour as downside would shift the focus of an institution’s resources that are currently set towards AVA calculations.
In respect of our answer to Questions 1 and 2 above, we would also highlight the difficulty we and others would have in providing the data required in a short timeframe. The uncertainty of the effect on banks’ ability to produce the information mentioned in the consultation is also compounded by the brief consultation period on the proposals.
We support in principle the consultation’s proposal to disclose fair value adjustments.
We are concerned however with the proposed breakdown, for example, of assets and liabilities by portfolios. We note that the reporting of most institutions is not built at this portfolio level and that therefore the approach suggested in the consultation document would add an additional layer of complexity which would go beyond what is required for the purposes of accounting valuation. This level of breakdown is also not implied by the Prudent Valuation RTS.
As an alternative, we suggest that the split be done at a business level, which would thus align the requirements with information that is used by firms for the purposes of their internal/management reporting.
We are also concerned with the list of required items including information on QTD Revenue for portfolios (in Column 150). We would not consider the revenue attributed to portfolios to which the fair value positions were assigned to be useful in meeting the objectives described in the explanatory note preceding question 3. The case has also not been made for why this would be relevant for valuation uncertainty, and we would therefore welcome the removal of this requirement from the template.
As explained above, there are practical difficulties to implementing the portfolio approach. In our view, the split should instead follow the business organisations and the way activities are reported (this would allow the alignment of supervisory reporting rules to the business activities performed). We note that this way of proceeding also appears to be more aligned with the approach adopted in the forthcoming Fundamental Review of the Trading Book. The portfolio-based approach would otherwise be unduly onerous on market participants and would not mirror any of the existing reporting processes in use for either regulatory or internal purposes. Similarly the requirement to report the Fair Value amounts for such portfolios is not currently possible at some institutions due to the different basis by which risk and financial data is collated. We would therefore welcome the EBA clarifying whether the intention of the wording on the requirements for Rows 040-160 (page 10 of Annex 2 to the consultation) referring to a “portfolio based allocation” is compatible with the way in which most businesses would currently report information on AVAs. If a portfolio approach is retained, at a minimum, we would welcome that it be clarified that firms have the flexibility and ability to exercise judgment in allocating their internal reporting breakdowns to the proposed regulatory categories.
Moreover, the split between the trading and non-trading books would also cause difficulties for most institutions. Requiring institutions to collate information for more levels of reporting would also lead to additional calculations which do not seem aligned with the initial intention of the RTS. The proposed definition for the non-trading books risks capturing products on trading desks which are seen from an IPV perspective to be a normal part of trading and would not need separating out for any other purposes. This would require additional work as the separation is not built in to existing systems. Stripping out the netting benefit is also likely to require calculations being performed on a group netted basis and on a portfolio basis, with a corresponding impact on banks’ resources. The final ITS should therefore clarify that banking book products that are part of a trading or capital market activity or offering do not need to be separated out into the non-trading book category.
On the question of the mismatches between portfolio-level and institution-level AVAs, we note that the level of mismatch would depend on how much offsetting valuation exposures existed between the different portfolios.
The effort required to perform the calculations increases significantly with additional splits of information required. Risks/valuation exposures should be able to be netted at the institution level for the AVA calculations. The significant extra effort required to perform these calculations for additional splits is the same regardless of whether the benefit is large or small since institutions would have to build and operate the additional calculation processes.
We note that column 050 requires a “Description of the main features of the valuation model that is used to derive the accounting fair value. This should include a description of any model-related Fair Value Adjustments where relevant”. While we welcome a standardised approach to the length of the description, and we agree that this could improve comparability across institutions, we note that only the basic features of the model could be explained in 60 characters.
In relation to the clarity of the proposed prudent valuation instructions, we would like to reiterate some of the points we have raised earlier in our response. For example the way in which trading book portfolios are explained for rows 040-160 would seem to preclude using businesses/desks as the driver for categorisation (to give an example, Structured Rates should be reported in Rates Exotics and Flow Rates in Rates Vanilla). We also note that, due to the short consultation period it is likely that we and others will have been unable to identify potential additional issues that may exist.
In answer to Question 9 of the consultation, we would reiterate our introductory comments to this response letter, highlighting the difficulty we had in analysing the proposed templates in detail given the short timeframe for the consultation. Our initial analysis did however raise a number of inconsistencies, some of which are mentioned below:
- The templates seem to be unduly onerous and not focused on the “essential context” required by the objective of capturing information which relates to reporting prudent valuation information.
- We are concerned with the requirements regarding the model risk and concentration AVA templates, and in particular with some of the wording in paragraph 4.1 of Annex 2 to the consultation document (requiring “report details of the top 50 individual model risk AVAs”). As an alternative, we would suggest either applying a materiality filter to the requirement (e.g. 1% of total AVAs), or restricting the detailed information to the top AVAs needed to ensure sufficient coverage of the model risk. A similar approach could also be used for reporting concentrated position AVAs.
- We agree that providing certain model related metrics alongside the model risk AVAs may be useful with respect to Prudent Valuation. We do not, however, think that it is natural to allocate all of the proposed metrics to a given model and are concerned that this will lead to confusion and inconsistent interpretation. For example, a given trade can reference a number of different models so the allocation of that MTM/notional between the models may be subjective. We also believe that disclosures around fair value adjustments and IPV should be limited specifically to those balances relating to model risk.