Primary tabs

European Savings and Retail Banking Group (ESBG)

Members are of the opinion that the additional information embedded in the upside potential might be useful to determine the level of prudence which is already included in the fair value of assets and liabilities. This additional information supports the interpretation of the total aggregated AVA and those institutions which value their portfolio closer to the 90% confidence level will be able to argue a smaller AVA impact that institutions valuing their portfolios closer to the 10% confidence level.
Concerns have been expressed that assessing the appropriateness of the reported downside potential of the AVA might only be possible for those AVAs where the range of plausible data can be retrieved. Also, the Regulatory Technical Standards (RTS) does not state for which AVAs the up-side potential should be calculated. The efforts required to set up the additional confidence calculation is expected to be burdensome and doubts have been raised amongst members whether it is appropriate to so significantly extend the reporting scope beyond what is included in the RTS by introducing a new concept. This introduction may cause the following problems:
• The nature of the complex Expert approaches in place for the separate Pru Val AVA does not facilitate a simple tweak to obtain the flip upside of the distribution. Nearly all AVAs would require extensive methodologies, and thus a standalone process alongside Pru Val, to obtain this value.
• The implementation of this concept would result in significant additional administrative costs.
• It is also unclear why a new concept is being introduced outside the scope of the original RTS at this stage, and via the ITS rather than the RTS.
• The argument for this new requirement is also challenged as Pru Val is a prudency concept which builds on the conservative principles of Fair Value accounting and IFRS 13. It seeks to protect banks from unexpected valuation shortfalls via a capital reserve, therefore, it is hard to see where in this context Upside Uncertainty fits. Will there be a capital saving?
• The applicability of an Upside Uncertainty to most of the AVAs, is also questionable, and would undoubtedly lead to differing interpretations and methodologies across institutions rendering the measure inconsistent and thus reducing its benefit.
The proposed definition appears suitable for a general definition but lacks clarification on how to apply it to the different AVA categories. The approach suggests showing the upside potential for the total aggregated sum of AVAs. As individual AVAs might be derived by using expert opinions it needs to be clarified how the expert based downside potential can be translated into an upside potential, especially for AVAs like “future administration costs” or “early termination costs”. In addition, some institutions do not calculate an adjustment for certain AVAs (e.g. Operational Risk AVA for those institutions applying the AMA model).
As mentioned under Q1, members do not believe it to be appropriate to so significantly extend the reporting scope beyond what is included in the RTS. Hence, they are sceptical about the inclusion of this definition in the reporting standards or if it were to be included it is proposed that the up-side potential is limited to those AVAs where a plausible range of data (market prices, net present values, etc.) can be retrieved and a confidence level of 90% and 10% can be calculated. Some members have raised the concern that this could also lead to potential risks on individual interpretations between reporting banks thus not supporting uniform requirements within the EU.
Splitting fair value assets and fair value liabilities between asset and product classes is not practical to implement, due to comparability issues between risk and accounting information. Also, Trading Book and Banking Book is not an accounting concept and hence the implication of having to present the information on this level will be most cumbersome.
Given the fundamental differences in the accounting and internal management framework (upon which Pru Val is currently built) it will be very problematic to provide Fair Value data beyond Group/accounting totals. Breakdown per asset classes or portfolios as required by template C32.02 are currently not supported by the accounting set up which stores the Fair Value data. In the case where the valuation input for calculating the Market Price Uncertainty AVA for a valuation position is a market price the split into the asset classes might not be possible. The following example illustrates this:
Example: The institution assesses the market price uncertainty for a bond portfolio on the basis of a range of available market observable price quotations. As for a price range below, how should an institution split the market price uncertainty between the asset classes Rates – Vanilla and Credit – Vanilla?
(Please see attachment for graphic)
The proposed attachment of the position to the portfolio where the instrument is booked is not always clear and members would appreciate a more explicit separation by product type.
The requirement set out in the ITS requires a fundamental change to align the accounting set up with internal reporting tools such as Risk management. This has an impact well beyond the remit of Pru Val. The cost would be considerable and in our view far outweigh the benefits. Some additional clarification is sought on the exact legal entities and expectations in terms of matching the Fair Values submitted via the Pru Val population being tested/in scope.
Furthermore the requirement to combine a capital adjustment concept, Prudential Filters with Fair Value Balance Sheet items does not take into account the differences between these diverging concepts.
As commented on above, there are significant obstacles to reporting AVAs per asset class and product category. In the risk based approach used for the RTS, valuation exposures are not in a direct relationship to the portfolio based product categories and the level of aggregation and net-ting is so significantly different that matching valuation exposures to fair value reporting is a significant obstacle. Consequently, it will also be challenging to allocate IPV differences to AVA categories.
The split between vanilla and exotic seems suitable. The portfolio split however would indicate that all products exposed to the proposed risk categories, could be aggregated into a portfolio, this portfolio based approach is not appropriate as the RTS supports a risk based approach when calculating in particular the market price and Close out Cost AVAs, which is completely contradicted by the portfolio approach described in Annex2.
In order to be able to assess the accuracy of the reported AVA values the fair values would have to be split to these portfolios as well. If the portfolio definition is neither instrument based nor risk based the split seems unpractical for IPV and Fair Value Reporting purposes. Members suggest that the reporting requirements should match the RTS, and that the AVAs should be split on a risk or a product basis. For which the stipulated asset and product classes are still relevant.
This is probably institution specific, the level at which netting is performed impacts the netting benefit and hence will impact the size of the mismatch. In order to be able to assess the difference between the portfolio AVA and the institution level AVA additional clarification on the portfolio term is needed, especially with respect to those AVA categories which can’t be easily attached to individual portfolios.
Given the guidance on page 10 in Annex 2, members would not expect a big mismatch between the portfolio and institution level AVA. Subject to working on a Risk based portfolio definition this is a realistic requirement, it will however have resourcing implications as any additional reporting sets create additional computations and the proposed split is not beneficial for the internal steering and capital allocation to individual trading desks.
Along with our answer to the previous question as stated in Q4 above the RTS supports a risk based approach, and stipulates that the total category AVA should be calculated by summing up the individual uncertainty AVAs, which in turn should be calculated on valuation exposure level. Valuation exposure is not simultaneous to portfolio level.
The size of the difference is somewhat irrelevant as the challenge is the extra work required to undertake an additional set of calculations.
The description of a model could be usefully summarised in a few key words although some details may have to be left out. In terms of categorisation, the main keywords should be enough to assist with categorisation and mapping. Members believe it will be a challenge to reach uniform reporting requirements, if each institute is free to set descriptions for the main features of models and products. Therefore it is suggested that the EBA sets out those features in order to align the reporting.
Meaningfully summarising the description of a product in a few words without compromising on accuracy may not always be. In the case of exotic and structured products, the complexity of the payoffs would, in most cases, require many more than a few keywords to be able to unambiguously differentiate between the different products in a way that would assist in a further categorisation/mapping and aggregation by the authorities. More detailed guidance would have to be provided to the institutions after further consultation with them and a final solution may require the introduction of more than one cell for the adequate description of all products, especially if the format has to maintain some forward looking compatibility/expandability as more new products with increasing complexity are introduced by financial institutions as time goes by.
Whilst members consider the requested information to be quite fair, and our members feel that they should be able to answers these kinds of questions, there is a considerable amount of work required to develop the existing set up to provide the additional information in an accurate, timely and efficient manner and to specifically provide the requested breakdown. Developments would be required in Model Risk as well as in the IPV process. This costs incurred in this would be substantial and not necessarily one off.
The most challenging fields are Observability, IPV Difference and IPV Coverage. Observability, for example, would require collection and aggregation of continuous intra-day communications (emails, telephone conversations, and meetings) of many traders and salespeople with brokers and counterparties in order to be able to confidently provide accurate information about the frequency of qualifying price observations for hundreds of products. It would require the creation of an application that would archive hundreds of emails and conversations that take place every day be-tween hundreds of front office personnel. Concerns have been raised by members about the lack of a name for the product group. The lack of guidance with regard to naming convention/model explanation will limit the usefulness of this information.
In general it can be seen that the proposed instructions follow the purpose with an alignment throughout the European Union Members. However, there are still many question marks over the proposals and a number of unclear areas, mainly when it comes to introducing new concepts and extending the scope of Prudential Valuation. Examples of those areas are:
• Split by Trading Book – Definitions of “Portfolio Based Allocation” are unclear.
• Concentration Risk AVA: Although the RTS and the DA clearly states when a valuation position is concentrated, it lacks guidance on how the concentration Risk AVA should be calculated in a second step. As the proposed COREP template aggregates valuation position on portfolio level, should the concentration risk be tested on portfolio level as well?
• Unearned credit spreads: the RTS as well as the DA lack clarification on how to calculate the valuation adjustment. The split into market price uncertainty, close-out costs and Model Risk increases the complexity and requires further details on the expected valuation adjustment.
• Investing and Funding Costs: The same obstacle as detailed above for unearned credit spreads.
• It is unclear whether Unearned Credit Spreads & Inv & Funding Cost only reported in row 200 & 210 (of C32.02) i.e. not allocated out at all.
• Exotic/Illiquid – There is a clear definition, however, it is yet to be confirmed whether existing categorisations complies with the definition. Thus changes may be required.
• Is the Portfolio reporting split required for all AVA’s as implied by the s/s?
• Upside uncertainty which is a new concept and not included in the RTS for Prudent Valuation.
• Over hedges which is not a general concept in Fair Value accounting nor in the RTS for Prudent Valuation.
• The reporting requirement of splitting the fair value adjustments between assets and product classes may be directly inappropriate.
• The focus on a portfolio based approach is not appropriate as the RTS for Prudent Valuation supports a Risk based approach.

The individual AVAs leave, in general, substantial room for interpretations which will be limited by the proposed reporting templates. Members would appreciate clearer guidance on the calculation approach for the individual AVA categories.
The application of the simplified approach for smaller institutions, which fall below the thresholds on individual level but not on consolidated level as outlined in Article 4(3) of the Delegated Act, is not fully clear. Feedback would be appreciated on the individual treatment of institutions falling below the threshold for applying the simplified approach. Is the institution allowed to apply the simplified approach for the individual calculation of Prudent Valuation or does the core approach have to be applied on individual entity level as well? If it is the case that the core approach is required for the individual calculation as well, the operational burden for consolidated banking groups, with a high number of fully consolidated smaller institutions, is extremely challenging.
Article 4(1) also states that the threshold has to be tested under the applicable accounting frame-work. In a case where an institution reports capital figures according to local GAAP on an individual basis and according to IFRS on consolidated basis, which accounting framework is subject to the threshold of EUR 15 billion? Which approach has to be applied on individual and consolidated basis in the following example?
Group Consolidated Fair Value Assets and Liabilities under the IFRS framework are above the threshold of EUR 15 billion. The group reports capital figures according to IFRS. Individual entities report their capital figures according to local GAAP; 99% of those entities are below the threshold. There is no consolidated statement under local GAAP. Therefore there is no breach of the threshold according to local GAAP. Which approach is applicable for the individual entities (under local GAAP) and which approach is applicable for the consolidated entity (under IFRS)?
The proposed templates post a high burden on one-off IT requirements and in addition to that ongoing operational costs. The expected changes in accounting treatment due to IFRS 9 should be incorporated or at least considered in the COREP Templates for Prudent Valuations.
The reporting burden would be substantially more that it is now. Financial institutions would be required to invest significant amounts of money and resources to introduce all of these additional reporting dimensions in an accurate and efficient way. Consultants would be required to assist with the new reporting/IT architectures as the contributions for the different fields would come from different parts of the institutions (Front Office, Accounting, Risk Management, Validation Groups, Mid-Office etc.) and new applications would have to be developed for this reason, the implementation and testing of which will last several months. Finally many other application (re-porting systems) would require upgrading to be able to provide the additional functionality as well as to maintain the integration with the IT infrastructure of the institutions.
The associated costs with implementing and maintaining such a comprehensive reporting template should only be compulsory for institutions above a threshold which goes beyond the EUR 15 billion thresholds for calculating the simplified approach. With this in mind, members have suggested an additional threshold based on the total aggregated AVA amount.
Seán Doyle
E