• Under IPV, usually the price verification with regard to the fair value as recognized in the financial accounts is understood. Since the aim of the prudent valuation differs from that of the IPV process, it appears unreasonable to request the use of the same data, in case this data is partially not appropriate, for instances in cases where the IPV relies on point estimates while the prudent valuation requires a range estimate that is not available from the particular data source. Consequently, we suggest amending the wording of Article 3 (1) as follows:
”Where institution calculate AVAs based on market data, they [shall] should take account of market data used in the independent price verification (‘IPV’) process of Article 105 (8) of Regulation (EU) 575/2013 subject to the adjustments described in this article, unless institutions can explain that this is not justified”.
• Some data mentioned in Article 3 (2) might be overly costly to obtain and a cost/benefit analysis may not always justify the usage of all data available or listed in Article 3 (2). We suggest changing the wording in Article 3 (2) as follows:
“The market data used to determine a prudent value shall include a [full] range of available and reliable data sources, [including all of the] among the following:
(a) Exchange prices in a liquid market
(b) Trades in the exact same…
The number of data sources used should be proportionate and related to market data observability and uncertainties.
• We suggest amending the wording of Article 3 (3) by deleting the word “all” and adding a reference to the proportionality principle.
• Yes, we believe such a simplified approach is helpful and its application should be subject to a risk sensitive threshold.
• The threshold appears to be overly conservative in view of large institutions with relatively small positions of fair valued instruments. We believe that the threshold could accommodate for these cases as follows: “If the sum of the absolute value of on- and off-balance-sheet fair valued assets and liabilities exceed € X bn and Y % of total assets or exceed € Z bn the institution applies the core approach” (with Z much larger than X).
• In addition, for the purpose of deriving the gross amount to be compared to the threshold we suggest considering the exclusion of positions with clearly negligible valuation risk. For instance, it would be advisable to take account of exclusions provided in Article 8 (2).
• The time period granted to implement the core approach of two quarters according to Article 4 (2) could prove challenging, unless it is the time limit set to submit the implementation plan.
• It is our understanding that no “partial use” shall be allowed on group level. In this case there are practical issues.
• Applying the core approach on a group level would require a subsidiary that is eligible to the simplified approach to implement both approaches or would force it into the core approach.
• Please note that in some circumstances the consolidation of institutions may differ under the regulatory and the accounting frameworks and different GAAPs apply on entity and group level. In such cases, this will pose additional operational burden.
• Allowing a partial use on group level would solve this issue. In order to reduce implementation cost we suggest considering the possibility of a simple aggregation rule of AVAs on group level for all or elected subsidiary institutions taking account of intra-group deals, unless a group choses to apply the core approach commonly throughout the group.
• The risk sensitivity of the unrealized gains is doubtful since it is a historical number. For instance, the unrealized profit could be locked-in by risk off-setting trades.
• In addition, the lifetime unrealized gain of a fair valued financial instrument is not required to be collected for other purposes in case fair value changes are recognized in profit or loss. Hence, institutions may incur large implementation costs for deriving a number that is not risk sensitive. Aged trades and positions with high turnover would complicate matters too. We recommend considering a metric that is derived in course of regulatory or financial reporting.
• The fair value of a financial instrument is risk sensitive and subject to the AVAs. In general, we would regard this value as an appropriate measure of valuation risk.
• We suggest clarifying the meaning of “matching, off-setting”. In the Q&A along the QIS it is stated that back-to-back transactions could fall in this category.
• The metric should not be enriched with complexity in first place. In case the unrealized profit is disregarded and the institutions could use categories of liquidity that have to be determined under CRR/CRD IV or IFRS anyway, the currently envisaged approach could be improved.
• We interpret Article 7 (4) that institutions need to calculate the AVAs according to Article 7 if it is not possible for just one single AVA being calculated according to Articles 8 to 16. The fallback requirement appears to be very conservative and would require an additional calculation approach as the one under the simplified approach (see comments above). We propose to consider the expert judgment approach for all AVAs as a primary fallback approach since it would force institutions to pay dedicated attention to illiquid or complex trades or exposures. Alternatively, we propose to consider a fallback solution that uses a combination between AVAs that can be calculated and those AVAs that cannot. For instance, if a particular AVA cannot be calculated the institution might be required to set an AVA to X % of the particular AVAs of similar instruments, with X > 100 (in proportion to fair value or notional).
• The “volatility” metric defined in Articles 8 (4.b.3) and 9 (5.b.3) is quite hands-on and limited. We believe that a principle based formulation improves tractability and appropriateness, for instance that variation of fair value shall not be materially underestimated. This would also allow institutions to apply techniques they have already in place.
• It does not become clear from the example how the volatility criterion is applied according to the respective definitions of Articles 8 (4.b.3) and 9 (5.b.3).
• In the example the valuation inputs and the exposures are mapped to a smaller set. The largest impact of inaccuracy typically comes from exposure mapping since the yield curve can to a large extent be described by three risk drivers. Due to inter-/extrapolation of valuation inputs the exposure mapping may often not be required achieving a higher accuracy. It is not entirely clear whether a smaller set of valuation inputs (justified due to data quality or liquidity considerations) without exposure mapping would be subject to the same strict requirements of Articles 8 and 9.
• The hierarchy set by Article 11 that calculation is preferable to judgment might be questioned in cases where the calculation effort requires undue costs in relation to the valuation uncertainty attached to the respective positions.
• In Article 105 (10) of the CRR, the AVA for concentrated positions is not mentioned. Therefore, it should be clarified whether this AVA belongs to a category outlined in the CRR, e.g. market price uncertainty.
• The methodology of the AVA for concentrated positions should be more detailed. For instance, the impact of the exit strategy on market price could be calculated using a Value-at-Risk at the targeted confidence level exceeding 10 trading days or any other appropriate risk metric. This calculation should take account of the exit strategy.
• The approaches chosen by institutions to Article 14 can be widespread and more guidance would be helpful. In most cases future administrative costs are covered by trading income, a positive carry or unrealized profit. The logic of incorporating a cost component without recognizing income components need clarifications in our opinion. EBA may consider limiting the AVA for administration costs to those that are not covered by expected income components.
• Usually, customers need to pay a fee for early termination to compensate the institution for the market price and a foregone future profit. It would be helpful to clarify whether Article 15 applies especially to early termination in case the institution would waive such fees on a regular basis or for other circumstances.
• In order to simplify matters, the EBA may consider granting the exception of zero AVA to level 1-instruments according to IFRS or equivalent local GAAP for some AVAs (e.g. market price uncertainty).
• In case EBA considers zero AVA for certain positions, it would be important though to allow netting between positions with zero AVA and other positions in the scope of the RTS (comparable to valuation uncertainty from hedged items designated in hedge accounting) in order to accommodate for hedging effects.
• We generally welcome less complex and/or internal risk-sensitive approaches.
• The option of a simple sum appears not unreasonable as a non-complex approach.
• Article 3 (2) of the CP prescribes the use of a “full range of available and reliable data sources” as listed. As mentioned before (see our response to Q1), we believe that the consideration of all possible data sources in the calculation of the AVAs might be in some cases overly burdensome with low added value. It should be noted that not all participants will have access to the same data sources by virtue of their level of participation in a given market. We believe a reduced and well selected range of data sources might be sufficient in certain cases. By adjusting the requirements on the data sources to be used, the monitoring process required by Article 20 might also be less demanding for the institutions using the core approach.
• Article 19 (3) states that institutions shall report documented valuation methodologies to its board at least annually. A reporting on the operational effectiveness and impacts could be relevant, whereas the reporting requirement to the board on valuation methodologies could focus on products and/or activities with the most significant or most volatile AVAs.
• Though we agree in general that a backtesting of prudent valuation would increase its reliability, the detailed requirements laid out in Article 20 (3) may practically prevent the application of a reduced set of valuation inputs.
• With the exception of banks having a robust and comprehensive IPV process already in place, we doubt that “many required systems and processes are in place” or “require only very few additional resources” captures the European picture accurately.
• Especially for small and medium sized banks the pricing process itself is often already set-up independently from front office and largely in an automated fashion. Hence, the IPV process in these banks is far from supporting prudent valuation and would require substantial modifications and enhancements to accommodate for the requirements of the CP. In this regard the implementation costs for the core approach would be substantial for less sophisticated banks.