EAPB

Generally the EAPB agrees with the prioritisation of EBA and welcome the approach to discuss important technical aspects of the SA-CCR and FRTB at an early stage with the industry. However, we believe that the definition of the revised trading book boundary should be addressed at an earlier stage too. In order to give guidance on implementation and investment decisions, especially the understanding of exceptional circumstances and reclassification has to be given early to allow for alignment of risk and accounting departments. Additionally, the boundary definition will have impact on trading desk structure.
EBA’s proposed allocation of products to different risk categories seems correct and suitable. From the EAPB’s point of view, the list is easily understandable and can be implemented well.
In general, bond-forwards on investment-grade bonds and cross-currency swaps are common OTC-products just like other products in the list. The EAPB thinks that bond forwards and investment-grade bonds should be assigned to the “credit” risk category. This would be consistent with the PFE-calculation in the context of the ISDA SIMM-approach which provides for an assignment of bond-forwards into the “credit” category.
Cross-currency swaps should be assigned to the “interest rate” category. Their components (swap legs) do not necessarily deviate from the components of an interest-rate swap. For instance, a EUR/USD cross-currency swap consists of a EUR-swap leg against the 3 months-Euribor and a USD-swap leg against a 3 months-USD-Libor. Currency-swaps are assigned to the “interest rate” category, so that an assignment of cross-currency swaps to that category also makes sense. Further, an assignment to the risk category “foreign exchange” also seems suitable.
Additionally, total return swaps and performance swaps should be added in the list. More specifically, they should be added to the category credit risk or equity risk.
We do not think that a list of criteria is useful, since it would be difficult to implement and since it would create too much room for deviating interpretation which would run counter a consistent implementation across the EU.
The rationale for the proposals is understandable but their implementation would certainly create a lot of administrative burden and/or complexity. Further, institutions could get to different results regarding the categorization. This would not be desirable for the use of a standardized approach. Moreover, the EAPB is afraid that “step 2” would be applicable rather often and not just in exceptional cases, as it seems to be intended.
The EAPB thinks that option 1 would be the simplest option since it is a standardized approach. However, we think that there should still be a limit for the number material risk-drivers. Further, option 3 seems to strike a good balance between administrative burden and precise risk measurement, as it makes the sensitivities of the different risk categories more comparable. For small institutions option 3 could however be too burdensome. All in all, the standardized approach should be relatively easy to calculate but still sufficiently risk sensitive.
Concerning threshold Y in option 2, we think that a percentage of 70% is adequate. This threshold can be derived from the alpha-factor of 1.4 which is a safeguard for imprecisions of models.
Moreover, the EAPB would have two amendment proposals for option 2 which aim at making the approach simpler and at avoiding that risk factors are identified as “material” for derivatives even though they are in fact of minor importance. These proposals would in principle also applicable to option 3 and option 4.

Proposal 1: relative sensitivities, grouped by risk categories:

For steps 1-4 in the calculation of a_(1,2,…,n) the sensitivities of the risk factors should be grouped by risk categories. From our point of view, a derivative for which several risk factors of a risk category are exceeding Y%, should only be assigned to that one risk category. That way the market risk is adequately reflected. Further, the derivative anyway only has to be assigned to the hedging set which provides for the highest sensitivity (according to Art. 277 para. 5 of the CRR-review proposal).

Example 1 (assumption Y = 70%):
For a specific derivative, five risk factors and relative sensitivities have been identified:

risk factor relative sensitivity per risk-factor a_⁄s_n
Interest rate risk-factor 1 65% (=a_1)
Interest rate risk-factor 2 10% (=a_3)
Interest rate risk-factor 3 6% (=a_4)
FX-risk-factor 1 15% (=a_2)
Credit-risk-factor 1 4% (=a_5)

Option 2 of the discussion paper would lead tot he fact, that the derivative would be assigned to the risk-category interest rate as well as FX, since only with the summation of the relative sensitivities of the interest rate risk-factor (a_1+a_2=80%) the threshold of Y = 70% would be exceeded, even though 81% of the sensitivities (=a_1+a_3+a_4) are based on interest rate risk. Further, the FX-risk should be considered secondarily compared to interest rate risk.

A summation of the relative sensitivities based on risk categories would result in the following relative sensitivities a_i:

Risk factor relative sensitivity per risk-factor relative sensitivity per risk-factor a_i⁄s_n
Interest rate risk-factor 1 65% 81% (=a_1)
Interest rate risk-factor 2 10% 81% (=a_1)
Interest rate risk-factor 3 6% 81% (=a_1)
FX-risk-factor 1 15% 15% (=a_2)
Credit-risk-factor 1 4% 4% (=a_3)

In this case, the derivate is only assigned interest-rate category, as the risk factors of the risk category interest-rate risk are exceeding 70%. Thus, the risk category FX would not be taken into account.

Proposal 2 (assumption: Y = 70%): introduction of an option 1 element:

This proposal would introduce an element from option 1 in option 2, by way of introducing a comparison of sensitivities from the main risk factor with the sensitivities of other risk factors. This would ensure that only the main risk-factor is relevant for the assignment to the risk category, even in case its relative sensitivity does not exceed Y. For this purpose the sensitivities of the other risk factors are put in relation to the main risk factors (threshold X).

Example 2 (assumption: Y = 70%; assumption X = 30%):

Risk factor Relative sensitivity a_i⁄s_n
Interest rate risk-factor 1 15% (=a_2)
FX-risk-factor 1 65% (=a_1)
Credit risk-factor 1 12% (=a_3)
Other risk factor 1 8% (=a_4)

When applying option 2 of the discussion paper the derivative would be assigned to the risk categories FX and interest rate, since the sum of a_1+a_2 would exceed the threshold Y = 70%. When applying EAPB’s proposal and the introduction of another threshold (X), the assignment would only be to the FX risk category because the sensitivity of the second highest risk factor compared to the main risk factor would not exceed the threshold X = 30% (15%/65% = 23%). From EAPB’s point of view, this would be adequate since the interest rate risk is relatively low compared to the FX risk.
Art. 277 para. 6 point a of the CRR proposal provides for a technical standards for the identification of the material risk driver for derivatives of the banking book. The discussion paper contains a respective proposal which however foresees that derivatives of the banking book which would not qualify for the catalogue of step 1 could only be considered under the fallback approach (step 3). The EAPB would therefore ask for the possibility that the assignment of derivatives of the banking book can be done through an internally defined qualitative approach under step 2. Procedures and decisions under this approach shall be adequately documented and should be subject to supervisory review.
Additionally, the review of the CRR will likely lead to a postponement of the date of first application of the FRTB. This means that the first application of the SA-CCR would precede the one of the FRTB. As a consequence, there would be no sensitivities for derivatives of the trading book when first applying the SA-CCR. This would however be necessary for the assignment to risk categories and hedging sets. Therefore, it is necessary to provide for transitional provisions.
In general, this definition seems suitable.


FRTB – Trading book boundary

The EAPB generally welcomes an early assessment from the EBA on the boundary between the trading and the banking book. Nonetheless, we have further remarks which go beyond the scope of questions in the discussion paper.
For instance, we do not agree with EBA‘s stated opinion in para. 101, that “in the event of conflict between a list in Article 104 and the institution’s purpose, it is expected that the list prevails”. This stands in contradiction to Art. 104 (4) CRR II which allows institutions to assign a trading book instrument to the banking book, if no trading intend is being followed. Thus, the relevant paragraph of the EBA discussion paper should be deleted.
Additionally, para. 102 states that a CIU should be assigned to the banking book, if look-through is not possible and daily prices are not available. This is also not in line with Art. 104 (3)(d) of the EU-Commission’s CRR II proposal. As this Art. states that a CIU has to be assigned to the banking book where look-through is not possible or daily prices are not available.
Our interpretation was that changes in instruments’ circumstances that imply a shift between the presumptive lists are not exceptional circumstances" and can thus be shifted. However, this could be accepted as ‘exceptional circumstances’, but institutions should not be required to move positions."
This depends on the reason for illiquidity. If only single positions become illiquid, this is not understood as exceptional circumstance". If an entire market breaks down, we think that this is an exceptional circumstance and re-designation should be possible."
Exceptional circumstances might be:
• break down of an entire market segment (market disruption, severe restrictions on tradability)
• restructuring of a bank's business model, e.g. close down of certain trading activities (or all trading activities)

We think it should be the goal to synchronize the regulatory trading book and the accounting trading portfolio. In IFRS (IAS 39) and German GAAP (HGB) are similar rules for re-designation based on rare" or "exceptional" circumstances."
In general, the identification of non-trading book FX positions is not an issue and carried out on a daily basis. Front office and risk systems flag these positions based on associated portfolio and the presence of foreign exchange risk. Nonetheless identification issues might come up for portfolios outside the scope of market risk when modified investment and hedging strategies lead to new foreign exchange risks. This could typically be the case for non-monetary items (equity investments, funds, real estate) with low materiality.
The valuation of non-trading book FX positions under FRTB standards is not different from trading book positions and will be carried out in front office systems. Currently, only the FX component of these positions is revalued. Valuation is performed on a daily basis.
The valuation of positions subject to the own funds requirement for market risk is based on the valuation used for the calculation of economic PnL (by risk department; independent of trading) for the vast majority of positions carrying FX risk. Therefore, most positions are marked to market on a daily basis, even if they are not accounted for at fair value, including loans in foreign currencies.
There are only few positions, like pension provisions of foreign branches or non-financial Assets, that are neither accounted for at fair value nor marked to market in the economic PnL. As a general rule, those positions are valued for accounting on a monthly basis at accounting date. These values are used for the purpose of computing the own funds requirement for FX risk.
The FX components of non-trading book positions subject to market risk are revalued as mark to market" per currency on a daily basis. In general non-trading book positions are market to market, only if a market value can be assigned meaningfully or observed otherwise. This is done irrespective of the accounting treatment. Exceptions to this are the credit components, loans not designated at fair value and equity investments. If there is no mtm-valuation for the economic PnL the most current available accounting value is used (see question 18)."
Yes, but regardless of the treatment for accounting purposes, most non-trading book items subject to FX risk are marked to market daily via economic PnL (see Q18, Q19).
Yes, the EAPB believes that there can be other factors that can affect the valuation of non-trading book FX positions. impairments and similar factors are often taken into account for valuation (e.g. valuation of loans comprises impairments on a daily basis).
In our opinion the requirements of “4.4 FRTB - Treatment of non-TB positions subject to FX or commodity risk“ in regard with the notional trading desks are not relevant for institutions with an asset and liability management of FX risks which is only in the banking book.

Additionally, it should be made sure that pure banking book FX positions are not included twice in the calculation of the regulatory capital requirements (e.g. a hedge of a fixed rate security in FX with a cross-currency swap).

In case of institutions with trading books, the number of notional trading desks needed might be dependent on the institution’s business model. Typically notional desks for treasury units will be clustered according to maturity of affected positions. Additionally, each branch might require a notional desk on its own, if management of positions is carried out individually. Restricting the number of notional desks allowed to just one would obstruct an integrated interest and liquidity management by forcing institutions to separate FX positions from multiple banking book units to one combined FX notional desk. This would reduce governance on affected positions.
However, if institutions have only one notional desk for FX, they should not be forced to have more desks. It really depends on the institution’s business model and size.
In our opinion the requirements of “4.4 FRTB - Treat-ment of non-TB positions subject to FX or commodity risk“ in regard with the notional trading desks are not relevant for institutions with an asset and liability management of FX risks which is only in the banking book.

Generally, the EAPB agrees that trading book positions should not be included in notional desks. Due to the different nature of a notional desk compared to a trading desk. Most of all the qualitative requirements of Article 104b(2) should not apply to notional desks. Where appropriate the notional trading desks may be included in desk reporting and limiting.
However, if an institution wishes to include trading book positions in notional trading desks they should be allowed to. This might be useful due to organisational constraints. Especially for smaller branches it is not always useful to separate access to the market from other units. For those combined desks not all trading desk requirements should apply, especially if a waiver according to Article 104b(4) CRR II is granted for SA only banks. Especially organisational and reporting requirements should adjust to banking books (point (a) and (d) from Article 104b(2) CRR II).
According to status quo, backtesting requirements might apply on notional trading desk level. However, we draw attention to the known issue (e.g. from partial-use models) in backtesting a single risk component (in this case FX risk). It needs to be well-defined how to deal with theta effects and what will be covered by theta effects. In the end, the coverage of the FX rates returns will be backtested only.
According to known issues regarding the P&L attribution for well-hedged desks, P&L attribution requirements should not apply to notional trading desks. Since FX risk in the non-trading book is extremely well-hedged (nearly zero exposure) for many banks, P&L attribution is not applicable in a meaningful way.
The EAPB agrees with the proposed general definition.
Complementing the definition with a list of instruments seems useful and appropriate. Still, the possibility to exclude instruments from RRAO is needed to reflect individual circumstances allowing institutions to define ‘other risks’ along the lines of their internal risk governance. This approach is also following the intentions of the trading book boundary definition.
We do not believe that there are any more instruments which need to be included additionally to the RRAO-list.
Assignment of risk factors to appropriate risk factor category/ subcategory is pretty clear. Uncertainties of some risk factors can be clarified by Q&As. Thus an explicit RTS is not mandatory.
The proposed triennial update of the list makes sense and is in line with the initial definition of the list. Additionally, an update of the list should be possible, if a significant market event happened (e.g. introduction of Euro).
The proposed definition of most liquid currencies" is comprehensible, but the list should be extended by application of suggested triangulation logic in para 161, which we find very useful."
The proposed triennial update of the list makes sense and is in line with the initial definition of the list. Additionally, an update of the list should be possible, if a significant market event happened (e.g. introduction of Euro) and the list should be extended by application of suggested approach in para 161.
Triangulation of currency pairs should be included in the definition of most liquid currency pairs. It is an appropriate and common way to hedge FX risks.
The proposal of combination of absolute and relative criteria (differentiation by equity markets; member in an equity index etc.) in the categorisation of companies makes sense. However, the EAPB believes that the proposed way of using the list of equity indices, which is published by ESMA, only is not sufficient. It should be extended at least by large companies, which are not part of an index and by companies, which have a reasonable size and are part of a liquid index, which is not on the list of ESMA (e.g. SDAX-companies).
see answer to question 45.
In this respect, retaining the existing Basel II standardised approach as a simplified alternative approach appears to be a more appropriate solution for achieving the stated goal of less complex calculation. The advantage of such an approach is that the institutions in question would not have to make any far-reaching adjustments to procedures and processes.
The EAPB would agree that a simple risk-sensitive recalibration could be based on different weightings of risk factors or groups of risk factors, as also contained in the proposed sensitivities-based methods.
But please note that the size of an institution’s trading book does not give any indication about the materiality of the associated market risk for a given institution. Therefore, other criteria should be used to determine which institutions should be allowed to use the simplified standardised approach. One example for an appropriate criterion could be a threshold for the relation of the profit or loss of the trading book and the level of own funds.
Jeroen van der Donck
E