French Banking Federation

French credit institutions are concerned regarding the backtesting at the trading desk level on Actual P&L (APL): although it could not be considered as a statistical test of VaR-measure (contrary to backtesting on Hypothetical P&L (HPL)), Basel Committee decided to include APL backtesting to assess the integrity of the metric.

As a reminder, the objective of backtesting on the Actual P&L is to demonstrate that the VaR-based capital (ES) is sufficient to cover market losses. Although capital is calculated at the entity level and not at the trading desk level, this test was finally retained by Basel Committee.

Differences between VaR and Actual P&L (time effect, non-daily adjustments, new/terminated trades) will most likely generate bad outcomes in backtesting results that can prevent eligibility to IMA for some trading desks. For that reason, French credit institutions are not in favour of using APL backtesting to determine eligibility to IMA. Hence, special attention will be paid to Article 1 of this consultation paper.

As defined in the Basel standard, Actual P&L excludes valuation adjustments for which separate regulatory capital approaches have been otherwise specified as part of the rules or which are deducted from Common Equity Tier 1 (CET1).

Hence, the following adjustments are excluded from the Actual P&L (as well as Hypothetical P&L):
- CVA (separate regulatory capital approach – CVA Capital Charge);
- Profit/losses generated by the default of a counterparty including bond or other security (separate regulatory capital approach – EAD (Exposure At Default) and DRC (Default Risk Charge));
- DVA (deducted from Common Equity Tier 1 – CRR Article 33(1)(c)).

French credit institutions consider that Own Credit Adjustment (profits/losses on own liabilities – CRR Article 33(1)(b)) should be excluded and consequently, same treatment should be applied in VaR and ES (no risk factor for Own Credit Spread). If the EBA believes that this interpretation is correct, a reference should be added in paragraph 3 of Articles 1, 2, 3 and 4.

There is no clear-cut opinion regarding the inclusion of FVA in the Actual P&L. Although the FVA is sensitive to funding risk at the first order, it is also sensitive to market risk at the second order. There is however a consensus to exclude FVA from Hypothetical P&L unless it is capitalized in the model: HPL is a statistical test of the VaR measure and should have the same adjustments. For this reason, French institutions believe that the wording used for the inclusion of daily adjustments in the Hypothetical P&L is not in line with the Basel standard:
- No systematic inclusion of daily adjustments: Basel [MAR10.32] “[…] Valuation adjustments updated daily should usually be included in the HPL”;
- Backtesting is a statistical test of the VaR-measure (Basel [MAR99.6]).

To be consistent with Basel text, Article 3 and 4 should allow flexibility on the valuation adjustments to include in the Hypothetical P&L.
Day-one P&L Reserve (see our answer to question 6 of this consultation) is a pure Accounting reserve which is clearly not in the scope of market risk. As a consequence, day-one P&L Reserve is excluded from Actual and Hypothetical P&L.

We would like to highlight that the reference to the sub-paragraph 3 of Article 33(1)(c) of Regulation (EU) No 575/2013 should be removed from Articles 1(3)(b), 2(3)(b), 3(3)(b) and 4(3)(b) of the draft Regulatory Technical Standards with regard to back-testing requirements under Article 325bf(9) and profit and loss attribution requirements under 325bg(4) of Regulation (EU) No 575/2013. Indeed, Article 33(1)(c) of Regulations (EU) 575/2013 and 2019/876 has no sub-paragraph.
In our view, all the adjustments not listed in our answer to Question 1 of the EBA consultation EBA/CP/2019/06 may be included in the Actual P&L (APL). Among them:
- Bid/Ask reserve;
- Model reserve;
- Independent Price Verification.

As reminded in the answer to Question 1, only daily valuation adjustments included in the VaR/ES should be included in the Hypothetical P&L (statistical test of the VaR measure).

Moreover, there is a consensus to exclude Independent Price Verification (IPV) from the Hypothetical P&L (HPL). If IPV is included in HPL, it will generate an asymmetry with Risk Theoretical P&L (even if we apply conditions under Article 6(2) of this consultation paper) that could lead to ineligibility to IMA through the P&L Attribution Test.
French banks consider the provision detailed in paragraph 4 of Article 1 is clear and consistent with the Basel standard MAR32.27.
The provision of paragraph 4 is clear, but we do not agree with it. Indeed, the Basel standard MAR32.28 offers more leeway than the EBA draft technical standard. It should be considered in the final version of the EBA technical standard, that some adjustments (e.g. Bid-Ask) must be computed at top level and a computation at the trading desk level is not aligned with IFRS 13 requirements.
We agree with the criteria listed in paragraph 5 of Article 1.
As specific terminology may differ across the Industry, French institutions propose the following definition to answer this question:
Day-one Profits/Losses correspond to the gains or losses at inception. It is measured as the difference between the transaction price and the fair value at initiation (fair value is generally observed at the end of the day, as a consequence day-one P&L includes intraday volatility between transaction time and closing time).

IFRS 13 requires day-one gains and losses to be recognized in P&L unless another IFRS specifies otherwise. IAS 39 and IFRS 9 prohibit the recognition of Level 3 day-one P&L (instruments for which valuation techniques rely on non-observable market data). As a consequence, on Level 3 instruments, the common practice is to calculate a day-one Profits/Losses Reserve (for accounting purposes only): Day-one Profits/losses is reserved and released through the life of the trade. Given that this reserve is not in the scope of market risk, it is not included in backtesting P&Ls.

Day-one P&L is generally split into:
• Trading Margin: hedge cost required by the trading;
• Client Margin: Margin at inception of the trade on top of trading margin;
• Market volatility between trade time and EOD (i.e. End-Of-Day).

Quantifying precisely each of the 3 elements above would require a closing valuation at the precise trade time. Given that such practice is unrealistic, industry members generally quantify client margins on a declaratory basis. On the other hand day-one P&L is generally computed as the difference between the fair value (observed at the closing time) and the trade price. Neither client margin nor day-one P&L is considered as commission.

Intraday trading as well as new and modified deals should be included in the Actual P&L. As a consequence, day-one P&L is included in the Actual P&L.

New trades are not part of the Hypothetical P&L. Hence, day-one P&L is not included in the Hypothetical P&L.

Client Margin is part of the day-one P&L and is consequently included in the Actual P&L and excluded in the Hypothetical P&L.

The inclusion (respectively exclusion) of day-one P&L and Client Margin in the Actual P&L (respectively Hypothetical P&L) is clear and does not need to be clarified further in Articles 1 to 4.
Please see our answer to question 4 of this consultation (EBA/CP/2019/06).
The French Banking Federation considers as overly burdensome the possibility to include in the portfolio’s changes the value of an adjustment stemming from the entire portfolio of positions, as outlined in paragraph 5 of Article 2 of the draft Regulatory Technical Standards.
The French Banking Federation considers as overly burdensome the possibility to include in the portfolio’s changes the value of an adjustment stemming from the entire portfolio of positions, as outlined in paragraph 5 of Article 2 of the draft Regulatory Technical Standards.
The French Banking Federation welcomes that the alignment of input data, introduced in paragraph MAR32.31 of the Basel standard, has been transposed in Article 15 of the draft Regulatory Technical Standards.

The example provided in the consultation paper (i.e. case 2 in page 19) clarifies the conditions under paragraph 2 of Article 15 may apply. However, French institutions believe that the application criteria should cover situations where the risk factors for the HPL are transformed into equivalent risk factors in RTPL, yet represented differently, for the purpose of risk modelling.

Typically, it is common practice to transform observations of IR futures contract with fixed maturity date (used for HPL), into equivalent price of “synthetic instruments” with fixed time-to-maturity (used in the RTPL) to exclude time decay effects from the calibration of risk factors and thus enable consistent shock distributions in the risk engine.

The industry considers that data alignment should be allowed in such situation as long as institutions can justify to the satisfaction of supervisory authorities that these inherent transformations of market data, performed for the purpose of risk modelling, neither reduce the granularity of risk factors nor distort the representation/dynamic of the PL function, and provide documentation thereof as required in Article 16(2).
Pierre-Loic BENOIT
F