The industry preference is Option 1b: Y= 60% and Z = 30% since we believe that Y=60% is a more appropriate option for the first threshold and provides a better representation of the risks and 30% provides an adequate backstop to identifying significant risk drivers.
The Industry believes that full reliance on FRTB SA sensitivities at trade level is not justified for the sole purpose of ranking risk drivers. We would argue for flexibility in the approach of ranking risk drivers and a way to leverage existing approaches considering the following four principles:
• Scope: The vast majority of transactions should be covered by approach 1 in order to minimize the scope of transactions subject to a quantitative analysis to determine the primary risk driver
• Timing: Any quantitative analysis under approach 2 should not have to be performed more frequently than at inception or quarterly basis
• Granularity: Banks should not be required to perform the quantitative analysis at a trade level but should have the flexibility to make determination at a portfolio level of similar transactions
• Flexibility: if for operational reasons FRTB sensitivities at trade level are not easy to implement, banks should have the flexibility to use other sensitivities as long as the bank uses a consistent set of sensitivities that are derived from bank’s independent process such as valuation or risk management.
Given that the scope of transaction that fall into Approach 2 should be kept small and that the quantitative information is only used for risk driver identification, the Industry believes such flexibility is justified.
In contrast, the quantitative methodology under Approach 2 proposed in the EBA Consultation Paper creates a clear dependency between the Market Risk framework and the Counterparty Credit Risk framework both in terms of methodology and IT systems. As such, depending on bank internal organization, using sensitivities for the identification of the most material risk driver may be burdensome and technically complicated. The Industry therefore urges the EBA to introduce appropriate flexibility also for Approach 2.
As mentioned in the executive summary, the Industry would encourage an expansion of the list for Approach 1, as well as maintaining the list as Guidelines rather than an RTS, thereby providing greater flexibility for the list to be updated more easily. With that goal in mind to expand transactions covered by Approach 1, the Industry recommends the removal of the following part of Article 1(1)(b): “where the currency of the underlying of the transaction is the same as the settlement currency of the transaction”, since the FX risk concerned here is either not material or already captured.
We do not have specific views in relation to appropriateness of this alternative specifically applied to smaller institutions. However, as mentioned above, general principle should be to apply any quantitative methods to as marginal as possible list of trades and to leave banks an adequate discretion in choosing the most fit-for-purpose methodology (please refer to our response to question 2).
We support alignment with the Basel FAQ as a minimum standard, i.e., calibration of the shift at currency and bank level - based on each institution set of forwards/strikes in its own population of options. In particular, some IMM institutions might opt for this option as it could bring alignment/consistency with their IMM model and this choice would therefore be seen as operationally efficient. For greater consistency, banks should be allowed to use the same currency level shift for SA-CCR at as the one used for IMM.
However, whilst a shift at currency and bank level offers consistency across the portfolio, this may suffer of threshold effect in case of an option with a negative strike. As such, only one transaction may impact the figures of the whole portfolio. Thus, the Industry is of the opinion that the option to calibrate the shift at transaction level should be retained since we consider transaction-level calibration to be superior to current-level calibration in the following respects:
• Operational simplicity of implementation
• Minimum distortion introduced as no adjustment would be applied to options that do not require it under this approach
• With currency level calibration, the shift has a dependency on the option with the most negative strike. A single option could trigger a recalibration of the shift which is not desirable. This is not the case when the calibration occurs at transaction level which could imply unwarranted volatility on the outcomes across institutions and through time
• Consistency of the shift at transaction level across institutions. Indeed at a point in time, the same trade in portfolios of different institutions will result in the same SA-CCR exposure across institution, which might not be the case with alternative 3a) approach at currency level, where discrepancies between institutions could arise within the market quote retrieval process
We consider that a unique shift across all institutions would be operationally difficult to implement and note that it would have the undesirable consequence that a single option with a very low strike in one institution could trigger an immediate recalibration of shift across the entire Industry.
The Industry strongly supports the adoption of option 4a which implies a threshold 1bp. Main rationales are:
• The larger the shift introduced, the larger the shifted log normal distribution could move from log-normal to normal.
• As evidenced in the EBA CP, the necessary adjustment of volatility for larger shift could be particular complex as it depends on the option details. The Industry is incline to put forward simplicity: no volatility adjustment but with a shift as small as possible.
• A threshold of 1 bp combined with the most negative combination of strikes/forwards would pose no issue from a computational perspective as it would be well within the bounds of any machine precision.
Setting the shift to prevalent market conditions would be operationally burdensome, in particular it would be extremely challenging to react sufficiently quickly to new options with negative strikes.The potential benefits are not sufficiently clear to warrant operational difficulties. The Industry is strongly against the potential adoption of this approach.
As evidenced by EBA CP, an adjustment of volatility is theoretically required to ensure that a shifted log-normal delta aligns with the corresponding log-normal delta with a volatility of 50% (when both are defined).The size of this adjustment depends on the option details, and is practically complex to put in place. We would favour simplicity and not adjust volatility while keeping the distortion as low as possible (via a threshold of 1bp, see response to question 5).
The industry considers the definition provided in Regulation (EU) 2019/876, “CRR2” to be sufficiently clear. Where the relationship between risk driver and the derivative trade is one-to-one and a regulatory prescribed list is used to map trades to risk categories, the definition in “CRR2” will be used. However, where sensitivities are used to assign trades to risk categories, sensitivities could be used to determine whether a transaction is long or short.
The method proposed by EBA for determining whether a transaction is a long or short position in the primary risk driver or in the most material risk driver in a given risk category shall allow the qualitative approach set out in Article 6(b) for transaction where the classification is done using article 1. The Industry suggests the removal of the following part of Article 6(b): « where institutions apply the approach set out in Article 3(1)(a), ».