Response to consultation on ITS on amending Commission Implementing Regulation on benchmarking of internal models
Go back
1. In respect column 0060 (Credit Quality Category) of the C120.04, the following additional allowable values are required to support the ‘securitisations that are not in the ACTP’ risk class as per the official mapping of SEC-ERBA credit quality steps .
• ‘Credit quality step 7’, ‘Credit quality step 8’, etc. through to ‘Credit quality step 17’
• ‘Credit quality step All Other’
2. In respect of C120.04, the allowable values for data points relating to ‘securitisations that are in the ACTP’ do not support the representation of non-tranched instruments that may be included in the ACTP (e.g., non-securitisation hedges and Nth-to-default instruments). This may not be required for the current set of instruments but may be required in the future.
3. FRTB-SA CRIF does not include Risk Weight as a data field as risk weighting is usually determined by the SA model itself based on the inputs, rather than being an input to the model. The risk weight applicable to a given row should be readily identifiable from the specification of the risk, excepting those cases for which there are methodological alternatives. In the case of 120.02, data fields to capture methodological alternatives have already been added to the template proposed by the EBA (e.g., (0090): Division of curvature risk components for foreign-exchange risk by scalar). It is proposed by the Industry that the addition of a data field to capture the methodological alternative “Division by sqrt(2) for liquid ccy/ccy pair”, in addition to those already proposed, would eliminate any requirement for a distinct data field to capture the applicable risk weight.
Other minor points of divergence in the proposed DRC representation with respect to the existing data standards used by the Industry (i.e., the ISDA FRTB-SA CRIF) are as follows:
• FRTB-SA CRIF uses Recovery Rate notation rather than Loss Given Default
• FRTB-SA CRIF captures Tranche Thickness rather than distinct Attachment and Detachment Points for securitisations not in the ACTP
Firms using this alternative methodology may not distinctly produce notional and/or P&L + adjustment and so reporting of these values would be extremely burdensome. Therefore, the Industry recommends that the following columns should be made optional: 0140 (Notional), 0180 (Notional in reporting ccy), 0150 (P&L + Adjustment), and 0190 (P&L + Adjustment in reporting ccy).
a) there is an existing industry gold standard (i.e., the ISDA Unit Test ) that has extensive use across the industry and
b) diminishing returns have been observed in respect of this validation element and hence the benefits of a year-on-year validation process are considered by the Industry to be negligible
Should the EBA introduce this element instead of relying on the existing industry gold standard (i.e., the ISDA Unit Test) as proposed by the Industry, the Industry recommends that the portfolios and representation should be fully aligned to the existing industry gold standard (i.e. the ISDA Unit Test).
• The SBM validation portfolios do not align with the existing industry gold standard (i.e. the ISDA Unit Test).
• Unlike the ISDA Unit Test, the proposed portfolios do not test all possible combinations & permutations for GIRR Delta. Note the following example test condition deficiencies:
o different inflation curves for the same currency
o intra-bucket flooring
o inter-bucket negative square root alternative specification
o inter-bucket aggregation with ERM II currencies
MR 1: Do you see any issues or lack of clarity in the definition of the data points of templates C120.04 and C120.05? Do you foresee any issues in terms of compatibility of template C120.04 and data standards used by the industry?
The Industry has identified the following issues with the data templates:1. In respect column 0060 (Credit Quality Category) of the C120.04, the following additional allowable values are required to support the ‘securitisations that are not in the ACTP’ risk class as per the official mapping of SEC-ERBA credit quality steps .
• ‘Credit quality step 7’, ‘Credit quality step 8’, etc. through to ‘Credit quality step 17’
• ‘Credit quality step All Other’
2. In respect of C120.04, the allowable values for data points relating to ‘securitisations that are in the ACTP’ do not support the representation of non-tranched instruments that may be included in the ACTP (e.g., non-securitisation hedges and Nth-to-default instruments). This may not be required for the current set of instruments but may be required in the future.
3. FRTB-SA CRIF does not include Risk Weight as a data field as risk weighting is usually determined by the SA model itself based on the inputs, rather than being an input to the model. The risk weight applicable to a given row should be readily identifiable from the specification of the risk, excepting those cases for which there are methodological alternatives. In the case of 120.02, data fields to capture methodological alternatives have already been added to the template proposed by the EBA (e.g., (0090): Division of curvature risk components for foreign-exchange risk by scalar). It is proposed by the Industry that the addition of a data field to capture the methodological alternative “Division by sqrt(2) for liquid ccy/ccy pair”, in addition to those already proposed, would eliminate any requirement for a distinct data field to capture the applicable risk weight.
Other minor points of divergence in the proposed DRC representation with respect to the existing data standards used by the Industry (i.e., the ISDA FRTB-SA CRIF) are as follows:
• FRTB-SA CRIF uses Recovery Rate notation rather than Loss Given Default
• FRTB-SA CRIF captures Tranche Thickness rather than distinct Attachment and Detachment Points for securitisations not in the ACTP
MR 2: Do you agree with the proposed format for the collection of DRC data in templates C120.04 and C120.05?
As stated in the EBA RTS on Gross JTD amounts (Article 2, paragraph 1): “The alternative methodology to estimate the gross JTD amount of an exposure referred to in Article 325w(7) of Regulation (EU) No 575/2013 shall consist in calculating the difference between the market value of the instrument from which the exposure arises for the institution at the time of the calculation and the market value of the instrument from which the exposure arises calculated under the assumption that the obligor defaulted at that time.”Firms using this alternative methodology may not distinctly produce notional and/or P&L + adjustment and so reporting of these values would be extremely burdensome. Therefore, the Industry recommends that the following columns should be made optional: 0140 (Notional), 0180 (Notional in reporting ccy), 0150 (P&L + Adjustment), and 0190 (P&L + Adjustment in reporting ccy).
MR 3: Do you agree with the proposed amendments to template C120.06 (former C120.03) to include DRC and RRAO OFR by portfolio?
Yes, the proposed amendments are pragmatic.MR 4: In your view, what approaches would be suited to benchmark banks’ implementation of the RRAO requirements more comprehensively?
The Industry has no recommendations in respect of approaches to benchmark RRAO more comprehensively at this time.MR 5: Do you agree with the proposed change to the reporting of vega sensitivities?
Yes, the proposed amendment is pragmatic.MR 6: Do you agree with the proposed clarification with regards to taking the reporting currency view for the consideration of FX risk? Do you agree with the proposed clarification with regards to converting reporting currency results to the EBA portfolio currency using the applicable ECB spot exchange rate?
The Industry does not agree with the proposal to take a reporting currency view with regards to FX risk. In the opinion of the Industry no significant benefits could be expected from the proposed change, while it could lead to a number of potential complications. These include an increase in the variability of results with a lack of visibility of the underlying causes (especially for results relating to Basel 2.5 measures), the unintended benchmarking of FX risk for portfolios that have not been designed for that purpose, and a requirement for additional clarifications in respect of instrument level instructions.MR 7: Do you agree with the proposed introduction of individual and aggregated portfolios for purposes of SBM validation?
The Industry does not recommend the introduction of an SBM validation component as currently proposed for the following reasons:a) there is an existing industry gold standard (i.e., the ISDA Unit Test ) that has extensive use across the industry and
b) diminishing returns have been observed in respect of this validation element and hence the benefits of a year-on-year validation process are considered by the Industry to be negligible
Should the EBA introduce this element instead of relying on the existing industry gold standard (i.e., the ISDA Unit Test) as proposed by the Industry, the Industry recommends that the portfolios and representation should be fully aligned to the existing industry gold standard (i.e. the ISDA Unit Test).
MR 8: Do you see any issues or lack of clarity with the instructions of Annex 5 defining the SBM validation portfolios?
The following issues are noted:• The SBM validation portfolios do not align with the existing industry gold standard (i.e. the ISDA Unit Test).
• Unlike the ISDA Unit Test, the proposed portfolios do not test all possible combinations & permutations for GIRR Delta. Note the following example test condition deficiencies:
o different inflation curves for the same currency
o intra-bucket flooring
o inter-bucket negative square root alternative specification
o inter-bucket aggregation with ERM II currencies