Final Report on Draft Implementing Technical Standards Amending Benchmarking of Internal Models for 2026

EBA/ITS/2025/07 08/08/2025 Final Report Draft Implementing Technical Standards on amending Commission Implementing Regulation (EU) 2016/2070 with regard to the bench￾marking of internal models – 2026 benchmarking exercise.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 2 Contents

1. Executive Summary 3
2. Background and rationale 4 2.1 Market Risk benchmarking 4 (i) Amendments to Annex 5 (instruments and timeline) 5 (ii) Amendments to Annex 6 (instructions) and 7 (templates) 6 (iii) ASA validation portfolios (Annex 10) 6 2.2 Credit risk benchmarking 7 (iv) Amendments to Annex II 7
3. Draft implementing standards 8
4. Accompanying documents 12 4.1 Draft cost-benefit analysis for changes related to credit and market risk benchmarking 12 (v) Market risk 12 (vi) Credit risk 15 4.2 Overview of questions for consultation 18 4.3 Feedback on the public consultation and on the opinion of the BSG 19 ▪ Summary of key issues and the EBA’s response 19 ▪ Summary of responses to the consultation and the EBA’s analysis 21

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 3

1. Executive Summary Article 78 of Directive 2013/36/EU (CRD) requires competent authorities to conduct an annual as￾sessment of the quality of approaches used for the calculation of own funds requirements. To as￾sist competent authorities in this assessment, the EBA calculates and distributes benchmark val￾ues to CAs that allows a comparison of individual institutions’ risk parameters. These benchmark values are based on data submitted by institutions as laid out in Commission Implementing Regu￾lation (EU) 2016/2070 which specifies the benchmarking portfolios, templates and definitions to be used as part of the annual benchmarking exercises. For the 2026 benchmarking (BM) exercise the following changes are suggested: • For market risk (MR), templates and instructions remain mostly unchanged. The scope of the data collection remains limited to IMA banks, as in previous years. Nonetheless, the data collection will focus solely on the ASA components. • For credit risk (CR), the asset classes definition is aligned with the breakdown of Credit Risk IRB templates adopted in the revised Implementing Technical Standards (ITS) on supervi￾sory reporting. The EBA supervisory benchmarking serves three major objectives, the first one being the above￾mentioned supervisory assessment of the quality of internal approaches. However, it also pro￾vides a powerful tool to explain and monitor RWA variability over time and horizontally as well to indicate related implications for prudential ratios and the relevant policy. Lastly, the benchmark￾ing results also provide the institutions with valuable information on their risk assessment com￾pared to other institutions’ assessment on comparable portfolios. Next steps The Annexes presented in this draft ITS replace or are added to the existing set of templates in order to create a consolidated version of the updated draft ITS package. These draft ITS will be submitted to the Commission for endorsement before being published in the Official Journal of the European Union. The technical standards will apply 20 days after publi￾cation in the Official Journal.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MOD￾ELS (2026 EXERCISE) 4 2. Background and rationale 2.1 Market Risk benchmarking

1. Due to the European Commission’s proposal to postpone the implementation of the Fundamental Review of the Trading Book (FRTB)1 , the structure of the Market Risk Benchmarking exercise has been significantly affected.
2. In light of this delay, the previously planned introduction of new templates for the Alternative In￾ternal Model Approach (AIMA) and the expansion of the exercise to include all banks applying the Alternative Standardised Approach (ASA) have been deferred. Specifically, the expansion to ASA banks is now postponed until the Own Funds Requirements (OFR) is applied.
3. Consequently, the 2025 data collection format has been reestablished, maintaining the same scope as in the previous cycle. This decision ensures continuity and comparability while the regu￾latory framework is being finalized.
4. The reasoning behind the choice of leaving the scope of the exercise unchanged – i.e. banks with IMA approval - is based on the increased uncertainty associated with the further FRTB delay, fol￾lowing which in 2026 there will be no banks using ASA for the calculation of their own funds re￾quirements. In this sense, the eventuality of a further delay in the implementation of ASA had hampered the preparation for participation in the exercise, which could have also affected the outcome, resulting in poor quality of submissions due to a lack of understanding of the exercise, which – if a larger scope had been confirmed - would have been the first experience for a consid￾erable number of participants to the exercise.
5. This choice is also supported by the literal reading of Article 78(1)(b) CRD as amended by Directive 2024/1619, that refers explicitly to Article 325a(1) CRR whose introductory wording was amended by Regulation 2024/1623 (CRR3) as it would have applied at the time of application of Article 78(1) CRD in the absence of the FRTB postponement. This demonstrates the intention of the legis￾lator to include in the benchmarking exercise the banks in question as soon as they used ASA for the calculation of their capital requirements for market risk and not only for reporting purposes. In particular, while the threshold in Article 325(1)(b) has remained unchanged (500 millions), its purpose has changed, as in CRR2 it refers to the threshold under which an institution is exempted from the reporting requirement set out in Article 430b (which specifies that ASA is used for re￾porting purposes only), while in CRR3 it is the threshold above which it becomes compulsory to 1 https://finance.ec.europa.eu/news/commission-proposes-postpone-one-additional-year-market￾risk-prudential-requirements-under-basel-iii-2025-06-12_en

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MOD￾ELS (2026 EXERCISE) 5 apply ASA (and not the SA) for the calculation of the own fund requirements. Consequently, Arti￾cle 78(1)(b) CRD should be read as applying to institutions that use ASA for OFR, and not to insti￾tutions which only use ASA according to 430b, as this scenario was not foreseeable when Article 78 was amended. 6. Pausing the expansion of the scope of the exercise is also preferable in light of proportionality. More specifically, from Recital 39 of CRR3 the legislator's desire for a non-fragmented but unified application of all FRTB elements is demonstrated. A reading by which banks that do not currently participate in an exercise should be burdened with such participation regardless of the FRTB post￾ponement and despite their very inclusion in the exercise appearing closely linked with such FRTB application would be disproportionate and, therefore, less preferable. 7. For the banks in scope of the exercise – the IMA banks – the participation to the exercise has not been in question. However, it is proposed that the IMA banks do not report the IMA figures con￾cerning the portfolios, acknowledging their participation in the exercise for a prolonged period of time and that the transition from IMA to AIMA (Alternative IMA) or ASA has been postponed. 8. For what concerns the ASA data collection for IMA banks, those banks have already been report￾ing ASA figures as part of the benchmarking exercise. This reporting has been in place even prior to the formal inclusion of ASA within the benchmarking scope in Article 78 CRD , and therefore the ASA data collection of the IMA banks should continue, considering that the cost of participa￾tion is marginal compared to the initial participation of the ASA data submission, and it allow the supervisors to properly monitor the progress achieved by those banks with respect the previous exercise. 9. Additional detailed information on the changes in the Market Risk Templates is in the following section. (i) Amendments to Annex 5 (instruments and timeline) 10.A series of changes, listed below, were introduced to Annexes 5. 11.In light of the decision to restrict the data collection for the banks in scope, IMA banks, to the ASA data collection, in section 3-4-5 of the Annex, the request for the portfolios of the IMA element was set to “No”. • Update of the reference dates to the new 2026 exercise (section 1 – letter b). 12.A series of amendments to few instruments, mostly equity options, so that the maturity is still 9 months (i.e. 3 months later than the risk metrics references dates) after the booking date, as it was in the exercise 2024 and previous (section 2 of the Annex).

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MOD￾ELS (2026 EXERCISE) 6 (ii) Amendments to Annex 6 (instructions) and 7 (templates) 13.A series of changes, listed below, were introduced to the Annex 6 and 7. • New Template 106.02 was introduced to collect solely the SBM validation information, as specified in Annex 5 – Section 7 and Annex 10. It should be highlighted that this data col￾lection remains substantially unchanged with respect to in the past, where it was done via template 120.02. The new template should increase clarity in terms of content to be pro￾vided for this specific part of the exercise, as well as the timing to provide it (i.e. at the IMVs submission date). • Template 106.01 collects information concerning SBM – Risk sensitivities by instrument. Here, the Additional identifier 1 were adjusted. For Credit spread risk, the LEI was intro￾duced to unambiguously identify the obligor referred to by the instrument. For Equity risk, the ISIN was introduced to unambiguously identify the referred instrument. • Template 120.01 collects information concerning SBM – Risk sensitivities by Instru￾ment/portfolio. Here, the Additional identifier 1 were adjusted. For Credit spread risk, the LEI was introduced as a standard identifier to unambiguously identify the obligor referred to by the instrument. For Equity risk, the ISIN was introduced as a standard identifier to unambiguously identify the referred instrument. Template 120.04 was amended to in￾clude the LEI as reference for the obligor in a consistent way for equity and debt instru￾ments.

(iii) ASA validation portfolios (Annex 10) 14.Institutions are required to submit the set of instruments in Annex 10 (SBM validation portfolios) so that competent authorities are able to control the capacity of the institutions to correctly im￾plement the SBM aggregation formulas, without the additional element of complication due to bucketing and sensitivities calculation. It was highlighted in the previous consultation that this data collection, while beneficial for the proper control of the correct SBM calculation, could be￾come burdensome and redundant once the institutions correctly report the figures. The EBA acknowledges the information of the validation data submissions will decrease with the time, once it is done correctly by the institutions, and will consider making this data submission optional in the future for institutions that achieve a sufficient degree of precision in their submission in this area in order to reduce the submission burden. 15.The Annex 10 was also resubmitted in consideration of a limited set of amendments as suggested in the feedback received from the consultation of this ITS (see also the feedback table of this ITS for more details).

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MOD￾ELS (2026 EXERCISE) 7 2.2 Credit risk benchmarking (iv) Amendments to Annex II 16. The benchmarking portfolios are obtained as a drill down of the asset classes used in the Credit Risk IRB templates defined under the supervisory reporting ITS. The Annex is amended so to pro￾vide a mapping between the asset classes used for the definition of the benchmarking portfolios and the breakdown of Credit Risk IRB templates adopted in the revised Implementing Technical Standards (ITS) on supervisory reporting in line with changes in the regulatory framework related to CRR3/CRD6. 17.In detail, the instructions regarding the C 102.00 (Definition of Low Default Portfolios) and C 103.00 (Definition of High Default Portfolios) templates were amended so to provide a mapping between the asset classes used for the definition of the benchmarking portfolios and the break￾down of Credit Risk IRB templates adopted with the revised ITS on supervisory reporting. 18. In the proposed mapping between the asset classes used in the benchmarking and the break￾downs adopted in the ITS on reporting, it is required to map the breakdown B.6.3 (Retail – Pur￾chased receivables) in both the asset classes: Retail – SME – Other and Retail – no SME – Other. It is then asked to exclude counterparties and their exposures classified as non-SME from the first asset class and to exclude counterparties and their exposures classified as SME from the second asset class. 19. The analysis in Section 5 provides the reader with an overview of the findings as regards problem identification, possible options to remove problems and their potential impacts. Given the nature and the scope of the draft GL, the analysis is high-level and qualitative in nature.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 8 3. Draft implementing standards

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 9 COMMISSION IMPLEMENTING REGULATION (EU) No …/... of [date] amending Commission Implementing Regulation (EU) 2016/2070 laying down implementing technical standards for templates, definitions and IT-solutions to be used by institutions when reporting to the European Banking Authority and to competent authorities in accordance with Article 78(2) of Directive 2013/36/EU of the European Parliament and of the Council THE EUROPEAN COMMISSION, Having regard to the Treaty on the Functioning of the European Union, Having regard to Directive 2013/36/EU of the European Parliament and of the Council of 26 June 2013 on access to the activity of credit institutions and the prudential supervision of credit institutions, amending Directive 2002/87/EC and repealing Directives 2006/48/EC and 2006/49/EC2 , and in particular Article 78(8),third subparagraph thereof, Whereas: (1) Pursuant to Article 78(1) of Directive 2013/36/EU, institutions are required to submit to their competent authority, at least annually, the results of the calculations of their risk weighted exposure amounts or own fund requirements under their approaches for expo￾sures or positions that are included in the benchmark portfolios, to enable that competent authority to assess the quality of those internal approaches (‘benchmarking exercise’). (2) Pursuant to Article 78(3), second subparagraph, of Directive 2013/36/EU, the EBA has to produce a report to assist the competent authorities in the assessment of the quality of the institutions’ approaches, based on the results of the benchmarking exercise. (3) The Commission specified the reporting requirements for the benchmarking exercise in Commission Implementing Regulation (EU) 2016/20703 . That Implementing Regulation should be amended regularly to reflect the changes in the focus of the competent author￾ities’ assessments and of the EBA’s reports. For the same reason, it is necessary to update once more the benchmark portfolios, and thus also the reporting requirements laid down in Implementing Regulation (EU) 2016/2070 to reflect prudential elements introduced in Regulation (EU) No 575/2013 as amended by Regulation (EU) No 2024/16234 and Directive 2013/36/EU as amended by Directive (EU) 2024/1619. 5 (4) In particular, for the market risk benchmarking, considering the deferred application of the Fundamental Review of the Trading Book (FRTB) standards for the calculation of 2 OJ L 176, 27.6.2013, p. 338–436 3 OJ L 328, 2.12.2016, p. 1–1422 4 Regulation (EU) 2024/1623 of the European Parliament and of the Council of 31 May 2024 amending Regulation (EU) No 575/2013 as regards requirements for credit risk, credit valuation adjustment risk, operational risk, market risk and the output floor (OJ L, 2024/1623, 19.6.2024). 5 Directive (EU) 2024/1619 of the European Parliament and of the Council of 31 May 2024 amending Directive 2013/36/EU as regards supervisory powers, sanctions, third-country branches, and environmental, social and govern￾ance risks (OJ L, 2024/1619, 19.6.2024).

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 10 own funds requirements for market risk set out in [Commission Delegated Regulation (EU) XXX/XXX of XXX amending Regulation (EU) No 575/2013 with regard to the date of application of the own funds requirements for market risk], the scope of applica￾tion of the benchmarking exercise remains temporarily limited to institutions permitted to use internal approaches (Internal Model Approach) as referred in Article 78 (1), point (a) of Directive 2013/36/EU. (5) Moreover, considering that pending the further postponement the composition of port￾folios is substantially unchanged, it is opportune that the data collection from Internal Model Approach institutions is restricted to the element of the Alternative Standardised Approach data collection. This would not only not create any issue in terms of compara￾bility of annual data series, which has limited value for this data collection, but it would in any case reduce the burden for the participant to the exercise, while providing data considered mostly useful by the supervisors. (6) For credit risk benchmarking, asset classes used for defining the benchmarking portfo￾lios is aligned with the breakdown followed by the credit risk IRB templates adopted in the revised Implementing Technical Standards on supervisory reporting6 . (7) Implementing Regulation (EU) 2016/2070 should be amended accordingly. (8) This Regulation is based on the draft implementing technical standards submitted to the Commission by the EBA. (9) EBA has conducted open public consultations on the draft implementing technical standards on which this Regulation is based, analysed the potential related costs and benefits and re￾quested the advice of the Banking Stakeholder Group established in accordance with Article 37 of Regulation (EU) No 1093/20107 . HAS ADOPTED THIS REGULATION: Article 1 Implementing Regulation (EU) 2016/2070 is amended as follows: (1) Annex II is replaced by the text in Annex I to this Regulation; (2) Annex V is replaced by the text in Annex II to this Regulation; 6 https://www.eba.europa.eu/activities/single-rulebook/regulatory-activities/supervisory-report￾ing/implementing-technical-standards-supervisory-reporting-changes-related-crr3crd6-step-1 7 Regulation (EU) No 1093/2010 of the European Parliament and of the Council of 24 November 2010 establishing a European Supervisory Authority (European Banking Authority), amending Decision No 716/2009/EC and repealing Com￾mission Decision 2009/78/EC (OJ L 331, 15.12.2010, p. 12).

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 11 (3) Annex VI is replaced by the text in Annex III to this Regulation; (4) Annex VII is replaced by the text in Annex IV to this Regulation; (5) Annex X is replaced by the text in Annex V to this Regulation; Article 2 This Regulation shall enter into force on the twentieth day following that of its publication in the Official Journal of the European Union. This Regulation shall be binding in its entirety and directly applicable in all Member States. Done at Brussels, For the Commission The President On behalf of the President [Position] ANNEX Annex I (Credit Risk Benchmarking) Annex II (Market Risk Benchmarking) Annex III (Market Risk Benchmarking) Annex IV (Market Risk Benchmarking) Annex V (Market Risk Benchmarking)

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 12 4. Accompanying documents 4.1 Draft cost-benefit analysis for changes related to credit and market risk benchmarking Article 78 of Directive 2013/36/EU (CRD IV) requires competent authorities to conduct an annual assessment of the quality of internal model approaches, used for the calculation of own funds re￾quirements, and requires the EBA to produce a report to assist them in this assessment. The re￾port of the EBA relies on data submitted by institutions in accordance with EU Regulation 2016/2070, which specifies the benchmarking portfolios, templates, definitions and IT solutions to be used by the institutions as part of the annual benchmarking exercise, when using internal model approaches for market and credit risk. The current draft ITS aim to update the previous ITS for the benchmarking data collection with the purpose of improving the exercises and adapting to the relevant policy changes which will be ap￾plicable by end-2025 and thus relevant for the 2026 exercise. With regard to the credit risk no metrics have been deleted or newly introduced. Therefore, no in￾depth impact assessment is considered relevant. (v) Market risk Regarding the EBA’s market risk benchmarking data collection, the purpose is to amend the set of templates to collect information on the FRTB Alternative Standardised Approach (ASA). The new template (106.02) concerns the validation portfolios information to be provided. As per Article 15(1) of the EBA regulation (Regulation (EU) No 1093/2010 of the European Parlia￾ment and of the Council), any ITS developed by the EBA shall be accompanied by an Impact As￾sessment (IA) annex which analyses ‘the potential related costs and benefits’ before submitting to the European Commission. Such annex shall provide the reader with an overview of the findings as regards the problem identification, the options identified to remove the problem and their po￾tential impacts. For the purposes of the IA section of the Consultation Paper, the EBA prepared the IA with cost￾benefit analysis of the policy options included in the regulatory technical standards described in this Consultation Paper. Given the nature of the study, the IA is mainly high-level and qualitative in nature including quantitative analysis when possible.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 13 A. Problem identification With regard to the market risk benchmarking data collection, the previous ITS for benchmarking data collection have remained stable, in terms of risk measures collected (i.e. VaR, Stress VaR, IRC). B. Policy objectives The general objective of the current ITS is to amend the set of templates to collect information on the FRTB Alternative Standardised Approach (ASA). The main objective of the implementation of the current draft benchmarking ITS is to extend the set of templates to have a data collection of the specific portfolio ad defined in the Annex X of the ITS, which was defined for the porpoise of validation of the SBM aggregation formula. This would foster the strategic objective of creating a supervisory and reporting environment to ensure that institutions apply consistent modelling and valuation techniques. The following sec￾tions examine the options that could create such an environment, as well as the net impact that the implementation of such solutions implies. C. Baseline scenario For the market risk part of the exercise, for the EU institutions which are meant to applying for the ASA-FRTB, the current status of reporting the results of modelling and valuations implies the potential operational costs and miscalculations, which lead to overvaluation or undervaluation of the reported values for the purposes of the benchmarking exercises. Since the extent and magni￾tude of overvaluations or undervaluations cannot be identified, the impact assessment focuses on the assessment of the net impact on the institutions’ operations. D. Options considered When developing the draft ITS, the EBA considered the following options: Option 1: do nothing This option implies that credit institutions continue reporting data for the benchmarking exercise using just the previous set of templates for the exercises to date. For the market risk part of the exercise, the continuation of the application of just the previous set of templates assumes that credit institutions and the EBA have the usual operational cost as￾signed to providing clarifications and ensuring the consistent submission of data. The ‘do nothing’ option would imply leaving the Implementing Regulation on market risk bench￾marking unchanged, Annex X, which would result in obtaining no information as the features of the ASA approach are not fully represented with the current set of templates.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 14 Option 2: revision of the templates relating to the benchmarking exercises The main arguments that support the revision of the templates in the market risk benchmarking exercises are: A. That the current template would not be appropriate to collect the SBM validation figures provided by the institutions applying ASA FRTB framework; B. providing clarity in the data collection of the different type of data collected (Validation portfolios vs standard portfolio of the benchmarking exercise. For the market risk part of the exercise, the current ITS could achieve the objective by expanding the set templates to be used for the data to be collected. With new additional template (106.02) would be suited for the solely data collection of the portfolios as defined in the Annex 10. E. Cost-Benefit Analysis The principle of proportionality applies to all aspects of the impact assessment, including method￾ology, depth of analysis, level of detail and necessity of quantitative analysis. Being consistent with this principle, the EBA staff follow the principle of proportionality when conducting the cost￾benefit analyses. Given that the implementation of the current ITS would have a detrimental im￾pact, the following analysis focuses on the qualitative characteristics. In doing so, it provides rough estimations of the net monetary impact that relates to the conduct of benchmarking exer￾cises. The net impact on capital requirements, implied by the implementation of the current guidelines, cannot be precisely assessed because, substantially, it would depend on further actions agreed by institutions with national competent authorities in response to the benchmarking exercise results; however, it is expected to be on average close to zero due to the hypothetical market portfolio exercise framework. Market risk: Option 1 Costs: a possible loss of informativeness in the data collection. Benefits: one-off benefits (reduction of the existing operational costs) of not dedicating human resources to the drafting the present ITS. Option 2 Costs: the one-off cost of dedicating resources to the drafting of the ITS. There is also a source of minimal cost that relates to the need for the EBA to explain the new template to the national

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 15 competent authorities and, through them, the participating credit institutions. However, it is to be noted that the data requested with the new templates should not be too burdensome, since the template is based on the template 120.02, for the data collection of the standard portfolios in the benchmarking exercise. Benefits: the benefits of this option arise from providing clear framework for the collection of the ASA-FRTB validation information and data, which would trigger the provision of additional insights to competent authorities and would keep the exercise relevant for the banks involved. F. Preferred option The EBA considers that, although these benefits are not directly observable and are spread over time, they are not negligible, and they are considered more important than the costs enu￾merated above. For this reason, the preferred option is Option 2. (vi) Credit risk A. Problem identification The benchmarking portfolios are obtained as a drill down of the asset classes used in the Credit Risk IRB templates defined under the supervisory reporting ITS. Since the ITS on supervi￾sory reporting has been modified following the implementation of Basel III in the EU, it is nec￾essary to align the definitions used in the benchmarking. B. Policy objectives The main objective is to avoid asking to the reporting entities to employ definitions that are not aligned with the ones used for supervisory reporting. C. Baseline scenario The current instructions for the benchmarking would require to the reporting entities to em￾ploy definitions (in particular the definition of the asset classes) that are no more used for su￾pervisory reporting. D. Options considered When developing the draft ITS, the EBA considered the following options: Option 1: do nothing This option implies that reporting institutions continue mapping the exposures using defini￾tions of asset classes that differ from the ones used for supervisory reporting. This would imply

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 16 a misalignment of the data reported for the benchmarking exercise and the data reported for supervisory reporting. Option 2: revising the asset classes used in the benchmarking exercise This solution would enable to ensure the alignment between the data reported for the bench￾marking exercise and the data reported for supervisory reporting but at the cost of revising the definition of all the benchmarking portfolios i.e. the Annex I of the benchmarking ITS. Option 3: mapping the asset classes used in the benchmarking exercise with the breakdown of Credit Risk IRB templates adopted in the revised Implementing Technical Standards (ITS) on su￾pervisory reporting. Under this solution, the reporting entities can continue to use the data extraction and aggrega￾tion procedures (programs) that they have used in the past as it only requires performing the suggested mapping between the asset classes used in the benchmarking and the breakdown of Credit Risk IRB templates adopted in the revised Implementing Technical Standards (ITS) on supervisory reporting. In this way it is possible to ensure an alignment (even if not perfect) be￾tween the benchmarking exercise and the revised ITS with minimal costs. Option 4: not doing the data collection only with reference to December 2025 Under this solution, the costs for the reporting entities would be the lowest possible but this would prevent the EBA and the Competent Authorities to fulfil the mandate in Article 78 – CRD and it would generate a break in the time series of the data collected that in turn could repre￾sent a limit for the future analysis. E. Cost-Benefit Analysis The following analysis focuses on the qualitative characteristics. In doing so, it provides rough estimations of the costs and benefits of each option. a) Option 1 Costs: for the reporting entities, the need to maintain definitions that would be used only for the benchmarking exercise. In general, data misalignment between the benchmarking exercise and supervisory reporting that in turns would make data quality checks more difficult. Benefits: any interventions to the programs would be required for both the reporting entities and the data users. Option 2 Costs: this is the most expensive option as it would require a complete review of the ITS and, in turn, a review of the programs and procedures used for this exercise by both reporting entities

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 17 and data users. It should also be noted that further revisions of the ITS on supervisory reporting are expected in the near future. Benefits: the benefits of this option arise from providing a better alignment between the bench￾marking exercise and supervisory reporting. b) Option 3 Costs: for the reporting entities the costs are minimal and for data users no costs at all are ex￾pected. Benefits: avoiding misalignment between the benchmarking exercise and the supervisory report￾ing. c) Option 4 Costs: for the reporting entities the costs would be minimal, but the EBA and the Competent Au￾thorities could not perform the assessments of internal models required by Article 78 CRD. Benefits: no benefits. F. Preferred option Considering the balancing between cost and benefits, the EBA preferred option is Option 3.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 18 4.2 Overview of questions for consultation Questions MR Q1: Do you see any issues or any missing information that should be required in the new tem￾plates suggested for the AIMA FRTB benchmarking exercise (i.e. Annex 6 & 7)? MR Q2: Do you see any issues with the reduced subset of instruments proposed for the AIMA ex￾ercise? Please elaborate? MR Q3: Do you see any issues with the new template 106.02? Please elaborate. MR Q4: Do you see any issues with specifying the specified timeline in the Annex 5 or with the reference date for new ASA institution in the exercise as defined in the suggested draft of Article 4.1.(b)? MR Q5: Do you see any issues with the changes introduced in the Annex 5? MR Q6: Would you consider it useful to clarify the type of SOFR rate (term, compound) to be used when booking related interest rate instruments? If so, please suggest a clarification. CR Q1: Do you think that the proposed approach aimed at including the breakdown B.6.3 is cor￾rect and it enables to avoid any double counting of the exposures?

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 19 4.3 Feedback on the public consultation and on the opinion of the BSG The EBA publicly consulted on the draft proposal contained in this paper. The consultation period lasted for three months and ended on 26 May 2025. Two responses were received, of which two were published on the EBA website. This paper presents a summary of the key points and other comments arising from the consulta￾tion, the analysis and discussion triggered by these comments and the actions taken to address them if deemed necessary. In many cases several industry bodies made similar comments or the same body repeated its comments in the response to different questions. In such cases, the comments, and EBA analysis are included in the section of this paper where EBA considers them most appropriate. Changes to the draft ITS have been incorporated as a result of the responses received during the public consultation. Nonetheless, most of the change are due to the delay of the FRTB, as officialised by the European Commission the 12 June 2025. ▪ Summary of key issues and the EBA’s response A respondent has several concerns and suggestions. The respondent reiterates concerns about reduced participation in the Internal Models Approach Benchmarking post-FRTB go-live. They sug￾gest a stepwise approach starting with IR and FX for 2026. Moreover, they remarked that the ex￾pansion of the EBA mandate to include FRTB Standardised Approach will increase participation, posing operational challenges. EBA already restricted substantially the volume of submissions for IMA banks, in view of the FRTB go-live and the reduced participation. EBA and supervisors are aware of the operational challenges on the expansion of the scope, and they are actively operating to minimize it (with the project to provide training and keeping the framework stable, i.e. not changing substantially the composition of the portfolios). For the moment, the June 12 Delegated Act of the European Commission8 postponed the imple￾mentation of the FRTB, and with that, the expansion of the scope of the benchmarking exercise was posed (see also the background and rationale section for more detail), so the majority of the issues raised need to be reassessed also considering this aspect. 8 https://finance.ec.europa.eu/news/commission-proposes-postpone-one-additional-year-market-risk-prudential-re￾quirements-under-basel-iii-2025-06-12_en

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 20 A respondent provided a general comment about the possibility to make reference, when report￾ing the columns historical default rate / loss rate of the last 5 years, to shorter period of time. The EBA confirms that, in case the competent authority has authorised an institution's model based on a number of years less than 5, it seems legitimate that the institution can refer in its reporting starting from the years used for the development of the model. A respondent answered to the question CR 1 confirming the appropriateness of the proposed so￾lution to avoid possible double-counting of the retail exposures but also making some suggestions aimed at further improving the quality of the reporting. Such suggestions will be kept in consider￾ation when drafting the validation rules for the IST 2027.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 21 ▪ Summary of responses to the consultation and the EBA’s analysis Comments Summary of responses received EBA analysis Amendments to the proposals General comments A respondent has several concerns and suggestions. The respondent reiterates concerns about reduced participation in the Internal Models Approach Benchmarking post-FRTB go-live. They suggest a stepwise approach starting with IR and FX for 2026. Moreover, there is the concerns on the expansion of the EBA mandate to include FRTB Standardised Approach will increase participation, posing operational challenges. The respondent proposes using ISDA’s capital models benchmarking platform to support this exercise (ISDA Analytics to handle the increased data and supervisory workload). The respondent recommends providing greater detail in all aspects of the HPE, including SbM Valuation portfolios, to minimize errors and improve data quality. Finally, they welcome the EBA's recognition of diminishing returns from recurring SBM Validation portfolios and suggest updating the ITS to allow for bilateral consultation with supervisory teams. Beside the pose on the expansion to the institutions in the scope of article 78.1.b of the CRD, the EBA has restricted substantially the volume of sub￾missions for IMA banks, in view of the FRTB go-live and the reduced application of internal model. EBA and supervisors are aware of the operational challenges on the expansion of the scope, and they are actively operating to minimize it (with the proposal to provide training and keeping the frame￾work stable, i.e. not changing substantially the composition of the portfolios). ISDA Analytics remains to be tested, and for the moment evidence that the use fit the EBA benchmarking exercise are lacking as well as on how it could simplify the process for the supervisors. Further facilitation in the SbM Valuation will be considered once this part of the exercise will be considered to be stable and having exhausted its primary purpose.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 22 A respondent proposed the following general comment. We would like the EBA to clarify the historical default rate / loss rate of the last 5 years for exposure classes where a change of approach has been experienced during that specific period (A-IRB to F-IRB, for instance). In these specific cases, we consider that a reasonable option in order to avoid excessively burdensome criteria on changes of approach already authorized by the ECB, would be to allow banks to report only the period of time during which the new approach has been applied. Can the EBA confirm this approach is valid. Article 1801 (1) (h) of the REGULATION (EU) No 575/2013 (CRR) states <<irrespective of whether an institution is using external, internal, or pooled data sources, or a combination of the three, for its PD estimation, the length of the underlying historical observation period used shall be at least five years for at least one source.>> Article 1801 (2) (h) CRR reiterates the same requirement for the case of Retail exposures. Article 181 (1)(j) of the << for exposures to corporates, institutions and central governments and central banks, estimates of LGD shall be based on data over a minimum of five years, increasing by one year each year after implementation until a minimum of seven years is reached, for at least one data source. If the available observation period spans a longer period for any source, and the data is relevant, this longer period shall be used. >> Arti￾cle 181 (2)(c) CRR reiterates the same requirement for the case of Retail exposures. These Articles seems to imply that at least five years of data should be available when validating a PD and an LGD model. However, Article 181 (3) states << […] a competent authority may permit and institution pursuant to paragraph 3 to use relevant data covering a period of two years when the institution implements the IRB Approach. >> Therefore, if a competent authority has authorised an institution's model even if it is based on a number of years less than 5, it seems legitimate that the institution can refer in its reporting to the period of time during which the new approach has been applied. Alternatively, the institution could perform the calculation of the 5Y-DR on a best effort basis, using a proxy for the retrospective identification of the portfolio at the past reference dates.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 23 Responses to questions in Consultation Paper EBA/CP/2025/03 Question MR 1. Do you see any issues or any missing information that should be required in the new templates suggested for the AIMA FRTB bench￾marking exercise (i.e. Annex 6 & 7)? Template 108 A respondent suggested that the template 108 requires 10d P&Ls and references 130.05. However, template 130.05 requires 1d P&Ls. This discrepancy should be reviewed and re￾solved. Another respondent suggests incorporating standardised metrics that capture operational and non-financial risk factors—such as system failures, data errors, and governance issues— which can significantly affect trading book risk profiles and model outputs. The template 108, as modified in the CP was re￾verted to the previous template because of the FRTB delay – no changes to the previous frame￾work because of this suggestion. EBA recall that the primary objective of the benchmarking exercise is to assess the compa￾rability and performance of internal market risk models framework. Expanding the templates to include operational events, governance issues, or non-financial risk metrics may introduce sub￾jective and non-standardised elements that could compromise the consistency and clarity of benchmarking outcomes. EBA believes that op￾erational and data governance factors areas are better addressed through separate supervisory processes and frameworks dedicated to internal controls, operational resilience, and data quality management. Template 108 as proposed in the CP removed, in place of the previous version of the same template.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 24 Question MR 2. Do you see any issues with the reduced subset of instruments proposed for the AIMA exercise? Please elaborate? Scope of the IMA data col￾lection A respondent appreciates the EBA’s reduced scope for the HPE portfolios and instruments in the AIMA exercise but reiterates concerns from last year’s consultation. Specifically, there’s a risk that post-FRTB go-live participa￾tion may fall below levels needed for meaning￾ful benchmarking. Due to the significant oper￾ational burden and uncertain feedback, the in￾dustry recommends a stepwise approach starting with Interest Rates (IR) and Foreign Exchange (FX) asset classes in 2026. Additionally, the respondent highlights issues in the proposed reduced instrument set. In the IR asset class, instrument 204 has shown high volatility, and instrument 205 is non-standard. They suggest removing such non-standard in￾struments and replacing them with more va￾nilla ones, such as instrument 207 (a long Ger￾man Government bond). In the FX asset class, instrument 303 (a cash position) is considered of limited standalone value. Another respondent suggests that standard￾ized tracking of non-financial risks—such as The burden for AIMA banks is already quite lim￾ited due the reduction of the scope of the data collection. While the low number of banks in the exercise could impact the production of public feedback, such as the EBA Benchmarking report, the provision of usable feedback to banks is still a goal of the exercise. The delayed implementation of the FRTB to 2027, has now changed the setting of the 2026 data collection, and the scope have been adapted to that, with the focus on the ASA fig￾ures from the banks subject to IMA approval. EBA in concerned for that introducing metrics related to system limitations, data feed depend￾encies, or operational vulnerabilities may intro￾duce variability and subjectivity that are incon￾sistent with the benchmarking exercise’s objec￾tive of consistency and comparability across in￾stitutions. The data collection for the IMA banks was restricted to ASA data – see Section 3-5 of amending Annex 5

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 25 system limitations or data feed issues—linked to specific instruments could provide valuable transparency. They recommend the EBA con￾duct regular reviews of the instrument subset to ensure it remains relevant and reflective of market and risk management realities. Question MR 3. Do you see any issues with the new template 106.02? Please elaborate. Issue on Annex X A respondent suggests that the Annex X, which defines portfolios from last year's exer￾cise, needs to be republished due to the inclu￾sion of new banks and existing inconsistencies. Issues include: • Duplicate use of a non-existent instru￾ment identifier (“S_FXD_e1#”) in portfolios F007 and F023. • Inconsistent definitions for curvature sensitivity inputs across portfolios (e.g., C080 vs. C062), causing confusion. Another respondent suggests the introduction of standardized operational exposure metrics, such as Risk Units (RUs), to further strengthen Issues highlighted with Annex X were ad￾dressed, removing the non-existent instrument and the inconsistency sensitivities. Introducing qualitative operational disclosures within the benchmarking framework may intro￾duce heterogeneity that challenges the tem￾plate’s core purpose of standardisation and ob￾jective comparison. We believe that operational risk, data quality, and governance considera￾tions are best addressed through other supervi￾sory channels that are specifically designed to evaluate internal control environments and data integrity. The Annex X was added to the final ITS for resubmis￾sion.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 26 the consistency and credibility of sensitivity re￾porting. Question MR 4. Do you see any issues with specifying the specified timeline in the Annex 5 or with the reference date for new ASA institution in the exercise as defined in the suggested draft of Article 4.1.(b)? Issue with the timeline – ad￾ditional time for submission of the data A respondent proposes extending the time be￾tween reference and remittance dates for all parts of the exercise, citing the new mandate and the need for thorough internal validation by participating firms. Specific suggestions in￾clude: Keeping a three-week gap between the first reference date and remittance, as in the 2025 exercise. Extending the gap between the last reference date and remittance to two months. Decoupling the submission of SBM validation portfolios from IMV results, preferring align￾ment with the full results submission date in￾stead. Another respondent suggests encouraging in￾stitutions to disclose any significant opera￾tional challenges or data reconciliation issues Since the expansion of the scope of the bench￾marking exercise was posed the rationale to postpone the deadlines for submission as sug￾gested was lacking. On the other side, given the migration to DPM2.0, institutions will start reporting in csv instead of xml, therefore the references date of the exercise were adjusted to allow banks more time for this transition. Therefore, the timeline was postponed of 1 months for IMV references date, and 2 weeks for Risk Measures refences dates. Regarding the proposal to incorporate supple￾mentary metrics for operational risk granularity, EBA feels they are better addressed outside the current market risk benchmarking framework. Please consult the amending Annex 5 – Section 1 letter b

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 27 faced during adherence to these timelines, as this would improve transparency and help dif￾ferentiate between model and execution dis￾crepancies. Question MR 5. Do you see any issues with the changes introduced in the Annex 5? Scope definition allows for interpretation and requires more guidance. A respondent recalls that the scope is defined via approved IMA desks. They seek for guid￾ance whether this considers PLAT results. In other words, the approval could change over time. In such case we propose to assess ap￾proval at a particular date, e.g. IMV reference date and keep the scope fixed afterwards. ‘Institutions in the scope of the ASA submis￾sion shall report solely instrument that are not forbidden to trade by internal policy decision or trading system limitations, and Institutions in the scope of the AIMA submission shall re￾port solely instruments that are not forbidden to trade by internal policy decision or trading system limitations and are traded on an AIMA approved desks.’ Because of the FRTB delay –the previous frame￾work remains in place; therefore this suggestion is no longer applicable. On introducing qualitative operational risk dis￾closures or supplementary metrics this does not seem consistent with the objectives of the exer￾cise. Instruction re￾verted to the pre￾vious framework, so the proposal in the CP in no longer suggested.

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 28 Another respondent suggests encouraging in￾stitutions to include brief operational risk dis￾closures alongside their submissions—such as qualitative notes on significant events or op￾tional quantitative indicators. Question MR 6. Would you consider it useful to clarify the type of SOFR rate (term, compound) to be used when booking related interest rate instru￾ments? If so, please suggest a clarification. SOFR clarification The respondents are supportive of further clarifications and propose to align with indus￾try best practice, i.e. using “overnight SOFR, daily compounded”. This would be useful given the increase in scope of participating firms who are inexperienced in the bench￾marking exercise and therefore the greater clarity provided in the definition of the HPE in￾struments the less likely to see variation from benchmarking process errors coming from booking differences. EBA agrees with the suggestion. Instruments with SOFR references were amended ac￾cordingly to the suggestion Question CR 1. Do you think that the proposed approach aimed at including the breakdown B.6.3 is correct and it enables to avoid any double count￾ing of the exposures? Template 103 A respondent supports EBA’s initiative to refine credit risk exposure reporting through the in￾troduction of breakdown B.6.3. We agree that The ITS 2026 is meant to last only for one year and to be substituted by the ITS 2027 where No changes

FINAL REPORT ON ITS (EU) 2016/2070 WITH REGARD TO THE BENCHMARKING OF INTERNAL MODELS (2026 EXERCISE) 29 a clearer and more granular separation of ex￾posures is essential to avoid double-counting and to ensure consistency in benchmarking in￾ternal credit risk models across institutions. From a risk aggregation and data governance standpoint, we respectfully observe that even with well-designed templates, exposure over￾laps or duplications can still arise due to opera￾tional issues. To further strengthen the EBA’s approach and minimize residual risks of double counting, we propose that institutions be encouraged to ap￾ply robust operational reconciliation proce￾dures between credit portfolios and reported templates. more fundamental changes will be introduced. The suggestions provided by the respondent are not easily translatable in terms of instructions however, it will be explored the possibility to in￾troduce Validation Rules inspired by such hints.