ECB consults on revised guide to internal models

July 2023

On 22 June 2023, the ECB launched a public consultation on its revised guide to internal models (PDF 2.63 MB). The revised version is similar in its structure to the original guide of 2019, containing an opening chapter on general topics and one chapter each on credit risk, market risk and counterparty credit risk. It aims to:

• Reflect recent and ongoing updates to regulatory requirements
• Learn from the supervisory experience of the past three years
• Provide additional clarification where required
  
  

Banks can submit comments on the revised guide until 15 September 2023, after which the ECB will release a feedback statement along with the final version of the revised guide. 

We now summarise some of the key revisions contained within each chapter.

1. General topics

• Where relevant and material, climate-related and environmental (C&E) risk drivers should be included in internal models and approved for use in calculating own funds requirements (OFRs) for credit and market risk.
• When changing or extending a model, institutions are generally expected to achieve implementation within three months from the date of notification, with exceptions only likely to be granted in certain circumstances (such as a joint implementation, or due to specific technical limitations).
• When institutions using the IRB approach wish to revert, under Article 149 of the CRR, to the standardized approach (SA) or the foundation-IRB (F-IRB) approach key activities should include:
  • Documenting the reasons or impediments leading them to reconsider using the IRB approach
  • Defining and formalising objective and intuitive criteria for deciding which approach to use for the calculation of OFRs
  • Providing convincing evidence that there is no intention to reduce OFRs, where reversion leads to a non-negligible reduction of capital requirements
  • Submit a single comprehensive request for all affected ratings systems

• When a consolidation (M&A) occurs, the revised guide includes new content on the use of Pillar 1 internal models for credit, market, and counterparty credit risk. In particular, if institutions need to submit a ‘return to compliance’ plan regarding all consolidation-related compliance issues, this should:
  • Clarify the strategy to restore compliance
  • Set out the target internal model landscape of the post-merger legal entity, including actions and timelines that banks plan to carry out to achieve the target internal model landscape
  • Describe the process for calculating RWEAs until the return to full compliance

2. Credit risk

• New expectations for institutions to provide evidence of IT readiness when applying for initial model approval (in a live production environment), and IRB roll-out or material model changes (in a non-live production environment) involving credit risk.
• Additional expectations around a range of technical points, including:
  • Definition of default (DoD) – consistency across banking groups; days past due and unlikeness to pay criteria; returning to non-defaulted status; consistency of external data; and adjustments to risk estimates in the event of changes to the DoD
  • PD risk quantification – including further principles and calibration tests regarding the calibration process of the long-run average (LRA) default rate
  • Treatment of massive disposals under article 500 of the CRR – such as using the incomplete workout approach when making adjustments to LGD calculations
• Details around including C&E risk drivers in credit risk parameter calculations, potential use of overrides, and the need for Margins of Conservatism (MoCs) to consider any deficiencies stemming from missing or inaccurate climate-related information.

3.  Market risk

• Clarification that instruments in the regulatory trading book that are lent out or repo’ed should be included in the calculation of OFRs for market risk under the internal model approach (IMA), but that instruments borrowed via securities lending or reverse repo should not since they do not transfer the market risk of the security.
• Further clarification over the use of probabilities of default (PD) in the incremental default and migration risk charge model.

4. Counterparty credit risk

• Clarifications of specific counterparty credit risk elements, such as the ECB’s understanding of the term ‘most recent exchange of collateral’ contained in Article 272 of the CRR.
• New indications regarding the ECB’s views of:
  • Applying use test requirements to model changes and extensions upfront
  • Identifying, monitoring and capturing ‘risks not in effective expected positive exposure’ (I.e., quantifiable risks not captured or inadequately captured by the IMM) 

How should banks respond to the ECB’s consultation on its revised guide to internal models?

First and most obviously, institutions should review the revised guide in detail and provide the ECB with any comments or suggestions by 15 September. Second, even though it is not clear when the final guide will be published, banks begin preparing for altered supervisory expectations.

In our view, six key areas of focus are likely to include:

• C&E risk drivers: Banks should ensure they can assess the materiality of C&E risks for each rating system and, if material, reflect these in the credit risk parameters through the use of appropriate risk drivers.  Where data is missing or inaccurate, this should be reflected in adequate MoCs.
• Reverting to SA or F-IRB: Institutions should review their internal models’ strategy and define objective criteria, identifying – where needed – any opportunities to revert to less sophisticated approaches for calculating OFRs.
• Implementing new or altered credit risk models: Banks should prepare to make non-live parallel IT environments available, in order to verify that systems are ready to implement new or changed models before actual implementation. Institutions should also define clear implementation roadmaps for meeting the expected timeframe of no more than 3 months from approval.
• Definition of default: For changes to the DoD, institutions should be ready to perform impact assessments on the data and on the model’s risk differentiation, by performing a granular retrospective simulation or with a parallel run of the new criteria going back at least 2 years. Banks should anticipate challenges from IT changes, data reconstruction, and other practical obstacles.
• Calibrating credit risk models: Banks should review their processes for estimating LRA default rates, and plan how to calculate them at both grade and calibration level. This could have a major impact on internal systems, procedures and data management. Where M&A         occurs, banks should use historical data from both entities for model calibration.
• IRC model for market risk: Banks should anticipate greater scrutiny over the use of PDs in IRC models, and be ready to demonstrate, on the basis of objective data and information, the appropriateness, accuracy and consistency of PD estimates within daily risk management processes. In case of differences between PD estimates methodology and expected losses methodology, institutions should document the underlying rationale and its approval by an appropriate management body.
• IMM for counterparty credit risk: Banks should implement comprehensive frameworks for the identification of trades involving either illiquid collateral or an OTC derivative that cannot be easily replaced and for the identification, monitoring and capitalisation of risks not in the effective expected positive exposure.