Overview of the ECB’s Final Revised Guide to Internal Models

Introduction

The European Central Bank (ECB) published its final revised Guide to internal models, following a public consultation which ended in September 2023. In light of the changes from the 2019 version, this article provides a summary of the new requirements banks will need to address regarding overall internal models along with credit, market, and counterparty credit risk models.

Summary of key points

General topics:

1. Institutions should include material climate and environmental risks in their PD and LGD models.

   • All missing or inaccurate information should be compensated with a margin of conservatism (MoC).

   • If material climate related information is not included in the rating system, then a more conservative approach should be taken, applying an override on the final rating assignment.

2. Institutions with the intention of reverting to the standardised approach should support their request with rationale, chosen approach, evidence that there is no intention to reduce the own fund requirements (even if they reduce), justification of not using slotting approach (in case of specialised lending). Requests should be consolidated if several rating systems are affected.

3. Use test requirements for initial permission should include evidence that the rating systems have been used for risk management for at least 3 years and, in case of extension, the compliance with the use test should be evidenced, if:

   • The exposures are significantly different from the scope of existing coverage;

   • The exposures are not significantly different from the scope of existing coverage, to calculate RWAs at individual level for a new legal entity.

4. Additional data requirements in case of consolidations must be followed, including:

   • In case of extension to a new entity, both default and loss history should be used to calibrate the IRB models.

   • Where workout processes are different, the data could be excluded, but the same should not be the case for defaults.

   • In case of not being able to access the default and loss histories of the acquired portfolios, that should be proved and signalised as a data deficiency with the required adjustment and MoC.

   • Institutions need to analyse and provide evidence of adequacy and where there are deficiencies or bias, they need to take further steps, such as adjustments and MoC.

Credit Risk

1. Institutions must follow definition of default in accordance with Article 178 of the CRR.

2. When an institution performs a rating transfer across different rating systems that do not share the same obligor rating scale, the mapping between rating scales is performed in such a way that the final PD estimate (including MoC) assigned to the guaranteed exposure amount is not better than the final PD estimate (including MoC) being transferred from a third party.

3. To offset massive disposals’ effect from the LGD and CCF, institutions are subject to the following requirements:

   • The deadline to require the adjustment has already passed, but current adjustments might still be changed.

   • The disposal date corresponds to the date of legal ownership transfer.

   • The effect should be estimated as the difference between the average LGD estimated for similar facilities which were not liquidated, and the average realised LGDs.

   • The average LGD estimated for similar facilities can be calculated based on the institution’s incomplete workout treatment applied to the exposures as of the date before the date of their disposal.

   • Additional observed defaults that are not part of the qualifying subsidiaries may not be adjusted.

   • Once most of the cases that were incomplete as of the date of the disposals have been closed or if the maximum period of the recovery process has been reached as of the time of the estimation.

   • If the inclusion of the additional defaults would wrongly inflate the maximum recovery period, then steps should be taken to use a more appropriate and shorter period.

   • Because of the necessary conservativeness related to the available data being unsatisfactory, a minimum value of 100% be applied as a final CCF estimate.

   • The margin of conservatism should not affect rank ordering, while ensuring monotonicity in their final estimates while still reflecting the uncertainty at grade/pool level.

Market Risk

1. Counterparty credit spread risk cannot be included in the scope of the IMA and is not part of the actual or hypothetical profit and loss (P&L) for back-testing.

2. Instruments in the regulatory trading book that are lent out or REPO’ed should be included in the calculation of own fund requirements for market risk.

3. Instruments borrowed via securities lending or reverse REPO should not be included in the calculation of own fund requirements for market risk.

4. The funding risk embedded in own liabilities held in the trading book should be modelled in the IMA, and that model must accurately capture all material price risks.

5. Both the general and specific risks of the CVA hedges are in scope of the IMA and are subject to its requirements.

Counterparty Credit Risk

1. The management and senior management should be involved in approving any limit change resulting from the prior implementation of an IMM model change or extension in the live production environment.

2. Institutions should start weekly parallel runs with the application, but before sending the application letter to allow internal validation to perform their analyses.

3. Several runs not only support testing but also allow for a more comprehensive view of impacts.

What banks should be doing and possible implications

• Institutions will need to incorporate climate and environment-related risks if deemed material. However, the materiality of these risk drivers is expected to be low, and the impact in capital requirements is difficult to predict at this point.

• The clarifications for massive disposal treatment will have positive effects, decreasing capital requirements through the offset of disposal effects on the LGD.

• Regarding the definition of default, it is expected institutions with internal models have already incorporated the definition present in the Guide, hence its impact is expected to be negligible.

• Although a return to standardized approach allows for simplification of models, the requirements for that change must result in an increase of capital requirements. Bear in mind IRB institutions should have an IRB coverage of at least 50% the exposure after concluding roll-out, hence, there is a limit to simplification while maintaining internal models for capital requirements.

• Clarifications on Market Risk and Counterparty Credit Risk reinforce the need for an in-depth knowledge of internal models and their impact in capital. This will, in turn, require additional resources to obtain and manage the additional information needed, increasing their maintenance cost. This ties in with the additional scrutiny and need for transparency regarding model risk management present in SS 1/23.

• Remaining clarifications point to tighter requirements for the implementation and use of internal models in risk management. More specifically, institutions will need to improve the standards on which technological environment are developed for internal models’ implementation, thus increasing costs of setting up and maintaining internal models.

If you’d like to learn more about how 4most can support your organisation, contact us – info@4-most.co.uk.