FAQ on the Targeted Review of Internal Models

What was the targeted review of internal models? What was its main goal?

The targeted review of internal models (TRIM), launched by the ECB at the beginning of 2016, was the largest project conducted by ECB Banking Supervision in coordination with national competent authorities (NCAs) to date. We assessed whether the internal models used by banks under direct ECB supervision complied with regulatory requirements, and whether their results were reliable and comparable.

Banks can use internal models to determine their total risk-weighted exposure amount. This is also referred to as risk-weighted assets (RWAs). Banks need permission from their supervisor to do so.

The main goals of TRIM were:

1. to ensure compliance with regulatory requirements and enhance the comparability and credibility of internal models;
2. to reduce unwarranted (i.e. non-risk-based) variability of RWAs;
3. to harmonise supervisory practices.

One important outcome of TRIM is the ECB guide to internal models. It helps to make sure that banks use internal models appropriately and consistently. The guide outlines the ECB’s understanding of selected applicable regulatory requirements for internal models. It also provides the basis for the common assessment methodologies that have been applied within TRIM.

The objectives of TRIM were in line with two major goals of ECB Banking Supervision:

1. to foster a sound and resilient banking system through proactive supervision and the use of best practices;
2. to ensure that supervisory practices are applied consistently across the euro area.

What were the main findings from the TRIM exercise?

Overall, the TRIM investigations confirmed that supervised entities may continue to use their internal models to calculate RWAs. TRIM identified numerous deficiencies (“findings”) across all risk types, that banks must remediate within deadlines imposed by the ECB. They need to continue to invest in maintaining the high quality of models achieved during TRIM.

While findings are in the process of being remediated, some models needed supervisory backstops to address the potential underestimation and high uncertainties of RWAs (such as some models for the calculation of loss given default (LGD) and conversion factors related to low-default portfolios).

We identified a number of findings per risk type during TRIM and about one-third of these had an estimated material impact on banks’ level of own funds requirements, internal governance or risk control and management. For example, for credit risk models, in almost all investigations we identified weaknesses with a high or a very high impact for the LGD parameter. For market risk models, we found deficiencies with a high or a very high impact in relation to the Value at Risk and the stressed Value at Risk methodology employed.

The findings per risk type identified through the TRIM can be found in Section 4 of the TRIM project report.

Has the exercise resulted in higher overall risk-weighted assets?

TRIM investigations were comprehensive both in breadth and depth. They resulted in numerous findings and a detailed supervisory follow-up. The supervisory follow-up focuses on the remediation of those findings to restore full compliance with the regulatory requirements.

Banks have received detailed and binding obligations with pre-defined deadlines. Additionally, we have imposed limitations to ensure a proper level of own funds requirements during the time banks are remediating findings and until they have reached compliance with regulatory requirements. We estimate that the aggregated impact of those limitations and the model changes approved as part of TRIM led to an overall absolute increase in RWAs of about €275 billion over 2018-2021. However, we cannot entirely estimate the impact of the supervisory decisions issued as part of TRIM in advance.

In fact, for obligations, which are at the core of the TRIM follow-up, the impact on capital requirements cannot be estimated in advance, as it ultimately depends on how each institution decides to address its specific obligations in practice.

Did TRIM reduce unwarranted (i.e. non-risk-based) variability of risk-weighted assets? If so, how?

TRIM reduced non-risk-based RWA variability in various ways, as this variability can arise from different sources, e.g. from regulation not being sufficiently specific to overcome such variability, or from institutions not complying with regulatory requirements.

1. The ECB guide to internal models provides transparency on the ECB’s supervisory understanding of existing regulation. It ensures consistency in the implementation of regulatory requirements and supports a harmonised assessment of internal models. In the context of TRIM, the ECB consistently applied its intrusive assessment methodology to all banks (particularly using standardised data requests and common inspection techniques and tools).
2. We issued supervisory decisions as a result of TRIM investigations following a consistent approach. These decisions include measures to bring the investigated models fully in line with regulatory requirements. During the remediation process, the measures compensate for any underestimation of risk. This way we ensure that the supervised entities remediate the observed deficiencies swiftly and do not underestimate RWAs during the remediation phase.

These measures ultimately strengthen the link between the risk of the underlying assets and the outcome of the model, and they contribute to the comparability of model outcomes. The measures therefore reduce non-risk-based RWA variability and contribute to restoring the credibility of internal models.

How has TRIM benefited future European supervision of internal models?

TRIM has delivered a range of benefits and results beyond assessing the compliance of internal models with regulatory requirements.

• Supervisors have gained a much deeper, system-wide, knowledge of existing modelling practices and related shortcomings, which will also help them to define areas for future investigation or monitoring.
• The ECB guide to internal models complements regulatory initiatives in the field of internal models from a supervisory perspective. The guide provides transparency on how the ECB understands the applicable regulatory requirements for internal models. It also contributes to sounder internal model frameworks and assessments within European banking supervision.
• TRIM fostered the development of a consistent approach to the supervision of internal models in use at significant institutions under European banking supervision. A similar approach for future internal model supervisory assessments will be retained. This will further contribute to ensuring a level playing field on an ongoing basis.

Going forward we will embed well established TRIM methodologies, supervisory practices and in-depth knowledge of the modelling landscape in the two pillars of regular internal model supervision within European banking supervision: internal model investigations and ongoing model monitoring.

In what sense was TRIM “targeted”?

TRIM was targeted in the following ways.

• It covered the most relevant risk types, while also taking further regulatory developments into account. The project covered models for credit, market and counterparty credit risk. Internal models for operational risk and credit valuation adjustment risk were excluded from TRIM. This is in line with the Basel Committee’s stance, as part of the reforms for finalising Basel III, that banks should not use model-based methods for these risk types.
• It covered the most significant areas in which unwarranted RWA variability was deemed to exist, or where a diverse array of practices existed which may not be in line with regulatory requirements.
• In the case of credit risk, it reviewed a sufficiently large number of the most material and critical internal models. This was subject to a proportionate use of time and resources, given that a full review of all existing models would not have been feasible within the scope of the project (unlike for market risk and counterparty credit risk models).

Did TRIM affect all banks directly supervised by the ECB?

We assessed the internal models at all directly supervised banks that used them at the date of the launch of TRIM. However, given the targeted nature of the project, and for reasons of proportionality and logistics, not all approved models at all banks could be checked. Moreover, we excluded from the review banks undergoing a merger and those which were no longer subject to direct supervision.

Similarly, we could not include in the scope of the project all institutions that received an initial model approval while TRIM was ongoing. Finally, we did not include less significant institutions (LSIs), which are directly supervised by NCAs, within the scope of the TRIM project (unless they had received a TRIM investigation before becoming LSIs).

Overall, a total of 65 banks were in the scope of TRIM.

How prevalent is the use of internal models? Is this expected to change as a result of TRIM?

With TRIM, we did not intend to persuade or dissuade banks from using internal models, but rather to assess how adequate the models in use were. Nonetheless, TRIM, together with the changes envisaged through the finalisation of the Basel III standards, may have consequences for where and how internal models are used by banks. TRIM will contribute to streamlining banks’ model landscapes. In particular, considering the investment needed to comply with the standards required in TRIM in line with the regulation, banks should reflect on their internal model strategies to inform future internal model developments or simplifications.

What was the timeline of the project? How many on-site investigations were conducted?

We started the project in 2016 by developing the underlying methodology and necessary tools as well as by identifying the models to be reviewed. Between 2017 and 2019, we conducted 200 TRIM on-site investigations. 161 of those investigations were for models related to credit risk, 31 for market risk models and 8 for counterparty credit risk models. In 2020, we concluded the horizontal analyses of the outcome of investigations and continued to issue supervisory decisions. In April 2021 TRIM was finalised with the publication of a project report.

What were the different phases of the TRIM project?

TRIM investigations had distinct phases of activity and quality assurance layers. Those were in line with the standard life cycle of internal model investigations, as described in the ECB guide to on-site inspections and internal model investigations:

1. defining the scope and preparing for the investigation;
2. conducting the on-site activity;
3. conducting off-site consistency checks;
4. finalising the investigation phase and sending the final assessment report to the bank;
5. following-up on the findings identified during the on-site investigation via the supervisory decision process.

The consistency checks ensured a consistent application of the common methodology developed for TRIM. We conducted horizontal analyses across the same risk type or categories of exposures to deepen our understanding of the existing modelling landscape in the respective areas. This provided an overview of the most recurrent or critical findings and ensured that supervisors could address similar shortcomings consistently.

How are the objectives and outcomes of TRIM related to other internal model initiatives, for example the EBA regulatory review of the internal ratings-based (IRB) approach or the Basel III standards?

Both TRIM and the Basel III standards seek to reduce the excessive unwarranted variability of risk-weighted assets. TRIM is addressing the issue from a supervisory perspective. The Basel III standards are addressing the issue from a regulatory perspective.

The implementation of the final Basel III standards alongside the remediation of TRIM findings together provide a safeguard against inappropriate or inadequate internal models. The restrictions and conditions that are contained in the Basel III standards are applied on a more general level (Basel III entails, for example, parameter floors or restrictions on the use of the advanced internal ratings-based approach). TRIM investigations on the other hand focused more on modelling-related aspects.

Both the EBA repair programme and TRIM seek to improve the adequacy of internal models with respect to regulatory requirements. For example, the understanding of the applicable law laid out in the ECB guide to internal models was developed in alignment with ongoing regulatory developments, such as EBA regulatory technical standards and guidelines.

The TRIM exercise has not been designed to frontload upcoming regulatory developments. But the implementation of Basel III standards, through the amendments to the Capital Requirements Regulation and the regulatory review of the IRB approach led by the EBA and the remediation of TRIM findings should provide complementary layers of safeguarding against inadequate internal models.

How were national competent authorities involved in the project?

Involving different NCAs was a key principle of the TRIM project. The aim was to promote consistent supervisory practices and to leverage on the wide span of internal model expertise across European banking supervision.

For example, NCAs have been involved in:

1. executing TRIM on-site investigations (the majority of which were staffed by NCAs);
2. the operational steering body of the project;
3. sending internal model experts to the TRIM centres of competence that were responsible for developing the TRIM methodology and conducting horizontal reviews of the TRIM assessment reports.