Following the financial crisis of 2007-2009 doubts had emerged on the relevance of the internal models to calculate Risk-Weighted Assets (RWA). The minimum capital required to absorb losses corresponds to a percentage of the RWA. To calculate the RWA, different type of risks (credit risk, market risk and operational risk) are taken into account and each asset class has different risk weights associated with them. As we can see in the scheme below, to calculate these risk weights, banks have the choice between the internal models or the standardized models. The internal models are composed of the Internal Rating Based (IRB) Approaches for credit risk, Advanced Measurement Approaches (AMA) for operational risk and Internal Models Approaches (IMA) for market risk. The internal models were created as part of Basel II, replacing the original Basel Accord of 1988 (Basel I) in an effort to create a better framework for regulating bank capital. Under Basel II and Basel III, banks were allowed to use these internal models for the calculation of credit and market risk exposure and corresponding regulatory capital. To be allowed to use the internal models, banks had to receive express permission from their regulators. They were eligible to the internal models when their rating and risk management systems were sound and had been implemented with integrity and subject to the bank complying with integrity and subject to the bank complying with various systems, controls and corporate governance requirements. Basel III revisions (informally known as “Basel 4”) are now dealing with the review of the internal model approach.
Fig1: Overview of Basel III approaches
The implementation of the internal model approach can be costly and is particularly advantageous for banks that have already an enterprise risk management system in place. Banks that estimate that internal models is the more appropriate approach for them, needs to request the authorization of the regulator to implement it. This approach allows to take into account the benefit from risk mitigation strategies that are implemented and have more flexibility on the calculation of the RWA. The problem is that the choice of the approach to use is mainly based on which method requires the least capital rather than the most appropriate level of capital for the risk a firm is taking. The internal model approach has been suspected to allow big banks to reduce the amount of capital need. According to regulators a majority of banks using the internal model approach would have to hold a bigger amount of capital if they used the standardized models as most smaller lenders. TRIM has been implemented in response to these doubts.
Reasons behind the Targeted Review of Internal Models
These last years, internal models have been under pressure as financial markets and supervisors claim to have lost confidence that banks are fully equipped to adequately assess their risk exposure and the corresponding required capital. Several studies have exposed unwarranted variability in RWA, financial markets and supervisors. This variability can be explained by all stages of the modelling and reporting process due to a lack of independence of (validation) functions, outdated and incomplete data and IT systems, variations and inconsistencies in key regulatory definitions and general methodological differences.
As a response to these problems, the ECB launched a review of the internal models, called the Targeted Review of Internal Models (TRIM). The preparatory phase began in 2016 followed by the on-site inspections during 2017-2018 and the follow up is intended to be finalized at the end of 2019. The TRIM project envisages about 200 investigations in 65 significant institutions across 15 countries. The aim is to enhance the credibility and confirm the adequacy and appropriateness of approved internal models permitted for use by significant institutions when calculating own funds requirements. Based on the Capital Requirements Regulation (CRR), the Capital Requirements Directive (CRD IV), the relevant Commission Delegated Regulations and Commission Implementing Regulations, the Regulatory Technical Standards (RTS), the European Banking Authority (EBA) guidelines, and the approved European Central Bank (ECB) Banking Supervision manuals and guidelines, TRIM will ensure the compliance of internal models to the regulatory requirements. And it will focus on delivering interpretation of the CRR and address current gaps of the interpretation of the regulation.
What is expected of TRIM?
Review of the internal models
The TRIM project is organized in different steps. The first step is to establish a common, standardized and methodological procedure for the on-site investigations. So, the ECB published a guide which lay out the framework to follow. The key objective is to be transparent and clear on how the ECB understands the regulatory requirements and how it will control that all banks meet these requirements. Common inspection techniques and tools are established as well as pre-defined areas of investigation. And to ensure that the inspection techniques and tools are applied consistently, a close interaction between the assessment teams and the risk-specific team is required during the on-site investigation.
The second step of the quality assurance consists of the checking of the report produced by the assessment team to ensure that similar shortcomings give rise to similar findings. The check is done by internal models’ experts from different National Competent Authorities (NCA) and the ECB, who have a horizontal view of the TRIM investigations for each risk type.
The final step consists in the follow-up decisions based on the findings of each on-site investigation. Every decision that is taken must be supported by regular exchange and alignment sessions for the model experts, which have access to a comprehensive overview of past cases. These decisions are also challenged and reviewed within the ECB before being sent to the banks.
An important expectation in the TRIM review concerns the estimation of PD and LGD. In fact, these two terms must be well defined, as different banks have different approaches to define “default”. The historic loan performance information of IRB bank will be compared to the definition of “default” used by the relevant member state regulator. PD is the probability of an obligor defaulting on its contractual obligations within one year and LGD is the estimate of loss that a bank will incur if its obligor defaults. In contrast with PD, LGD must be based on a conservative view of long-run averages.
Internal model investigation
TRIM will investigate if a bank meets the minimum requirements to use the internal model approach and the on-site investigation will aim to reduce inconsistencies and unwarranted variability when banks use internal models to calculate their RWA.
The on-site investigation will focus on the compliance of the model requirements, examine the portfolio on which the internal model is applied and examine the adequacy and adaptability of business processes related to the model. It will also assess the model’s economic appropriateness, the performance of the model’s IT infrastructure. Finally, it will compare model outcomes where feasible. The on-site investigation process is depicted in the graphic below:
What are the current outcomes of the TRIM investigation?
Following the on-site investigations, the ECB published the initial findings of the review of the internal models. There is a first part of the findings on the outcome of the governance review. The governance review concerns the review of principles overarching different risk types and internal models’ framework. Then, there is a second part on the outcome of the credit risk investigation.
Concerning the governance review, a certain number of shortcomings were identified, such as the absence of a model change policy or a lack of evidence of annual back-testing for some rating systems. The investigation also puts into light potential misalignments regarding the implementation of a model risk management framework. Despite some measurement of model risk and partial controls in place, practices were not formalized or documented and there were no model risk management in place.
Institutions should have a procedure and policy in place to ensure compliance with the requirements for permanent partial use (PPU). Also, the regulation requires that the criteria used to determine the application of the IRB approach to the proposed asset classes should be clearly documented and agreed with the competent authority. The findings showed that no monitoring of the PPU condition were identified in some institutions and no clear criteria could be defined in the decision of the application of the IRB approach. The level of detail in the decision-making responsibilities and internal reporting were not appropriate. Also, management body didn’t approve all risk management policies.
In some cases, non-rated exposures and outdated ratings were not monitored by the institutions. Additional potential misalignments were identified in the scope and frequency of the audit review of the rating systems. These misalignments were due to a lack of resources to allow a relevant assessment of the IRB requirements and there wasn’t any review of the rating systems. Also, change policy and re-rating process were not formalized.
The second part covered the credit risk investigation and as mentioned before, the focus was on the determination of PD and LGD. The determination of margins of conservatism and the review of estimates was subject to a significant number of findings. These included the calculation of default rates and the definition of the period representative of the long-run average. Also, some findings were related to a lack of consideration of relevant risk drivers or a lack of an appropriate definition of the grades
Most of the findings related to the calculation of realized LGD. The investigation mentioned the use of an inappropriate discount rate and the treatments of multiple defaults. Also, specific aspects of the calculation were identified during the intensive walk-throughs performed by the assessment teams during the on-site investigations. In addition, the estimation of long-run average LGD was subject to a significant number of findings, such as the treatment of incomplete work-outs, the downturn adjustment and the treatment of defaulted assets.
What will be the impact of TRIM?
Some findings are clearly recurrent in every institution that has been investigated. For example, issues with the calculation of the realized LGD have been detected in 100% of the institutions investigated and issues in the estimation of the long-run average LGD have been detected in 96% of the institutions investigated.
These issues arise from the misinterpretation of the internal model evaluation methodology or are due to deficiencies in the implementation of the internal models. Follow-up letters have been sent to the institutions that have been investigated in order to update their internal model approaches. Following the investigations, these institutions should be able to:
- demonstrate that, on the basis of the validation results and recommendations, measures are initiated to remedy the identified deficiencies of the rating systems.
- ensure that senior management and the management body are informed about the conclusions and recommendations of the validation results.
- have adequate processes in place for tracking the status of the measures adopted to remedy deficiencies.
- apply to the competent authority in the event of changes to their validation methodology and processes.
The guide published by the ECB must be well understood and areas where measurement uncertainty and model deficiencies are identified must be corrected. Banks are expected to develop and establish group wide policies and guidelines related to the maintenance of internal models and also ensure adequate audit mechanisms are in place. They are also expected to establish a robust model management framework and may have to make changes to their current model methodologies, model governance policies and data quality practices.
Under the revision of the IRB approach, the output of banks own calculations under the internal models is required to have a minimum of 72.5 percent of the standardized version. The way banks measure similar RWA is subject to large inconsistencies with some lenders relying on overly “optimistic” assumptions. The ECB will also examine assets on the balance sheet that cannot be determined by market price and that rely on banks’ internal assumption such as credit default swaps, long-dated derivatives and private equity investments.
The revision of the internal models may affect the performance of banks and see a fall in their CET1 ratio. For example, banks with a high degree of exposure to residential mortgage loans where the use of internal models and low historic default rates resulted in significantly underestimated risk weights may be heavily affected. These changes will also imply re-examining certain areas in their business. The portfolio composition could be revised, as some products may be less attractive based on the associated cost of capital. The controls put into place will require new control procedures and the revised calculation will need to be supported by new systems.
TRIM is expected to have consequences beyond 2019. Initio can support banks in building stronger control framework, reviewing and updating their internal model approaches based on the TRIM findings.
The execution of the TRIM project is not yet finished, the ECB will run its on-site investigation on low-default portfolios models through 2019. When all on-site investigations will be finalized, further updates on TRIM will be shared with the banking industry. Banks must be well-prepared and ready to adapt to new regulatory requirements. The BCE has invested a lot in this review of internal models and it represents a real challenge for banks. The TRIM project and all the initiatives around the review of the internal model’s framework raises the question of whether this type of model will even survive through time as there is a growing regulators’ desire to see internal models converge towards standard methodologies.
About the Author
Lorenzo Friz, completed a Master in Business Management at ICHEC and at the Louvain School of Management as part of the Track Financial Management Program. Before joining Initio in 2019, Lorenzo spent 1 year as an auditor at Mazars where he implemented audit works in different sectors.
- K. Sumpter E. Katz D. Carolan (2014) Internal Ratings Based Approach to Credit Risk in the Banking Book Allen & Overy Client Briefing Paper 4
- European Central Bank (2018) TRIM: reviewing internal models
- European Central Bank (2018) Internal models
- European Central Bank (2017) Guide for the Targeted Review of Internal Models (TRIM)
- European Central Bank (2017) What is the targeted review of internal models?
- European Central Bank (2017) Overview of outcome of general topics review and credit risk on-site investigations
- G. Conn (2014) The Benefits and Challenges of Adopting Internal Models
- KPMG (2018) Basel 4: The way ahead