The targeted review of internal models – the good, the bad and the future
The ECB’s targeted review of internal models (TRIM) is a multi-year project to ensure that capital requirements for banks using internal models are calculated correctly, consistently and in a comparable manner. TRIM will therefore contribute to the soundness of the banking system, create the essential conditions for a risk-sensitive approach to supervision and help create a level playing field across banks under ECB supervision.
Last but not least, the structured investigation framework developed for TRIM and the in-depth knowledge gained during the exercise will also establish high standards for future internal model supervision and help ensure the application of consistent supervisory practices across constituencies.
In this way, TRIM contributes to reducing unwarranted (i.e. non-risk-based) variability when calculating the minimum amount of capital that banks must hold by law and helps restore the credibility of internal models as a tool for banks to assess and manage their risks.
By the end of the project, ECB supervisors will have conducted about 200 on-site investigations at 65 banks that it supervises directly and will have assessed whether the internal models banks use to calculate their capital comply with regulatory requirements. These inspections cover the most material and critical models for credit risk and all internal models for market and counterparty credit risk.
So far, about three-quarters of the missions have been completed. The findings relating to the general conditions for the use of internal models showed that some banks lacked a policy for model changes, did not back-test their internal models annually, used the standardised approach to calculate own funds requirements without formal approval by the supervisor, and/or failed to ensure that staff validating the models were not involved in developing those same models. In short, banks need to do more to improve the way they implement and use internal models, and address the deficiencies detected.
Regarding credit risk, TRIM investigations for retail and corporate small and medium-sized enterprise portfolios produced, on average, 13 findings relating to the loss-given-default parameter and seven relating to the probability-of-default (PD) parameter. In particular, banks showed deficiencies in calculating realised losses on credit risk exposures and in the use of long-run average default rates for the calibration of the PD. A more complete list of interim findings can be found in the second update the ECB sent to banks in April.
The scope of the credit risk investigations also included a review of data quality. The findings of this review include several shortcomings in banks’ policies regarding data quality management and processes; in the allocation of roles, responsibilities and ownership in data management; and in the processes for reporting and resolving incidents.
Regarding market risk, the issues found in the TRIM investigations mainly concerned the methodology banks use to measure value at risk (VaR) and stressed value at risk (sVaR), regulatory back-testing of VaR, the scope of the internal model approach, and the incremental risk charge methodology. For instance, typical shortcomings detected relate to deficiencies in the data quality for VaR and sVaR models and the computation of the actual and hypothetical profit and loss for regulatory back-testing.
The exercise has already led to around 80 supervisory decisions, each including about 20 legally binding obligations for banks alongside other supervisory measures to remediate the identified shortcomings in the internal models.
Banks must remediate cases of non-compliance identified in the course of the investigations. Where appropriate, the ECB recommends that banks apply best practices and raise their awareness of ongoing regulatory developments. The resulting measures and their potential impact on capital vary widely between banks. For example, some banks were well placed to meet the requirements as they had already worked towards improving their models in preparation for the upcoming TRIM investigations. And while for a number of banks TRIM has had a relatively limited capital impact, others have seen TRIM and its follow-up result in a material capital impact. TRIM is not intended to increase capital requirements per se, but it does aim to ensure that regulatory requirements are implemented consistently.
Looking ahead, the next set of on-site investigations should be completed in the second half of 2019. Expected to be formally concluded in the first half of 2020, TRIM will require banks to continue working on addressing the deficiencies identified so far and should help shape the agenda for the future development of internal models and related supervisory activities.