TRIM is ready: The Risk-Type Specific part of the ECB Guide to Internal Models (TRIM) July 2019

By Thomas Obitz, FRM

Yesterday (July 8 2019), the ECB has published the final version of the risk type specific chapters of TRIM. Apart from duly referencing the latest versions of CRR and various RTSes, changes have been applied across all risk types.

The changes related to market risk partially respond to the FRTB2019 regulation which to some extent they pre-empt (e.g. trading book/ banking book rules). In some areas they even go further, e.g. with regards to TB/BB rules for funds. The RNIME framework is further elaborated. Data now has to be “objective” instead of “observable” – which does not help if the word “objective” is not defined anywhere.

The most notable change on the credit risk side is probably a framework on the use of human judgement and of overrides.

Counterparty Credit Risk rules will be covered in a subsequent version of this article.

Below an excerpt of the changes. Do let me know your thoughts about the new version of TRIM.

Market Risk

• FRTB-style Trading/ Banking book rules appear influenced by FRTB 2019 – specifically, equity funds require daily prices or daily look-through. (Daily look-through is significantly more restrictive than the “sufficient and frequent” information of FRTB2019) (6), (7)
• Visibility by instrument category where IMA is applied (across trading and banking book – which des not appear completely compelling) (8)
• Excluding positions from IMA requires to demonstrate that the SA adequately capitalised the risks of a position (23)
• “Objective” instead of “observable”, e.g. for correlations (46) and CIU information (40(b)).
• Valuation adjustments and reserves to be aligned between economic and
• Clarifications on the requirements for Hypo-P&L calculation (footnotes 100 – 103 – again, some resemblances with FRTB)
• You cannot withdraw a P&L outlier notification because you have changed the calculation method afterwards (82(f)) – sic!
• Banks should identify sub-portfolios which would migrate to IMA trading desks under FRTB (90(b))
• Clarification of the uniformity test on p-values (92(b))
• IRC: Permission to exclude defaulted issuers (161(b))
• Elaboration and clarification of RNIME framework (“The estimation of ???????? should be as accurate as possible using reasonable effort. Therefore, the ECB understands that the ???????? estimation methodology can use appropriate approximations, assumptions, or a stress methodology when duly justified and documented.”, 178 and 179). Notification requirements on RNIME methodology changes (186)

Credit Risk

• Retention obligation for rating system documentation of at least 3 years
• If multiple obligor databases are used, they need to be reconciled or understand the impact of not doing so (34)
• Proportionality requirements on the analysis of external ratings (38(b))
• Validation requirements applicable both to internal and external input variables (38(c))
• Tightening of requirements on use of data pools
• A significant extension of the framework to formalise the use of human judgement and overrides (45 – 48)
• Explicit framework on use of internal vs external PDs (88)
• Simplification on use of risk parameters (108(b))
• Calibration of LGD (117)
• Removed detailed guidance how to estimate a downturn LGD and referred to further EBA guidelines for downturn estimation

Thomas Obitz is founder and director of RiskTransform, a London-based consultancy that works with banks on regulatory change. Any opinions, findings and conclusions are those of the author and do not necessarily reflect the views of RiskTransform or any of its clients. They should not be considered actionable advice.