An Overview of the EBA's Initial Margin Model Validation

The European Market Infrastructure Regulation (EMIR) 3 Refit comprises many moving parts. One notable element is a process revision for the validation of the models used to calculate initial margins. One year after the European Banking Authority (EBA) released the final draft of Regulatory Technical Standards (RTS) addressing how to validate initial margin models outsourced to external providers, additional provisions of the regulation are now taking effect.

The EBA RTS final draft contained a series of processes for validating models. These processes included both qualitative aspects pertaining to governance and quantitative aspects in the form of a backtesting process designed to assess the model performance. Additional elements included:

• Dual processes: Due to the diverse range of market participants and the simultaneous requirements for regulatory compliance, the final draft set out a twin-track approach. The most significant market participants use the standard process while smaller counterparties use the simplified process.
• AANA threshold: The threshold that determines which process a counterparty should use is based on Aggregate Average Notional Amount (AANA), with a current threshold of EUR 750 billion.
• Assessment method: The validation process, whether in its simplified or standard model, begins with the submission of a validation request. It then progresses to an assessment of aspects in terms of governance and backtesting methodology.

Currently, most market participants use the Standard Initial Margin Model (SIMM) harmonization used by the International Swaps and Derivatives Association (ISDA), offering the possibility for competent authorities “to avoid assessing the same core methodology more than once."

“The European Banking Authority (EBA) published new rules for validating initial margin (IM) models used by banks to calculate collateral requirements on over-the-counter derivatives. These rules aim to enhance the accuracy and reliability of IM calculations across Europe.”

Before looking at the differences between the standard and simplified procedures, let’s consider the different sorts of backtesting in play. A key element of the initial margin model validation requirements is the administration of periodic backtesting, which must be carried out by the relevant counterparties. There are two types here: static and dynamic.

Static Backtesting

This involves comparing the initial margin against changes in market value, which are calculated by applying the returns of the risk factors used in OTC derivatives pricing. The overshooting of the Initial Margin, as well as counted and netting sets, are classified using specific Basel traffic light categories based on a normal distribution methodology. This methodology assesses whether the initial margin adequately covers theoretical losses.

Dynamic Backtesting

This involves comparing the initial margin calculated for a 1-business day Margin Period of Risk (MPoR) against the daily market data changes between each valuation date within the period and its subsequent business day. The backtesting period refers to the most recent 250 days or the most recent dates available.

The overshooting is counted for the static backtesting, and a similar classification method must be used. The same pricing methods as those deployed for official validation must be applied in the computation of market data changes; however, a close approximation, such as a Taylor expansion, can also be used.

As a new addition, margin average shortfall measures and margin average relative shortfalls will be calculated for both static and dynamic backtesting. Since backtesting requirements are likely to become more widespread as the Basel III update comes into effect, EU financial institutions that understand what the new RTS requires of them may be in a better position to comply with impending regulations.

“Static backtesting compares the calculated IM to historical market data to ensure it adequately covers potential losses. Dynamic backtesting assesses the IM's performance over a short time horizon. These backtesting results are crucial for ensuring the accuracy and effectiveness of IM calculations and contribute to a more stable and resilient financial system.”

As mentioned, the regulations established two sets of procedures: a standardized approach for larger institutions and a simplified approach for smaller ones.

Standardized Procedure

The standardized procedure is applicable to larger counterparties identified through the AANA calculation, which are more complex and capable of meeting the more demanding requirements. Other counterparties will be subject to the lighter-touch framework known as the simplified procedure.

According to the standardized procedure, the IM must undergo static backtesting to assess its performance, as well as continuous monitoring to identify any shortcomings. Dynamic backtesting is required to complement the static backtesting process, and it must be run continuously. The results of the dynamic backtesting need to be published at the end of every quarter.

Simplified Procedure

For counterparties in the scope of the simplified supervisory procedures, only the dynamic 1-day backtesting will be needed. For the qualitative governance aspect, the framework outlines a comprehensive list of documents and procedures required, with a reduced burden for the simplified procedures.

“According to the RTS, the validation of internal margin models is “an important step in improving the accuracy, relevance and effectiveness of initial margin calculations across the EU.” The hope is that it will do so by creating a “level-playing field” where initial margin calculations are concerned.”

Under Article 31 of the RTS, the general provisions (Section 1) of the regulation came into effect in July 2023, while other provisions roll out over time.

Section 2, which outlines the standard procedures, is only now coming into force, one year after the RTS publication. The effective date of Section 3, which outlines the simplified procedures, is based on the AANA, with a threshold of EUR 50 billion.

• For counterparties with an AANA > EUR 50 billion, Section 3 comes into force in July 2025, two years after the RTS publication.

For counterparties with an AANA ≤ EUR 50 billion, Section 3 takes effect in July 2026, three years after the RTS.

With the EBA’s new IM model validation framework now in effect, financial institutions must ensure their systems can handle its rigorous backtesting and governance requirements. The increased frequency of static and dynamic backtesting, coupled with tiered validation procedures, demands robust compliance strategies and scalable technology solutions.

To stay ahead, institutions should prioritize RegTech solutions that integrate regulatory logic, automate calculations and support evolving methodologies. Given that backtesting is now a central compliance function, its optimal implementation within a consolidated middle-office framework will be key to meeting both current and future regulatory demands.