Breaking down the PRA’s latest Dear CFO letter: Thematic feedback on accounting for IFRS 9 expected credit losses (ECL)

On 30^th September the PRA published a Dear CFO letter containing the latest thematic feedback on accounting for IFRS 9 expected credit losses (ECL).

Many of the themes are continued from previous letters, such as model risk management, recovery assumptions for LGD models, and the incorporation of climate risk. However, there is a new and increased focus on data quality and the measurement of ECL risks for securitisations.

We summarise the key points and outline strategies to help enhance alignment with the PRA’s expectations. We also share themes we have observed across the industry that warrant further consideration by lenders to strengthen their IFRS 9 solutions.

Model risk management

The PRA’s expectations for managing model risk continue to grow as SS1/23 becomes more firmly embedded. The regulator remains focused on ensuring model risk frameworks are robust enough to address new and emerging risks. A key area for any bank is its Post Model Adjustments (PMA) process, where the PRA expects greater completeness and rigour.

• It is important that CFOs and CROs regularly challenge the PMAs that are in place to ensure that they are comprehensive (capturing all new and emerging risks).
• Rigorous policies and procedures that ensure strong linkages between any emerging risk analysis and the IFRS9 ECL process is key to success. We still see many portfolios where this link from emerging risks to the ECLs is not a mature process, undermining the ability to describe to Board, auditors, and other stakeholders a coherent ECL narrative.

The PRA continues to encourage banks to enhance or replace models with longstanding limitations, ensuring investment is directed towards better capturing risk and meeting supervisory expectations. In particular, the PRA expects new models to be more robust across different economic conditions and, crucially, to clearly identify any limitations at the time of internal model approval. These areas of focus aim to support the development of a modelling framework that is stronger from a model risk management perspective.

• Robust model development plans, which ensure the most material limitations (including highest PMAs), are regularly tracked to ensure continued improvements.
• Regular model monitoring is key to ensuring investment is targeted correctly, with development plans being updated where new limitations are identified. We are seeing firms enhance monitoring, analysing all components of the ECL calculation – including PMAs – with clear actions being identified to mitigate newly identified weaknesses. For example, we observe more granular monitoring of the performance of borrowers with higher repayment-to-income ratios, capturing higher vulnerability to increases in repayments.
• Clear definition of model operating boundaries to help identify economic or other scenarios where the existing models would not be expected to be fit for purpose. Specific actions would then be taken to ensure appropriate ECLs (PMAs or model enhancements).

Recovery strategies

By highlighting it as an ongoing issue, the PRA continues to be concerned about the robustness of recovery assumptions driving LGD models, especially for vulnerable sectors or borrowers.

• The PRA’s view is that, in general, firms have not developed sufficient internal reporting to understand how recovery strategies can change during a default event, along with how the outcomes are impacted.
• There is an expectation that early warning systems and supporting processes should be in place to identify when the recovery strategy outcomes are diverging from the expectations embedded into the LGD models.
• As LGD outcomes are more difficult to monitor than PD outcomes, we are seeing banks increasingly tracking interim milestones throughout the strategy. This can then be used to challenge assumptions in the LGD models to ensure more “real time” feedback to identify if there is a material discrepancy. This may lead to more granular LGD models as recovery strategies mature.

Data governance

A foundational element to a strong ECL process is high quality data processes and structures. Given the increasing levels of data feeding the ECL process (both models and PMAs), the PRA is committed to explore the quality of governance and processes around source data, data flows, and data aggregation.

• The PRA will focus on accountability (ownership) of data risk, robustness of data quality policies and procedures, appropriateness of the complexity of aggregation and handling, knowledge and oversight of third-party solutions.
• The key steps to introduce control across an IFRS 9 journey are:
  • Identify Critical Data Element (CDEs) in models and the wider calculation:
  • Trace and document lineage of CDEs back to source, varying depth of lineage based on complexity and criticality of data elements:
  • Define Data Quality rules and tolerances for CDEs based on the impact on the model and calculation outputs.
• Many organisations find a holistic Data Risk maturity assessment a good first step to highlight focus areas of improvement in their approach to data governance and managing Data Risk

Climate risk

The 2025 letter places a much greater emphasis on climate risk. While the PRA acknowledges progress in incorporating climate risks into ECL processes, it continues to encourage further efforts to improve the identification, assessment, and modelling of climate risk drivers.

Lenders that assess these risks typically do so outside the core IFRS 9 models, relying on expert judgement along with outputs from stress tests and climate scenario analysis as the main drivers.

As expected for retail exposures risk drivers focus on physical risk impacts, which we know tend to be much further into the future, and transitional risk – with focus on low EPC ratings and the impacts of remediation activity required on house process and customer affordability.  For corporations, the driver remains carbon tax and impact on a company’s bottom line.

To better understand these risks, the PRA highlighted the need for more granular insights, such as property-level considerations for retail portfolios. We know that tail-end effects ultimately drive potential climate-related losses, so a bottom-up view is essential to capture these accurately. For corporate portfolios, the PRA stressed the importance of analysing sub-sectors to understand portfolio dependencies — think GVA impacts rather than GDP.

Other areas identified for improvement included:

• enhanced considerations to SICR – moving away from simplistic broad-brush approaches.
• The development of climate transition scorecards and their alignment with transition plans.
• Review of the flat hair cut approaches to collateral valuations.
• Greater consideration to physical risk collateral data within corporate portfolios.
• Scenario development and expansion to produce relevant climate related variables.

Additional themes

Our view is that there are additional themes that pose some risks for an effective IFRS 9 implementation. We believe that the following areas should remain in focus for all firms:

• Forward look – we see limited evolution of thinking around scenarios, in particular geopolitical risks or risks associated with cyber-attacks and IT risks (e.g. external vendor issues and disruption of IT infrastructures) are not typically considered.
  • Best practice here includes strengthening the emerging risk process (not just repeating the previous year’s risks) and then actively identifying the risks that could have the largest impact on the balance sheet.
• The use of Covid data and/or high-inflation/high-interest rate data in model recalibration remains a significant challenge across lenders.
  • Currently we are still seeing lenders using this data as a “testing” scenario for the models without directly embedding into the model due to unusual interactions observed during those periods. We expect that where scenarios include similar outcomes, model limitations may need to be supplemented by PMAs that are strongly linked to the outcomes observed during these periods.
• Staging remains one of the most judgmental and challenging areas for the implementation of the IFRS 9 standards.
  • We continue to see differences in interpretations and some inconsistencies across the industry, which affect comparability between lenders.

Get in touch

Send us an email if you would like to learn more about how our team of experienced consultants can support your organisation navigate the evolving regulatory landscape and optimise your IFRS 9 strategy – info@4-most.co.uk.