On 30 September, the PRA published its latest Written Auditor Reporting (WAR) feedback. The report builds on last year, with a focus on “the importance of recognising changes in credit risk in a timely way in a dynamic and challenging environment and enhancing capabilities to quantify the impact of climate-related risks”. The feedback concentrated on three areas: model risk, recovery strategies and climate risks.
Under WAR, the PRA asks the auditors of the major UK banks and building societies (greater than £50bn balance sheet) to submit an annual report about particular aspects of their audits. The PRA summarises the findings from the reports and their other work in an annual feedback letter addressed to CFOs, published around the end of September each year. The latest feedback letter reflects auditors’ reports received in 2025 (i.e. covering the 2024 year-end financial statements, except Nationwide which now has a 31 March year-end).
The PRA’s aim is that firms adopt high quality accounting practices for ECL, including the timely inclusion of climate risk in accounting valuations, to help ensure the safety and soundness of firms.
While the feedback is particular to the major UK banks and building societies, the PRA thinks it will be helpful to all firms applying IFRS 9 ECL.
Overall, the PRA reported that
Firms have made continuous improvements in their capabilities and governance frameworks, but exhorted firms to challenge and to continue enhancing their processes. The PRA did not, however, repeat two overall findings from last year: “we continue to see variation in practice and scope to further embed high quality practices”, although it is unclear if this suggests there has been an overall reduction in variation between firms, as the PRA identified variation in approaches for a number of granular matters.
Areas of concern and future focus
Model risk
The PRA outlined five areas of interest that will be their future focus:
- The robustness of processes to ensure the completeness of Post Model Adjustments (PMAs), including the ability to respond or identify new and evolving risks that might not be captured in existing models. The current economic and geopolitical uncertainties were highlighted as potential sources of risk, and where firms were at an early stage in assessing the risk – the PRA may expect these assessments to be considerably more developed for 2025 year-ends.
- The delivery of strategic model redevelopment to better capture risks and reduce reliance on PMAs.
- The monitoring of clearly defined model operating boundaries (ie the economic circumstance under which the model is expected to perform reliably) for new and recalibrated models, based on the range of economic variables used in their development, to help identify performance issues.
- The enhancement of model monitoring and validation, though the PRA noted it will take several years for firms to align with the expectations in SS1/23 (model risk management principles).
Specifically, the PRA will focus on embedding more granular monitoring and embedding changes to align with SS1/23.
Generally, firms were more advanced in developing their retail than corporate models.
Recovery strategies
The PRA outlined two areas of interest that will be their future focus:
- That early-warning indicators and governance processes are in place to support timely identification of risks of recovery strategies failing for potentially vulnerable sectors or borrowers.
- The continuing development of LGD models, including more granular models, using more diverse recovery data sources, and considering how models can be more responsive to the range of recovery paths.
Climate risk
The PRA is focusing on four ‘capabilities’:
- identifying the risk drivers most likely to impact ECL;
- quantitative analysis to assess how specific climate risk drivers affect ECL;
- adapting economic scenarios and weightings used for ECL calculations to incorporate the [climate] risk drivers; and
- the collection and management of data and the development of models to factor the risk drivers into loan level ECL estimates.
Unsurprisingly this area is still work-in-progress. Generally, the PRA found firms to be at varying stages of development with more work required across all capabilities, including to enhance the sophistication, granularity and breadth of analysis (eg, consider a broader range of economic scenarios). It was noted that less progress had been made with ‘data and models’ and that the availability of data remained a pervasive issue.
WAR next steps
For the next round of WAR reporting (2025 year-ends, with feedback around September 2026), the PRA will focus on the progress made on the issues identified above, along with processes around data quality and the effects of securitisation transactions.
Three thoughts …
One: it is likely that the methodologies used to estimate ECL can never stand still. In this regard it is appropriate that much of the PRA’s ongoing focus is on ensuring effective ECL processes: for example, to challenge whether a model remains relevant and reliable; that can respond and adapt to change; and that can identify new risks and their drivers.
Apart from death and taxes, the other certainty is that the environment is complex, dynamic and ever changing (potentially significantly so). In the last fifteen or so years, there have been significant economic and geopolitical events (eg, a global financial crisis, a global pandemic, a cost of living crisis), individual country events (eg, Brexit), the rapid growth in new risks and opportunities (eg, digital currencies and AI, and climate risks), as well as a myriad of other lesser, but still impactful events (eg, the late 2022 gilt market and LDI crisis). Moreover, as the environment and risks change so will the relationships between the various risk drivers and economic variables.
Firms will also evolve their business models to respond to changing risks and to take advantage of new opportunities: new products may be created, and the size and composition of lending portfolios will be adjusted.
While there can be periods of apparent benign economic conditions, history tells us these periods may store hidden risks. Firms, therefore, need to be alive to the possibility for change, that change may be significant and play out in unexpected ways, and be capable of managing change when it happens.
Two: the PRA indicated it wants reduced reliance on material PMAs. While the estimate of PMAs is potentially more subjective and lacks the rigour of a well-developed model, there are several reasons why it may not be possible to eliminate them completely and it may also be difficult to significantly reduce them (except in really benign periods).
As noted above, the environment can be dynamic and volatile. So, there is always a possibility that current models will not capture new risks (including those associated with new products) or deal well with changing risks. It then takes time to work through the implications of change and accumulate a sufficient time series of data to enable a new model to be built and calibrated or to modify an existing model. Finally, firms must balance the benefits of adapting or building new models with the (potentially not insubstantial) cost of doing so. For example, as a loan portfolio that represents a very small proportion of the overall balance sheet will have very impact on the overall ECL charge, an expensive sophisticated model represents a disproportionate cost. It becomes important to have a strategic approach to model development.
A further consideration for firms when developing IFRS 9 ECL models is the interaction with regulatory IRB models. While there are some fundamental differences between IFRS and IRB they are related. Firms may also, therefore, need to factor into their strategic planning for ECL models, their approach to developing regulatory IRB models, including the timeframes for regulatory approval of IRB models. And regulatory expectations for ECL should reflect the regulatory approach to IRB.
Three: the PRA noted a number of variations in practice between firms. It is reasonable that variations will exist, as differences in the size and composition of the different bank portfolios will mean banks are not necessarily exposed to the same risks, or that the potential magnitude of the effects differs. Banks will therefore have different priorities for which models and characteristics to develop, determined by different lending types and risk drivers.
