On the 15th April the PRA wrote to the CFO’s of the top seven banks to give their initial opinions on the implementation of the new impairment requirements under IFRS 9. As expected for such a substantial subject, the findings, based on their written auditor reporting work, are varied in nature. Through discussion with a range of lenders and client engagements, we see many of the issues raised by the PRA. However, there are also some omissions that we expected to see.
The thematic findings from the PRA relate to implementation concerns and these are tied to comments made in previous PRA statements. The key points enforce the message to implement ECL well, aim for outcomes that are consistent, reported comprehensively – again urging consistency – and a firm’s expectation that practices will continue to evolve beyond the current state. In respect of ECL implementation and the evolution of methods, the points raised comprised the need to strengthen controls and management information over ECL models, to reduce reliance on governance as the means for identifying and addressing weaknesses and adjusting provisions, to further consider downside scenarios, particular in number, incorporate judgement more thoughtfully via models and further consider the robustness and sensitivity of SICR choices.
Interestingly, there is little direct regards to the overall method chosen for calculating ECL. In our experience there are broad categories for estimation that are quite disparate in nature. This is particularly true for lifetime PD where approaches can range from simple linear extrapolations to migration matrices to fully independently developed decomposition survival models. The main modelling concern highlighted is with regard to discrimination of up-to-date accounts and consideration of predictive variables. This is generally not something we have observed with our clients, who generally have developed predictive models for this population, whether it be scorecards or decision tree type approaches.
The identification of required improvements in Management Information should be seen as sitting at the centre of a number of issues as, in our view, effectively solving this should support the remediation of the other identified areas. For example, strong reporting regarding the incremental impact of economic scenarios and sensitivity analysis, should allow firms to identify the appropriateness of their selections and identify scenarios, where they may be underprovided. Similarly, an understanding of the continual performance of SICR triggers against determined rules (a subset is highlighted in the letter) will ensure that these triggers continue to be appropriate. Finally, ensuring that Management Information includes the performance of the models against observed outcomes should reduce the reliance on Post Model Adjustments as, if required, remediation can be applied in-model reducing uncertainty of the model driven allowance.
The development of this enhanced Management Information can then facilitate more robust governance and controls through appropriate escalation triggers where weaknesses are identified in either model input (data), model application (sensitivity and component testing) and model output (overall ECL performance monitoring and allowance coverage). A broader observation is that the we see some firms struggling with finding a stable operating rhythm – for the monthly reporting cycle, for more substantive quarterly updates – say for economic changes, for demonstrating adequacy of provision, for making changes to models and parameters. This requires similar robust controls and governance to what is expected from an IRB framework. There are many parallels and certainly the need for the Board to be equally engaged in their duty of oversight.
In addition, to the above there is clearly also a view regarding to the consistency of decisions across the industry with the two key areas, as expected, being the application of SICR and economic scenarios. With regards to these points it is very difficult for firms to remediate without 3rd party support, whether that be via auditors or industry experts. The cumulation of the PRA letters signals they expect management to not let up but to remain acutely focused on improvement, evolution, understanding and managing between accounting and regulator measures and systems.
We are already engaged with clients on generation 2.0 versions of models and frameworks and helping to prioritise what their next evolution should be. The thrust of this work covers:
- Revisions to controls and governance over IFRS 9 and where relevant integrating this with other similar structures through to the Board, including the use of expert judgment;
- Improvement to forecasting and stress testing of IFRS 9 that aligns with similar needs for capital;
- Developing bespoke monitoring and reporting that demonstrates both the adequacy of the provision and explains the period on period movements; and,
- Revising the organisation design and operating model to create a workable cycle for production, reporting and incorporating change, including considerations of scenarios and impacts.
Some expectations – possibly further out – will be consolidation of approaches for Significant Increase in Credit Risk that deliver suitable sensitivity of risks through and ahead of time, standardisation of core disclosure tables and specification of economic scenarios. In the meantime, expect continued interest in provision reporting particularly when conditions start to change!
If you would like to discuss the above or other regulatory issues, please contact Chris Warhurst, Technical Director, firstname.lastname@example.org