The EBA’s 2018 stress test will not just be a re-tread of previous exercises. Nor will banks simply be able to rely on their internal stress testing models and processes. The arrival of IFRS 9, together with some specific changes to the EBA’s approach, will make this a challenging exercise. Given the impact the stress test results will have on the 2018 SREP, banks should start their preparations now.

The only thing that is constant is change. Today’s concern may be Brexit and non-performing loans, but what about tomorrow - or the day after that? Given the pervasive uncertainties facing both banks and regulators, it is hardly surprising that stress testing has gained prominence as a supervisory tool.

For most European banks this is not a new story. Regulatory stress testing has become part of the fabric of risk management for banks within the SSM, albeit less frequent than their own internal stress tests. The 2018 EBA stress test exercise, due to commence early next year, will follow on from similar biennial exercises in 2014 and 2016.

If this sounds like a low-effort exercise in running existing stress testing models and processes, think again. The draft methodology and templates published by the EBA in June¹ imply a number of challenges for banks’ stress testing processes, in particular different calculation approaches and more onerous data requirements.

The draft methodology reveals a number of changes from previous exercises. Arguably, the most demanding will be the introduction of IFRS 9 expected credit losses to the forecasts. In general, banks are expected to apply their own accounting approach for IFRS 9 projections. However, some aspects of the EBA methodology are likely to deviate from many banks’ internal approaches. Two examples are the lack of cures for Stage 3 accounts, and the requirement for a blanket 200% relative increase in the probability of default to be treated as a significant increase in credit risk, resulting in stage transfer.

Such requirements for credit and other risks reflect a seemingly intractable problem for supervisors. If banks are given free rein to use their own internal methods, the exercise risks becoming a test of who can “game” their models most effectively. However, imposing a detailed common methodology would create tremendous operational overheads for banks and could divorce the results from bank’s own views of financial risk.

Despite the “draft” nature of the EBA’s methodology and templates, it is still valuable for banks to begin preparing for the 2018 stress test, and especially to data integrity and –availability, dry-run processes and look for efficiency gains in areas that have traditionally proved difficult. We would recommend that banks pay particular attention to the following areas:

• Data quality. Stress testing and IFRS 9 are data hungry processes, requiring collation and processing of data from a multitude of systems and time periods. In addition, EBA is requiring more input data. Having tools and control processes that can validate and quality assess submission data is a powerful response to the increasing scrutiny on data quality by management and regulators.
• Resources. Given the timing of the exercise relative to year-end reporting, banks are very likely to feel a squeeze on their risk and finance resources. Resource planning, front-loading whatever can be front- loaded and clear lines of responsibility will be crucial for success.
• IFRS 9 methodology. Given the potential differences in IFRS 9 methodologies, assumptions and accounting decisions between banks, the EBA will request quantitative and qualitative information on IFRS 9 implementation as part of the stress test. Developing relevant documentation in advance of the exercise can aid internal understanding and lead to useful feedback from management.
• Model risk management. The scope of stress testing models and newness of IFRS 9 models represents a challenge for model governance. Previous KPMG benchmarking² indicated that many stress testing models are left out of the scope of model validation processes. This could lead to key calculations being dependant on models that are not fit for purpose.
• Benchmarking / challenger models. Several banks have expressed a desire to be “in the pack” when it comes to IFRS 9 projections and stress test results. Having top-down challenger models and simulation tools can help banks to sense check their results and key sensitivities.
  

In summary, the 2018 EBA stress test will give banks a lot to consider. The outputs of the exercise will be used to inform the 2018 Supervisory Review and Evaluation Process (SREP), so a high quality, on-time submission is essential. Effective preparation now will help to identify potholes in the road ahead and the resulting areas for action - mirroring the nature of the stress test itself.

¹EBA issues 2018 EU-wide stress test methodology for discussion

²Stress testing (PDF 833 KB)