Banks calculate expected credit losses (‘ECLs’) under IFRS 9 using forward-looking judgements, models and data. Overlays, or post-model adjustments, are often used to address shortcomings where models or data have limitations. As a result of severe economic conditions and uncertainty arising due to coronavirus (COVID-19), there is an increased need to apply overlays in calculating ECLs. In this publication, we provide considerations that might be helpful in developing and monitoring such overlays.
Banks estimating ECLs under IFRS 9 often use a three-step process: 1) develop judgements about the future; 2) apply those judgements to (statistical) models developed based on historical relationships; and 3) use relevant data to feed into the models. This often involves more statistical modelling and data than most other accounting estimates, and it might be very difficult in the current environment. Extreme economic conditions – coupled with uncertainty around the duration of the pandemic, potential for relapses, effects of government support and what recovery will ultimately look like – mean that forward-looking judgements are highly uncertain and challenging to make. At the same time, historical relationships between key variables might no longer hold, and comparable economic conditions might not have existed in the past. Lockdown and social distancing effects and timeframes will need to be expressed in terms of impact on macroeconomic drivers and, ultimately, on default rates. It will not likely be possible to revise models in the short term to capture all of these factors and uncertainties. Banks often use overlays, or post-model adjustments, where risks and uncertainties cannot be adequately reflected in existing models. We expect that such overlays will necessarily play an even more important role and will be higher-level in today’s environment.1
What does ‘overlay’ mean?
‘Overlay’ is a term that can be used to describe a spectrum of adjustments that are made outside the primary models. In some cases, the term can refer to straightforward adjustments in order to correct known model errors or data deficiencies. In others, the overlay is far more subjective and judgemental. For example, it sometimes refers to the application of expert credit judgement to address gaps in models, data or both (for example, as new risks or uncertainties arise). It can also refer to adjustments made to capture risks and uncertainties which are not captured by the models because the models were not designed to address them (such as Brexit). While potentially applicable to all overlays, the considerations in this publication will be of most relevance to overlays towards the more judgemental end of the spectrum.
What questions should banks ask in establishing overlays?
Since they are inherently judgemental, overlays require robust process, governance and internal controls, supported by transparent and high-quality documentation. Key questions to consider include:
- What is the limitation that is being addressed, and why?
- How was the overlay quantified, and what rationale was used?
- What are the underlying assumptions, and how were they developed and supported?
- What data was used, and how was it determined to be appropriate and consistent with similar data used for other purposes?
- How will the overlay be consumed over time (for example, through model development / redevelopment, new data becoming available, or loan-level losses having transpired)?
- How will reasonableness / performance be assessed (for example, by using back-testing, KPI monitoring, comparison to stress-testing and stand-back tests)?
- What has been done to determine, at a sufficiently granular level, the exposures to which the overlay relates?
- How have the staging implications of the overlay been addressed?
- Has an end-to-end review of the ECL modelling process been completed, to ensure that all potential model limitations which might indicate the need for an overlay have been considered? For instance:
- Is recent borrower information available?
- Is the data used to calibrate ECL models statistically valid?
- Are recent forecasts of relevant economic factors available?
- Are macroeconomic scenarios complete?
- Are the scenario design and probability weightings appropriate?
- Has the impact of modifications to existing loans (for example, payment holidays, covenant waivers) been reflected?
- Are the staging approach and triggers to determine significant increase in credit risk (SICR) appropriate?
- Are there any other simplifications and, if so, are they appropriate?
- Have government relief programmes been appropriately taken into consideration?
- Have events arisen post-model-run that require adjustment?
- Has the overlay been reviewed, together with the end-to-end ECL modelling process, to ensure that there is no potential for double-counting? For instance, taking into account any:
- top-down adjustments already incorporated by worsening economic forecasts;
- staging adjustments due to economic expectations already included within PDs;
- expectations about future losses included in historical data during model calibration; and
- adjustments to ‘days past due’ data already included in other top-side adjustments.
- What individuals and committees have provided input or review?
Addressing and documenting these questions helps to ensure adequate processes upfront and to prevent challenges over time as initial limitations (that is, those giving rise to the need for an overlay) are resolved. For instance, among other challenges, the absence of a documented rationale might make it more difficult to determine in future periods whether the overlay is still required.
How do overlays affect disclosures?
Overlays might require additional disclosures, and they might impact others. For interim periods, if there has been a significant change since the most recent year-end in the approach to estimating ECLs (whether due to changes in core models, overlays or otherwise), IAS 34 could require additional disclosures. These could include disclosures about inputs, assumptions and estimation techniques under IFRS 7 and IAS 1. The effect of overlays on disclosures that provide information on a granular basis (for example, disclosures by stage, segment and so on) will need to be thought through in order to determine whether and how the overlay is pushed down, or whether it is (or can be) presented separately. Careful consideration of the disclosure implications of overlays at an early stage might help to ‘tell the story’ later on and ensure that potential complexities are addressed.