ICAEW.com works better with JavaScript enabled.

IFRS 9: Six variables to watch in a stress test

The Financial Services Faculty looks at six aspects of the Bank of England stress test and how the interaction with IFRS 9 Financial Instruments may differ in a real stress.

In the UK, 2018 was the first year banks reported their results under the new International Financial Reporting Standard, IFRS 9. However, it has so far only had a limited impact on banks' credit provisions and their common equity tier (CET1) capital ratios, both in financial reporting and under stress.

Low unemployment and low interest rates may be part of the reason. Secondly, a long period of benign conditions means models used to predict losses have been built on unrepresentative data, which does not reflect how credit risk changes through an economic cycle. Providing the data points to calibrate the models appropriately may unfortunately require a real recession.

The assumption of perfect foresight v reality

The Prudential Regulation Authority’s (PRA's) stress tests use the concept of perfect foresight. This means that the end of the economic downturn is known and banks can respond and reduce their expected credit loss provisions after the lowest point of the stress. This is because they know conditions will be improving.

In a real stress, banks, regulators and auditors may be reluctant to call the bottom of an economic cycle and take actions such as reducing provisions at the trough (low point). This would be for prudential reasons.

We are likely to see higher provisions under IFRS 9 at the low point of the next economic cycle than those reported in the 2018 stress test results. When coming out of a real downturn, it will be difficult to release provisions at the right speed as banks will not be able to precisely anticipate an economic recovery. 

Will a real stress mean a larger impact on expected credit losses?

Yes No

Provisions will remain higher as banks and auditors will be uncertain or pessimistic about economic prospects during a downturn and it will be unclear whether poor economic conditions will continue or improve. A prudent approach will mean provisions remain high.

There may be material increases in credit provisions as loans move through the IFRS 9 stages and provisions increase from 12 months (stage one) to lifetime expected losses (stage two and three). In particular, the change in probability of default (PD) could drive cliff-edge effects.

As the downturn begins, banks would start moving assets from stage one (performing) to stage two (under-performing) as assets experience a significant increase in credit risk. As assets move to stage two, banks provide for lifetime expected losses (LEL).

As the stress worsens, and loans move from stage two to stage three (non-performing), there should not be a material change to credit provisions or a cliff-edge effect. LEL provisions would be held when assets are classified in both stage two and stage three.

Influence of human factors on forbearance decisions

As seen in the 2008 financial crisis, the human factors influencing banks’ decisions may mean more instances of forbearance (repayment relief) than repossession.

Banks may revise their forbearance policies materially if a stress emerges. The evolution of house prices and the approach to forbearance will influence how long banks are willing to give creditors to recover from difficulties. For example, forbearance policies could be extended over a five to 10-year period.

Different approaches may become more aligned under a stress

Banks have taken different approaches to some of the judgements and decisions required under IFRS 9. For example, the trigger for moving loans into stage two (a significant increase in credit risk) is interpreted in many ways, with no single right answer. This may become more consistent over time. As there has only been one year of IFRS 9 reporting in the UK, the accuracy of signals pointing to a significant increase in credit risk have not been tested. The variability of these triggers might lead to more market volatility. To avoid this, regulators may seek to promote a particular stage two trigger. For example, the European Banking Authority (EBA) has indicated a doubling or tripling of the probability of default (PD) is an appropriate trigger for a significant increase in credit risk.

In practice, we are likely to see herd behaviour from banks in reporting their overall provisions through the economic cycle. It is likely that banks will seek to move at the same time on provisioning through a stress, rather than stand out by releasing provisions earlier than others.

The difficulty judging an economic cycle’s low point

It is vital auditors demonstrate professional scepticism, even if the prevailing economic view is for an improvement. If a bank has experienced recent losses, but is confident that recovery is imminent and decides to release provisions, the auditor must conclude how the decision was made and consider the robustness of the data and analysis. This will probably be assessed in conjunction with specialist economics teams.

Banks’ capital structures add complexity to the process and mean making an accurate call is essential. The use of contingent convertible (Cocos) bonds, which can be bailed in as additional tier 1 (AT1) capital, adds pressure to all parties. It increases the need for professional judgement and scepticism on the part of auditors, particularly if the bonds are close to trigger levels and a reduction in provisions would improve earnings, capital and capital ratios. It would therefore help the bank avoid triggering its AT1s and conversion.

Challenging accepted economic data and assumptions

There are challenges around the consensus economic outlook and the accuracy of available economic data. The house price index (HPI) path may actually be trending downward and might be flat at best. However, some banks might adopt a mean-reversion model to HPI, meaning they continue to expect a rising house price trend.

This might reflect the position in an economic cycle where a downturn is at the early stages and some banks may predict conditions will not deteriorate much further. The variability of forced-sale discounts, for example, demonstrates the various opinions across banks. Similarly, while GDP has fallen, the consensus economic outlook in this scenario remains positive.

The UK may be going through a real-life experiment on how banks interpret mixed data and what it tells them about where we are in the economic cycle. Auditors (and banks) must be sceptical about current economic data, which is now included in financial reporting.

The impact of HPI on expected credit losses (ECL) will be non-linear. This means there could be minimal, if any, impact if prices start to fall, but credit losses could increase substantially if greater price falls are expected.

Auditors should consider:

  • Has an appropriate downside scenario been included in the multiple economic scenarios used to determine expected credit losses?
  • Has an appropriate probability been attached to the scenario?
  • Do the percentages used fit a reasonable distribution?
  • Has non-linearity been accounted for?
  • Which scenario is driving the outcome, for example, the central or the downside scenario?

Economic impact of regulatory action in a stress

Regulators are able to use different tools to mitigate the impact of a recession or economic stress. Given its relatively recent introduction in 2014, it is unclear how the counter cyclical capital buffer (CCYB) will operate through a stress. If the CCYB is released slowly, this may create a drag on any potential economic recovery. The regulator’s use of CCYB will come under scrutiny as it is, or not, used to offset the effects of IFRS 9 requirements. Supervisory concerns for individual banks and the economy will have to be reconciled to the Bank of England’s Financial Policy Committee position.

More creative policy responses may be needed to prevent undesirable outcomes. There are other regulatory measures that may act to limit economic recovery. The EBA says that while assets can move from stage three to stage two, three years must pass before stage two assets can move to a stage one classification. Such a policy could magnify the economic stress and cause a drag on recovery.

Practical implications

The PRA stress test made an important assumption on perfect foresight to aid comparability. There are other differences in bank approaches that will have a significant impact on their stress test results. These are considered below:


The probability that a customer would pay off their mortgage early in a stress would fall dramatically. This could impact the time elapsed to default and any income derived from the early repayment charge. Prepayment models will vary by bank and by asset class (for example, mortgages, small and medium-sized enterprises (SMEs) and unsecured).


When an asset moves to stage two, many banks will consider deploying forbearance tools. It is unclear whether banks should be predicting the probability of using forbearance and how it varies through the cycle.

Credit cards and revolving credit

Due to the number of interpretations and judgements around revolving credit, there are numerous potential outcomes during a stress. Modelling revolving credit is difficult, so the accounting is more problematic. For example, what is the life of a credit card (the life of the piece of plastic, the life of the account, or a re-agreement of terms). Until the life of a credit card is clear, it is difficult to conclude what the amount of lifetime expected losses should be.

The interaction between IFRS 9 and stress testing will be covered at the Faculty’s 15 October banking CPD work shop.

Further reading