IFRS 9 bank disclosures
Half-year reporting adds new insights on impairment
30 August 2018
Most large European banks have now issued their half-year reports – the first ones under IFRS 9. These have helped us to start building a picture of the dynamics of reporting under the new standard.
It’s a picture that has been developing gradually: through banks’ 2017 annual financial statements and first quarter reports, along with the valuable information in transition reports issued by large UK and some other European banks.
Now, with half-year reports being issued, new insights have emerged – largely consistent with what we’ve seen previously, but with more detail.
Most of the disclosures are for impairment as this is the largest area of impact. Banks have made limited disclosures on the impact of the new classification requirements. In this blog, we’ve selected three areas of interest regarding IFRS 9 implementation.
Criteria for transferring assets between stages
Under IFRS 9, financial assets are allocated to one of three stages.
Click to enlarge graphic (JPG, 88 KB)
Each bank develops its own criteria for when an asset is transferred from Stage 1 to Stage 2 – this is one of the most significant judgement areas in the new impairment model.
All banks we looked at consider both quantitative and qualitative factors for Stage 2 transfers. The quantitative element is usually probability of default (PD). The qualitative factors include placing assets on watch lists, or adverse credit bureau scores for retail portfolios. Two banks in our selection apply a statistical approach based on quantile regression.
Almost all banks apply the back stop of 30 days past due – at which point an asset is always transferred to Stage 2 – but it is generally not used as a primary indicator.
There’s diversity in how the quantitative analysis is applied. A number of banks regard a doubling of the PD estimated at initial recognition as a threshold at which an asset is moved to Stage 2, but most of these disclosed that they also have more granular criteria.
Many banks have different thresholds for different portfolios, often differentiating between retail and wholesale exposures. One bank applies an 8x increase for retail mortgages and 4x for consumer lending. Another uses a range of PD increases: from 15 basis points to 2x, depending on the credit risk at initial recognition. Most banks use absolute thresholds in addition to a relative increase.
Is there a ‘right’ approach?
A question often asked is: Are my transfer criteria right and how do I check whether they are?
Banks are using a number of indicators to test how appropriate their transfer criteria are. One is to what extent the transfer timing aligns with an exposure being 30 days past due. Generally, exposures would be transferred to Stage 2 before they become 30 days past due, because the ECL model is forward-looking while overdue status is a backward-looking indicator. I have looked at disclosures in this area in my previous blog.
Another indicator is to what extent exposures frequently move between Stage 1 and Stage 2. If they do frequently move backwards and forwards, this may indicate that the transfer criteria are too sensitive.
We’ve tried to develop our understanding of this by looking at the amounts of ECL allowance transferred from Stage 2 assets to Stage 1 assets as a percentage of the opening ECL allowance balance for Stage 2 assets – as shown in the chart below. These transfers took place over the first six months of reporting under IFRS 9. The selection comprises large European and Canadian banks that have disclosed this information.
Click to enlarge chart (JPG, 116 KB)
The range shown above for all assets has been made wider due to the inclusion of one bank for which the ratio was 80% (the same bank’s ratio for retail mortgages was 60%). Excluding this bank, most others showed a ratio of 20–30% for all assets. It will be interesting to see more detail in this area when the annual financial statements for 2018 are released.
Forward-looking economic scenarios
As IFRS 9’s impairment model is forward-looking, banks are required to consider future economic scenarios to calculate ECL. We looked at how a selection of 20 European and Canadian banks have approached this task. Most banks use three to five economic scenarios – as illustrated in the pie chart below1.
Click to enlarge chart (JPG, 106 KB)
The chart below illustrates, for a selection of seven European banks that disclosed this information, the probabilities assigned to the economic scenarios that they used.
Click to enlarge chart (JPG, 131 KB)
The probabilities allocated to the base scenarios in our selection vary (from 30–80%). The base scenario would normally align with the inputs used by a bank for budgeting, strategic and capital plans and for other aspects of managing and reporting.
Disclosure of the sensitivity of ECL numbers to changes in assumptions has been an area of some debate. The impairment model in IFRS 9 requires a lot of judgement and various stakeholders are interested to see what the impact would be if judgement were exercised differently.
The large UK banks have started to disclose information in this area. The most frequent type of disclosure we’ve seen is how much the ECL has changed by including additional forward-looking economic scenarios to the base scenario. We’ve seen six large banks disclose this information and it’s interesting to observe that the range of sensitivities of the amount of impairment to inclusion of additional scenarios is relatively wide. The chart below provides more detail.
Click to enlarge chart (JPG, 88 KB)
One bank explicitly stated that in their 2018 annual financial statements they will disclose more quantitative information on sensitivities to management judgement relating to forward-looking scenarios, probability weights, Stage 2 thresholds and post-model adjustments.
We’re still busy analysing information that has been issued to date. We’ll be back before the year end with additional insights and some thoughts on presentation and disclosure in the annual financial statements.
1. We included a similar chart in a previous blog but the selection was only 15 banks.