Over the last few weeks we published a series of three articles analyzing the IFRS9 allowances and provisions of the Big Six Canadian banks based on their disclosures for Expected Credit Losses (ECLs) and our experience with IFRS9 implementations. Although IFRS9 has been in force since 2018, it is only recently that it has shown how it could significantly impact provisions and earnings.
- The purpose of this article is to point out key areas of improvement for IFRS9 disclosures based on our analysis and recent guidance from a UK Taskforce. The recommendations should also be useful for the forthcoming CECL disclosures, starting with 2020Q1 for the SEC filers (excl. SRCs).
The timing of this article is perfect as the US Securities and Exchange Commission (SEC) recently issued a proposal to update the statistical disclosures for banking registrants and eliminate disclosures that overlap with Commission rules, U.S. GAAP or IFRS. Even though the Commission is not proposing any new disclosures related to CECL at this point, it has sought comments for refinements of disclosures vis-a-vis the new standard.
As a refresher of our series of articles, Part I dealt with the possible causes of the reported increase in provisions and the significant impact of provisions on earnings. Part II examined the sensitivity analysis that the banks provided in their disclosures with a focus on macroeconomic scenarios and procyclicality. Part III analyzed the variance in ECL across banks and products and identified emerging patterns with respect to the credit cycle.
None of our analysis and conclusions would have been possible without the disclosures provided by the banks under IFRS9. They are similar to the disclosures the US banks will be providing under CECL. Thus, the adoption of the new accounting standard allows the reader of the banks’ financial statements to better understand the credit risk in the balance sheet of the banks. It therefore enables higher transparency and comparability, despite the variance in the reported allowance and provisions across the banks.
- Is there room for improvement in the disclosures? Which parts of the disclosures need improvement the most?
The answer to the first question is yes. As we were writing the above articles, a UK Taskforce on Disclosures about Expected Credit Losses issued an updated guidance (Dec 13 2019): Recommendations on a comprehensive set of IFRS 9 Expected Credit Loss disclosures. This taskforce was established in November 2017 by three UK regulators as a partnership between the preparer community and the investor and analyst community. Kudos to the taskforce for the excellent set of recommendations.
The recommendations, organized along nine themes, proved very useful as we were analyzing the existing IFRS9 disclosures from the Big Six Canadian banks. They helped us identify the parts of the disclosures that need improvement the most in order to further enhance the transparency and comparability of the allowance and provisions across banks.
We therefore list few key recommendations which can be readily adopted by banks without creating extra reporting burden. We also cite the related recommendations from the above report. The recommendations apply for CECL as well as IFRS9, unless stated otherwise.
Forward-looking information
- Disclosing the weightings of the scenarios and explanations for any period-on-period changes in those weightings (see C.3 & C.6 recommendations). Canadian banks only provide a sensitivity analysis of reported vs. baseline ECL. This is helpful but it is not sufficient for understanding the level of prudence in the ECL figures, and how much the non-linear impact from macroeconomic factors weighs in on ECL.
- Presenting the key macroeconomic factors for a scenario in graphs (see C.4 recommendation). Such presentation is better than the existing tabular form of average annual rates from the Canadian banks. A graph can show to users when any peaks or troughs are assumed to occur and whether values are assumed to revert to a long-term rate. All things being equal, a forecast trough in the near future is expected to yield a more severe ECL impact than one of a few years out.
Movement and coverage across stages (IFRS9 only)
- Disclosing how much of the movement from Stage 1 to Stage 2 was caused by quantitative and/or qualitative assessments versus the 30+ days past due backstop (see D.1 & F.5 recommendations). The portion of the transfers due to the backstop provides insight into how forward-looking the staging criteria are. Although Canadian banks provide delinquency information the latter is not combined with ECL and staging.
Changes in the balance sheet ECL estimate
- Providing more information about the quantitative impact that changes to models and changes to risk parameters have on the reported numbers (see E.1 recommendation). A risk parameter is an input to a credit risk model and is expected to change from period to period. In contrast, model changes are expected less frequently. In their FY2019 annual reports the Big Six Canadian banks attributed the changes in ECL and allowance to changes in risk parameters and models. However, only three of the six banks have distinguished between the two sources of change in their disclosures. As stated in the Taskforce guidance, users have indicated they would like to see a distinction between the two impacts on ECL. This is because changes in ECL due to risk parameter changes, which indicate changes in the credit quality of the assets, are different in nature to changes in ECL due to methodology and models which are under the control of the bank. This recommendation can be used as feedback to the request for comment, #48, in the SEC’s proposal.
- E.1 Example 1 in the Taskforce report illustrates the disclosures stemming from the E.1 recommendation, including the distinction between model changes and risk parameter changes. It is the ECL attribution rollforward that ABA has also advocated in its related discussion paper. As pointed out in the latter paper, a proper attribution rollforward can go beyond explaining changes in credit risk. It can serve as a tool for evaluating future profitability and risk and hence assisting in business planning.
Credit risk exposures
- Quantitative disclosures of credit risk rating by asset class and for each stage that includes corresponding ECLs and gross carrying amounts (see F.1 & F.2 recommendations). The Big Six Canadian banks do provide such disclosures. The F.1 & F.2 example in the Taskforce report can be used as feedback to the requests for comment in the SEC’s proposal, in particular requests #42, 44 & 45.
Measurement uncertainty, future economic conditions, and critical judgements and estimates
- Qualitative and quantitative disclosures of sensitivities to key assumptions in forecasts of future economic conditions (see G.1-G.4 recommendations). The Big Six Canadian banks perform some of the multi-factor sensitivity analysis related to the economic scenarios as per G.4. These four recommendations can also be used as feedback for SEC’s request for comment #48, particularly for the disclosure of the material qualitative adjustments.
Last but not least, drawing upon the experience from the Great Recession, where some of the losses were due to changes in loan origination and portfolio mix, FASB is requiring vintage information in the ECL disclosures for the banks that are public business entities (PBEs). This is not yet a requirement for IFRS9 disclosures.
As pointed out in the aforementioned ABA discussion paper, understanding how the various loan portfolios perform based on loan age is a critical aspect of credit risk. In addition to non-linearity to the macroeconomic factors, loans in general and mortgages in particular behave non-linearly with respect to age. In fact, age of the mortgage is one of the key predictors for loan default. This is why vintage-based forecasting methods are most appropriate for asset classes with strong aging effects on repayment behavior as we have shown elsewhere.
Hence, measuring and disclosing the expected loss at the vintage level, namely a vintage-based attribution rollforward, provides the proper level of abstraction for assessing credit risk in the loan portfolios of a bank both internally, by the board of directors, and externally, as deemed necessary, by users.