IFRS 9 models in financial instruments and impairment regulations: The new reality and lessons learned (2024)

The COVID-19 pandemic emerged just a few years after the debut of International Financial Reporting Standard (IFRS) 9. The pandemic undoubtedly stressed the model and framework in unforeseen ways, posing significant challenges to banks’ loan-loss provisioning levels. In the wake of the pandemic shock and its raft of regulatory and government measures, as well as the recent unparalleled set of risk events, including insecurity in the European energy supply and global inflationary pressures, banks have been gradually planning to recalibrate their IFRS 9 expected-credit-loss (ECL) framework to improve their accuracy and embed lessons learned. Yet, even as they contemplate required changes to the models, new shocks are emerging given the highly uncertain macroeconomic outlook.

Now more than ever, banks need to act and build on some of the lessons learned. The European Central Bank has also undertaken a retrospective to see how well banks are prepared to deal with an expected increase in distressed debtors. With this in mind, we have outlined a five-part modular approach banks can use to evolve IFRS 9 to calculate provision levels more accurately, develop efficient governance, and improve pricing and the customer experience. One bank has already leveraged this approach to reduce its retail IFRS 9 provisions by 20 percent.

Trends are posing challenges

A recent report from the European Banking Authority (EBA) evaluated the quality and consistency of the ECL frameworks implemented by EU banks during the COVID-19 pandemic. The report discloses the financial conditions of 130 banks across 26 EU countries, including information on banks’ exposure and asset quality over time. Key takeaways from the report include the following:

• The coverage ratio (or loan-loss rate) for stage 1 and 21Stage 1 assets are performing; stage 2 assets have a significant increase in credit risk. credit exposures edged up in 2020, while the nonperforming loan ratio declined. The phasing out of pandemic-related measures, such as moratoria on loan repayments and public guarantees, is likely to affect asset quality (Exhibit 1).

1

• The cost of risk (COR) during the pandemic varied considerably across banks and geographies. For banks, this was due to a combination of risk-based drivers (for example, risk profile of the portfolios), and nonrisk-based drivers (for example, applying different methodologies to measure ECL impairments). For countries, different government support measures were largely the reason for COR variations.
• The proportion of stage 2 exposures and ECL amounts increased during the pandemic. This is largely because banks applied overlays to cope with data ambiguity, lack of prior history on the performance of moratoria, forbearance measures, latency of the underlying credit-scoring models, and the perceived decrease in borrowers’ resilience. In the second quarter of 2021, these stage 2 proportions continued to increase (Exhibit 2).

2

• COR jumped 50 percent in the first quarter of 2020 compared with the previous quarter because banks booked significant provisions on performing loans. There has not been any reversal in the COR, with aggregated levels stable at around 70 basis points.

As a result of these trends, lenders using IFRS 9 models faced several challenges. One is that the assessment of a significant increase in credit risk (SICR) was based on inaccurate or incomplete information. Another is that the probability of default (PD) was not sufficiently sensitive to forward-looking information and nonlinearities. Third, lenders needed to apply overlays more frequently, yet they didn’t always justify or quantify them. Finally, assets were inaccurately classified and measured.

Managing IFRS 9 issues with a modular approach

With this context, our 2021 survey of commercial banks across Europe helped us to better understand how they managed throughout the pandemic and how they plan to evolve their books of business. As we suspected, and in keeping with the EBA report findings, a majority of our survey respondents said that their next priority is model redevelopment to make ECL estimates more accurate and to meet supervisory expectations. Our survey questions fell into four main categories:

• Impact of the COVID-19 pandemic and application of policy response. Less than 30 percent of the respondents’ newly originated/refinanced exposures—across retail, small and medium-size enterprises, and corporate sectors—are subject to COVID-19-related state guarantee measures. And, generally, banks don’t expect COR to increase once the public support measures are fully withdrawn.
• Modeling implications. Two-thirds of banks plan to redevelop IFRS 9 models in the next two years, and almost three-quarters plan to incorporate sector differences. All banks in our survey have adjusted their SICR criteria temporarily and 76 percent use overlays to do so. They also expect heightened regulatory scrutiny around overlays and manual adjustments, which may be why 86 plan to return to prepandemic criteria within two years. Also, banks are now considering how to mitigate their models given the highly uncertain macroeconomic outlook.
• Implications for bank capital. Banks remain very concerned about how creditworthiness and portfolio quality will change when support measures are withdrawn; about 70 percent of them use existing IFRS 9 transitional arrangements to help alleviate the impact of the pandemic on their portfolios.2Under the International Financial Reporting Standard (IFRS) 9 transitional arrangements, institutions were allowed to fully add back to common equity tier 1 any increase in expected losses recognized in 2020 and 2021 for financial assets that are not credit impaired. The original ending of the transition has been extended from December 2022 to December 2024 via the cash reserve ratio quick fix in 2020.
• Impact of supervisory developments. Regulators have already begun a dialogue with 43 percent of respondents regarding upcoming on-site inspections for credit risk and provisioning, and banks have already started to prepare for them.

In light of these findings, we have developed a five-part, modular approach (diagnostic, staging assessment, forward-looking PD, overlays, and monitoring) that banks can use to evolve IFRS 9 to calculate provision levels more accurately, develop efficient governance, and improve pricing and the customer experience (see sidebar, “Use a modular approach to evolve International Financial Reporting Standard 9 models”).

Using the modular approach to improve provisioning models

A large European bank has already leveraged the modular approach to enhance the accuracy of its provisioning models. Rapid diagnostics of the bank’s methodology and approach established that its provision levels were too conservative. Eight levers were identified to drive accuracy in the bank’s measurement of IFRS 9 provisions, which were subsequently validated by the auditor.3IFRS 9 methodology and calculation results require auditing in order to be formally approved. The findings and remediation were aimed at mitigating the conservatism inherited from the pillar 1 models, as well as addressing weaknesses in the macroeconomic modeling supporting the forward-looking component. A new baseline was set, revising the staging criteria (thresholds for SICR), resulting in a 20 percent reduction in total retail IFRS provisions—10 percent immediately and an additional 10 percent the following year.

EU legislators and regulators have issued clarifications about how to apply IFRS 9 classification criteria and ECL models in the context of the COVID-19 pandemic, yet they’ve left considerable discretion to banks. So there is no time to waste. Banks need to evolve their models quickly—especially given the impending withdrawal of government support for many debtors. Getting these models right has significant competitive advantages, including more accurate provisioning levels, better and more efficient governance, improved pricing, and a better customer experience.