COVID-19 impacts and considerations for expected credit loss provisions, scenario analysis and credit-risk modelling.
COVID-19 impact and considerations
The global COVID-19 crisis affected businesses and lives in unimaginable ways. Countries were affected differently, with many at different stages of the curve and experts anticipating first and second waves over a prolonged cycle. Banks—which are pivotal stimulants for economic development and growth—were at the heart of the crisis.
IFRS 9 ECL stress testing and scenario analysis is expected to provide a range of outcomes that allows for this uncertainty. However, even with macroeconomic and forward-looking scenarios in place, it would have been difficult to imagine any model convincingly forecasting the pandemic scenario the world woke up to. ECLs are expected to be neither overly optimistic nor overly cautious, but unbiased, probability-weighted best-estimate outcomes. This could imply drawing from a set of plausible scenarios. COVID-19 could arguably qualify as an outlier event.
This article focuses on expected credit loss provisions for loan exposures as one key area of financial disclosure under the spotlight. It considers possible impacts and key questions arising from the crisis. Similar principles could apply to other bank exposures and financial sectors.
Governments and regulators guiding the way
For most jurisdictions, including ours, public policymakers and regulators led the way by providing guidelines. There were several reasons for this.
COVID-19 could have turned out to be a short-term phenomenon. Adjusting IFRS 9 ECL models too quickly could therefore have been premature. These models contain lifetime default likelihoods that are difficult to adjust properly using supportable past and future information. Consistent and stable approaches remained preferable.
There was also a need to prevent a rush towards overprovisioning due to assumptions built into ECL models. COVID-19-induced effects on businesses were broadly being prevented from flowing mechanically into model outcomes. Many borrowers were ready to invoke force majeure, and where adjustments were considered, they were being made cautiously.
Banks operating in Zimbabwe were asked to prepare business impact assessments by 30 June 2020. IFRS 9 ECLs were potentially among the most difficult exercises to conduct with confidence.
Impact on ECL models
Assuming no external relief measures, COVID-19 could warrant a deterioration in probability of default (PD), loss given default (LGD) and exposure at default (EAD) parameters. A key driver would be the outcome of credit assessments that identify significant increases in credit risk (SICR) for different borrowers.
Credit-risk assessments
The effects would vary across sectors and portfolio segments. Even after the crisis, some sectors could recover more easily than others. Stay-at-home and social-distancing orders severely affected transport, for example, while increased reliance on digital platforms could present a different picture for the technology sector. Similar distinctions would apply to individual borrowers.
One major challenge was separating borrowers genuinely affected by COVID-19 from those already due for a downgrade to stage 2 or stage 3. This could require more thorough credit assessment and potentially revised credit policies.
In Romania, some sectors were issued Certificates of Emergency Situation by a government ministry. Banks could potentially rely on such certificates when estimating expected-loss impacts. Similar mechanisms could help institutions define who should receive concessions and how performing and non-performing exposures should be classified.
Model adjustments for COVID-19
Several considerations arise when model adjustments are contemplated.
Macroeconomic scenarios and the weights applied by banks need to be developed using supportable information. The IASB advised that it was difficult to isolate the specific effects of COVID-19 and government measures at the time. Where scenarios could not be developed reliably, post-model overlays or other adjustments could be considered.
Any overlay approach should be backed by sound credit-risk policies and governance, which may also form part of the disclosures made.
The European Securities and Markets Authority (ESMA) noted that where issuers faced difficulty isolating COVID-19 effects at instrument level, assessment on a collective basis could be sufficient under IFRS 9 paragraphs B5.5.1–B5.5.5. ESMA also noted that where temporary borrower relief did not significantly affect the net economic value of a loan, the accounting modification could be considered non-substantial.
The European Central Bank recommended that banks give greater weight to stable, long-term outlooks supported by historical information and consider relief measures granted by public institutions.
The temporary nature of many COVID-19 measures could have had a limited effect on lifetime ECLs. Most relief measures were designed to minimise short-term liquidity strain on borrowers caused by public-health interventions.
Care was needed to avoid procyclical assumptions and undue short-term volatility in reported ECLs. Such volatility could also make parameter back-testing difficult to understand and justify.
General regulatory guidance cautioned against automatically moving borrowers into stage 2 solely because of measures such as payment holidays.
The combination of relief measures could also lead banks to revisit—and potentially rebut—the 30-days-past-due presumption in their staging decisions.
Different products would experience different increases in risk. Secured and unsecured loans, for example, would behave differently, as would revolving facilities and term loans. The effort required for individual assessment could also be prohibitive, leading institutions towards probabilistic adjustments at portfolio level.
Parameters might need to be remodelled after an observation period. The appropriate waiting period would depend on the pandemic ending and normality returning sustainably—an inherently judgemental decision. Readily available and supportable historical information would remain a primary source for building models and testing their robustness.
Sovereign-backed guarantees used as stabilisation measures could result in LGD adjustments. Bank guarantees might have a similar effect.
Collateral valuations could also experience significant haircuts, increasing uncertainty in ECL outcomes.
Going forward
At the time, the pandemic was still unfolding. Projections ranged from V-shaped to U-shaped and L-shaped recoveries, all with high uncertainty. Governments and regulators were likely to continue taking steps to contain the virus and support their economies. Because those steps were introduced progressively, IFRS 9 ECL models faced substantial practical adjustment challenges.
Most banks had already built buffers into their capital reserves. Regulators therefore had to consider whether banks should be allowed to use those reserves to provide further lending during the crisis. The Basel Committee on Banking Supervision advised that transitional relief for IFRS 9 could be applied at jurisdictional level, supported by enhanced reporting that illustrated full capital positions.
Where a loan was modified, changes in the present value of contractual cash flows would be reflected as a modification gain or loss. This could change the exposure amount used to calculate risk-weighted assets.
What will the post-pandemic period be like?
How suitable would existing models be for the world emerging from the crisis? New loan originations could become constrained as banks reduced appetite for certain exposures unless government guarantee schemes supported borrowers. Collections could become more difficult, while loss given default could increase as asset values fell.
Separating the impact of COVID-19 from genuine business failure would remain challenging. Even facilities that remained in stage 1 and appeared healthy could later experience difficulties in sectors that were slow to adapt. Tourism and agriculture, for example, might face prolonged recovery periods.
IFRS 9 does not provide bright-line rules or a mechanical modelling approach for stressed conditions. The standard contains significant areas of judgement. If a bank develops a bespoke approach for its exposures, supports that approach with evidence and applies it consistently, ECL provisions remain a powerful tool for communicating with users of financial reports.
If you would like to comment or discuss this article, contact the author at actuary@claxonactuaries.com.
Authored by the Claxon Actuaries Team.