Expected credit losses for banks under IFRS 9 - what next?

04 September, 2020

Chris Wood headshot photo Chris Wood
Partner, Accounting Advisory Services, Toronto, PwC Canada

Considerations ahead of 2020 year-end

I’m somewhat of a connoisseur when it comes to good tequila and, as luck would have it, my favourite anejo (not sold in stores) is top-shelf at a local cantina. During my first visit there in months last week, it was a relief to feel that things might become normal again. Sort of. Like many, I’d hoped for a quick lockdown, V-shaped recovery and rapid vaccine. Yet with the first two ‘crisis quarters’ behind us, it’s becoming clearer that’s unlikely to be the case. For estimating expected credit losses (ECLs) for banks under IFRS 9, that warrants a pause to think about where we are and what challenges might be next, including any necessary changes to adapt ahead of year-end reporting if crisis uncertainty continues to loom.

Thus far, billions have been booked in incremental provisions, while defaults are yet to rise. In many cases, delinquencies haven’t even transpired thanks to payment holidays and enormous government support. So enormous, in fact, that some individuals have found themselves better off than before. For banks, that means a few things. Firstly, losses are likely coming. Provisions set aside will be consumed as delayed defaults come to fruition. Secondly, risk will continue to be very challenging to measure. Put simply, normal rules don’t apply. Being jobless or unprofitable usually reduces one’s ability to pay, which leads to delinquencies and ultimately to defaults – not true (immediately, at least) when money’s free and repayments are on hold. 

What’s next? Optimistically, I still hope for a quick resolve – a fast return to life like we once knew (with a few exceptions, like in-store shopping and travel, perhaps). Rationally, I acknowledge that may not be the case. Of course, in many ways, all that we really know is that we don’t really know. That’s important since what comes to pass will determine the credit losses that eventually transpire. In estimating ECLs today, we average the good, the bad and the ugly by probability-weighting different outcomes. In real life, we don’t. For that reason, provisions are sure to evolve as reality unfolds. 

Given all of the above, to date most have adapted their estimates by moving away from core models and doing new or different analytics, adding a lot of (necessary) judgement along the way. Thinking ahead to year-end, things to focus on now are:

  • continuously refining the estimate as conditions evolve (for example, reflecting epidemiological developments and changes to government stimulus packages); 
  • assessing whether adding more alternative scenarios, when modelling estimates, might be a better way to address increased uncertainty; 
  • testing judgements, including by comparison to peers, market inputs (such as credit default swap spreads), current pricing of similar loans, and consistency with fair value disclosures; 
  • focusing on governance, process and controls, in order to bring the same level of rigour to new or changed methods as would normally be applied to core models (a particular challenge for Sarbanes-Oxley or similar requirements); 
  • thinking differently about validation, including whether overlays and other adjustments can be covered by similar processes; 
  • considering what new data might become available (such as cash flow information from the borrower’s different accounts or changes to government programmes) and how it will be collected and incorporated into estimates; 
  • planning for an agile future by rethinking the ‘model’ definition and implementing frameworks that allow for quick pivots from core models to risk analytics without having to compromise on quality; and 
  • putting transparency first, by considering what disclosures might be most useful for users, depending on how things might shake out between now and year-end. 

The only certainties right now are that none of us really knows where we’ll be in six months’ time, and that what worked in the past isn’t working now. Focus to date has been on finding ways to estimate until things become clearer. In the event they become less clear, it’s worth figuring out how to migrate from short-term fixes into longer-term solutions. 

Our guest blogger is Chris Wood, Banking Partner and IFRS 9 specialist, connect with him on LinkedIn here.

All views expressed are the authors

Contact us

Follow us