Disparate times and dialectic methods: credit risk today

19 October, 2020

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

Remember Plato’s cave? Prisoners saw shadows on a wall, but they couldn’t turn to see the objects being used to cast them. In the end, the prisoners mistake the shadows for the objects themselves and the point, I think, was that understanding ‘what is’ requires not only seeing, but also thinking. In the context of credit risk, that’s more the case now than ever.

My recent conversations with banks have centred on two key issues. First, macroeconomic conditions for some seem to have fared better than expected, widening the gap between current allowances and modelled outcomes. Naturally, those banks are wondering whether to increase overlays, or to begin reversing allowances. The second is that broad use of payment deferrals and government stimulus mean a general absence of delinquencies that are usually key in identifying loans that have become ‘credit impaired’. Banks must now try to understand what alternative indicators could be used. At first, I thought these two were discrete and possibly contradictory – on the one hand, things might look better than we’d expected (potential releases) and, on the other, maybe they’re really worse than what we’ve seen (potential defaults). What I’ve come to realise is that the underlying issue is the same – when it comes to credit risk today, our mind is at odds with our eyes.

Modelled outcomes are getting better because macroeconomic indicators appear to be getting better. True, when economic indicators improve, credit outcomes usually do too and, if losses decrease, so too should allowances. So, why challenge the models? For starters, we’ve questioned their performance in the current environment all along and, while the direction of output may have changed, the underlying limitations might persist. Plus, the coronavirus (COVID-19) pandemic has given rise to a whole new suite of risks that aren’t yet fully understood (or even known). Coupled with the fact that (like many of you) I’m still writing this in my living room, it makes sense to think hard about what we think we see before booking recoveries.

How? I’d focus on making sure that experiences with past crises are leveraged, to adequately capture the relationship of payment holiday duration and probabilities of default (PDs). I’d also consider whether sufficient time has passed to evidence a true recovery is underway, or whether it’s the manifestation of government support, which might not be sustainable. Of course, I’d also ensure that all model limitations have been identified and addressed, such as:

  • unidentified defaults – current default emergence periods might be much longer than normal, such that models might reflect one part of the picture (improved macroeconomics) but exclude another (current defaults);
  • segmentation – sub-sector, geography or other risks might not be captured by PD and macro models that segment at a higher level;
  • fallen angels – specific borrowers or industries might be affected in ways not adequately captured by models alone;
  • data gaps – complete data might not be available; and
  • workouts – governments might restrict strategies that banks can use to workout non-performing loans.

On defaults, just because we don’t see them yet doesn’t mean that they haven’t occurred. We’re used to looking at days past due, but IFRS 9 and the Basel rules also require us to consider things like ‘significant financial difficulty of the borrower’ and ‘unlikeliness to pay’. In some jurisdictions, regulators have been explicit in their expectation that the outcomes (for accounting and capital) should be the same. In short, even if the loan’s current, it’s impaired or in default as soon as you’re no longer going to be repaid in full. Today, when payment holidays are pushing half a year or more, that’s likely to be a significant judgement needing whatever data and information might be available.

The fault of Plato’s prisoners wasn’t seeing the shadows, it was mistaking them for reality. Hopefully, we’ll be different. After all, in our case, shadows aren’t all we have. As The Stranglers put it, ‘There’s always the sun (always the sun)’.

All views expressed are the authors

Contact us

Follow us