Expected credit losses model – a recipe for more confusion?
22 November 2013
One thing that everyone seems to agree on following the Financial Crisis is that the incurred loss impairment model is past its sell by date and needs to be replaced by an expected loss model. Unfortunately this consensus starts to disappear when the next question – what do you mean by an expected loss model? – is asked. I blogged a few months ago about the differences between the IASB and FASB versions of an expected loss model. Sadly, no signs are emerging from the Boards’ re-deliberations of their respective exposure drafts that they are getting any closer to a single solution. Different models in the two internationally recognised GAAPs can only lead to confusion.
This however is not the only source of confusion. Regulators talk openly about the need for banks to set aside more capital for “unrecognised” losses and large numbers are quoted. It is rarely clear whether these numbers reflect the difference between an incurred loss model and an expected loss model as envisaged by either the IASB or the FASB, or whether they reflect an allowance for future possible stresses that goes beyond accounting models. This can create an expectation that banks should be recording these stressed numbers in their impairment provisions today. But how does this fit with an accounting framework designed to report past performance? Delays to finalising the accounting models have fuelled this lack of clarity, meaning that confidence in the strength of bank balance sheets is further undermined.
There is also a danger that users will fail to appreciate the degree of subjective judgement required in an expected loss model. It sounds quite precise but in practice requires even more judgment than an incurred loss model. Judgment inevitably brings the likelihood of differences in application by preparers, even where the same model is applied. As an auditor I recognise that auditing expected loss models is a major challenge, requiring us to develop some new approaches.
How do we resolve this recipe for confusion? I certainly don’t have the answer. While a single accounting model would help, we are already past the point where this would be enough. We have a major communication challenge explaining what each model, whether accounting or regulatory, is seeking to do and what the key sensitivities are. If we fail to meet this challenge we face a real risk of challenge as to whether the accounting is delivering useful information.
What do you think?