IFRS 9 Financial Instruments and procyclicality

Earlier this year the IASB launched a call for evidence as part of its post implementation review of the impairment requirements in IFRS 9 Financial Instruments. Some respondents to the call for evidence pointed out that the impairment model in IFRS 9 is procyclical, a point also discussed by the ECB in a working paper published in June 2023.

Background

IFRS 9 was issued in July 2014 and implemented for annual periods beginning on or after 1 January 2018. The standard replaced IAS 39 Financial Instruments: Recognition and Measurement.

IFRS 9 introduced an expected credit loss (ECL) model for impairments to replace the previous incurred loss model within IAS 39. The change was a response to the global financial crisis in 2007/08 and the criticism that the incurred loss model was backward looking and delayed the recognition of credit losses – ie was procyclical.

Procyclicality matters from a financial stability perspective

A note for the Financial Stability Forum Working Group on Market and Institutional Resilience (2008) explained that ‘The term “procyclicality” is generally used to refer to the mutually reinforcing (“positive feedback”) mechanisms through which the financial system can amplify business fluctuations and possibly cause or exacerbate financial instability.’

Banks can stimulate and provide momentum on the upswing of the economic cycle. There is demand for funding and as banks compete on lower rates and weaker covenants, they add further impetus to the demand for funding. Perceptions of risk decline when times are good and so the supply of funds can be plentiful, including more risky equity and leveraged stakes. As the cycle turns, banks tend to withdraw higher risk funding, but the withdrawal of funds can be exacerbated if banks start to incur credit losses and need to rebuild capital positions. A credit crunch can ensue – with funding drying up generally, not just that which is higher risk - accelerating any downturn and/or exacerbating the severity of any downturn.

IAS 39 was inherently procyclical as credit losses were not recognised until incurred. As an economy went into recession, banks could find themselves flooded by losses eroding their capital and impeding their ability to lend.

It should be noted that, while an incurred loss model is procyclical it need not actually enforce procyclical behaviour on banks. IAS 39 was the public face of a bank’s financial reporting of its performance and financial position. The standard did not permit the early recognition of future expected credit losses, but it equally did not preclude a bank from considering the effects of a future recession and putting in place measures to mitigate the future risk (eg building its capital position in the upswing in readiness for future losses).

IFRS 9 and procyclicality

The approach to provisioning in IFRS 9 incorporates a forward view of credit losses by calculating an ECL. This should dampen the procyclical effects of IAS 39, relating to the recognition of real economic losses – by recognising those losses ahead of the recession.

IFRS 9 does, however, have procyclical features, although overall the ECB working paper found IFRS 9 (and the equivalent US Current Expected Credit Loss (CECL) model) to be less procyclical than IAS 39.

IFRS 9 could be procyclical in different ways: the construct of the ECL calculation and the use of staging; the consequences of making an estimate of the future; and behaviourally due to the application of judgment in the calculation.

IFRS 9 adopts a two-stage approach to ECL. For performing assets, a 12-month ECL is calculated, whereas if there has been a significant increase in credit risk (SICR) since initial recognition, a lifetime ECL is calculated. As a trigger event, SICR could be procyclical in the way that IAS 39 was. As the economy heads into recession, more assets are likely to experience a SICR, and so move to a lifetime ECL calculation. The US CECL does not have the SICR trigger as it is a lifetime loss calculation for all assets. However, as noted in the ECB paper, the consequence of a lifetime loss model might be the holding of a larger stock of provisions.

When calculating an ECL, judgements have to be made about the future which is inherently uncertain. If the future is considerably uncertain the range of projected plausible ECL outcomes can be wide. With the passing of time the range of outcomes will narrow and, eventually, the real economic effect will be known, which in some cases might reveal there was an initial significant over estimation of the ECL. This over estimation is most likely to arise following an adverse shock to the macroeconomy causing significant volatility and past experience provides no guide. This was the case with COVID-19, when banks initially made high provisions due to the elevated levels of uncertainty (eg around the consequences of lock down on the economy), but subsequently released those provisions.

Finally, the standard is principles based which has led to the use of highly complex models with significant judgements. As a result, there is considerable scope for the application of the ECL model to be more or less procyclical depending upon the models and judgments made. So while ECL should be forward looking, there remains scope for a recent benign period during which credit losses have been low to influence future expectations, and so ECL starts to resemble the incurred loss model in practice. Alternatively, as the development of the models and judgements was influenced by prudential regulatory models, and regulators rightly take a close interest in a bank’s provisioning approaches, the ‘expected’ calculation might become biased on side of prudence and so have a tendency to over-estimate actual credit losses.

With mild economic cycles, procyclical effects might not matter – it took a significant financial crisis in 2007/09 to generate the momentum to replace the IAS 39 incurred loss model with an ECL model (although it should be recognised that an incurred loss model does have certain favourable characteristics, such as greater objectivity and certainty as there is no need to make subjective judgments about the future, it minimises the risk of manipulation, and the purpose of accounting is not to mitigate procyclicality).

Summary

Whether IFRS 9 has a ‘problematic’ procyclical element perhaps depends upon the extent to which there might be a significant over estimation of the ECL arising from the three features noted above. In practice the application of the models and judgements might be key, as there seems considerable scope here to affect the calculation – both to amplify and dampen procyclicality.

It is perhaps too early to assess whether there is significant procyclicality issue in IFRS 9, as the standard has yet to experience a full economic cycle, including a conventional economic stress. While COVID-19 led to a significant over estimation of credit losses, it was unique and unprecedented. We don’t wish it, but a nice conventional recession might be useful in this regard.