IFRS 9 and bank credit loss explained

Richard Crecel and Daniela Thakkar: banks must act as credit loss remains divergent under IFRS 9.

By Richard Crecel, Executive Director and Daniela Thakkar (pictured), Methodology & Membership Executive, Global Credit Data

As banks’ first financial statements under the new IFRS 9 reporting standard are released, a recent survey by Global Credit Data shows that their Expected Credit Loss (ECL) estimates under the new standard vary wildly – even when assessing identical portfolios of assets. The complex framework will require further benchmarking on a larger scale, including developing a range of standard methodologies and reference points. 

The study, which takes in submissions from 26 banks around the world, including South African banks Investec, ABSA, FirstRand and Standard Bank, shows that banks’ estimates vary by a factor of at least four – and often significantly more. For instance, looking at a loan to a single, hypothetical borrower, assuming the same macroeconomic scenario and forecasts, banks’ estimated expected credit losses ranged from almost nothing to around 55 basis points – a staggering difference. Outliers aside, the majority of the participating banks still estimate an expected loss value somewhere between five and 25 basis points. 

Figure 1: Range of banks’ 12-month Expected Credit Loss under IFRS 9 for a single, hypothetical corporate borrower under a common macroeconomic scenario

Since banks’ credit loss estimates directly feed into their calculations as to how much financing they can offer and at what rate, there is also a significant knock-on effect for corporates, who might see a greater spread in the rates available to them while banks adjust to the new standard.

While some degree of variability in these estimates is inevitable – indeed, it is to be encouraged as a necessary corollary of banks’ different perspectives and business models – the current disparity is significant. Banks will need to bring their estimates into closer harmony or risk seeing stricter interpretations imposed.

Shifting away from “too little, too late”
How has this disparity come about? In response to the fallout from the financial crisis in 2008, the International Accounting Standards Board (IASB) recognised the need for a forward-looking approach to account for credit losses.

This led to the IFRS 9 financial reporting standard, a principles-based approach that requires banks to assume and account for a degree of credit loss for all their assets, estimating losses on each asset for both the coming year and the lifetime of the asset. The previous standard, IAS 39, which only required banks to book a provision in the case of a negative credit event, was widely criticised as “too little, too late”, delaying the recognition of potential issues while favouring forbearance. 

IFRS 9 came into force on 1 January 2018, marking one of the biggest ever shifts in accounting practices. 

Understanding the impact
As the first IFRS 9 statements are released, banks, investors, auditors, regulators and other financial industry participants are attempting to understand the variability of these new loss projections, provision charges and ECL estimates. Given this remains a relatively new framework, there is a need to understand and fine-tune the standard and educate financial institutions on its implementation and impacts.

With a view to gauging how banks are adapting to the standard, Global Credit Data’s benchmarking survey asked participating banks to provide ECL estimates for a representative reference portfolio of assets under IFRS 9. 

As a new, complex and principles-based standard, the IFRS 9 framework inherently allows for variability in ECL estimates. However, data from the Global Credit Data study suggests, variability is higher than foreseen across all asset classes. Crucially, while different macroeconomic forecasts are a predictable source of variability, this variability is also driven to a significant extent by differences in model assumptions, data sources, techniques and processes among banks.

The figures for South African borrowers bear out this trend. When calculating a 12-month ECL for a defined large South African corporate borrower under a common scenario, ECL estimates varied from almost 0 to around 60 basis points. While the majority of estimates are concentrated at the lower end (circa 0-20 basis points), a handful of outliers push the overall variability to higher extremes. 

These 12-month estimates are of particular interest to corporates, as these are the figures calculated first and they, therefore, have a tangible impact on banks’ finance offering – determining how much money needs to be set aside, how much the banks are prepared to lend and at what rate. 

The path ahead
While this data is useful in helping identify the issues, more is needed to accurately pinpoint how much variability should be considered too much. As such, we would encourage all banks – both in South Africa and across the wider African continent – to join ongoing benchmarking efforts that will help enrich their understanding of IFRS 9 and help advocate for appropriate regulatory treatment.

As financial institutions develop more precise methods to improve future credit loss estimates, we can expect regulators and auditors to closely focus on any differences and push for greater consistency. Dialogue with regulators is therefore key if the industry is to avoid overly stringent and restrictive standards on credit modelling and loss provisions.

There remains a lot of work to be done to properly understand the impact of IFRS 9, and more data is required to assess its intricacies. A one-size-fits-all approach would be counterproductive – potentially placing certain business models at a disadvantage. If the industry can consistently benchmark ECL calculations on a large scale, however, they can bring their estimates into alignment organically – creating certainty and clarity for regulators, banks and the corporates they finance.