PFRS 9 expected credit losses: How can banks apply pre-pandemic models?

Redgienald G. Radam

The pandemic has created disruptions affecting industries at a global scale. As an offshoot, we have seen unprecedented levels of government relief measures to help curb the pandemic’s economic impact. The pandemic and the subsequent government actions to aid borrowers on their loan payments, have, in turn, affected how banks apply their Expected Credit Loss (ECL) models under PFRS 9 Financial Instruments, which had been developed prior to the pandemic.

The economic disturbances have led to liquidity issues for many entities and individuals, putting into question the credit quality of the financial assets or receivables currently held by banks. Yet events continue to unfold, which means that we have not seen the full extent of COVID-19 impact. This makes the measurement of ECL more challenging as PFRS 9 requires that the calculated ECL capture expectations of future economic conditions that will affect borrowers’ ability to pay.

Under PFRS 9, ECL is a probability-weighted amount determined by a range of possible outcomes (scenarios) and requires the incorporation of reasonable and supportable information about past events, current conditions and forecasts of future economic conditions (i.e., forward-looking information) that are available at the reporting date.

Relating these to the current crisis, banks should be making assumptions in their ECL calculation about the extent of the pandemic’s impact on the collectability of financial assets. As such, it is clear that banks will need to update the ECL models or assumptions previously applied to their reporting as of Dec. 31, 2019 reporting because those would not have foreseen the economic impact of the pandemic.


Based on the EY survey results on COVID-19 benchmarking undertaken in March 2020 across selected global banks, we share some insights on how the impact of the COVID-19 pandemic was considered in the surveyed banks’ ECL models.

While there are variations in the approach to quantify the impact of COVID-19, a majority used portfolio level overlays. This involves incorporating the effect of the pandemic to the forward-looking adjustment to be applied to a group of borrowers with similar credit risk characteristics.

A majority of the banks surveyed also defined specific COVID-19 scenarios for their ECL models. Many have revised the probability weightings applied to the economic scenarios used in the ECL measurement.

Some banks have captured staging movements (i.e., assessment of significant increase in credit risk) through management overlays while others are taking a “top down” approach by migrating part or all of the most impacted portfolios from Stage 1 (requiring 12-month ECL) to Stage 2 (requiring Lifetime ECL). Banks are also evaluating enhancements to credit risk monitoring measures (e.g., forbearance, watchlist, etc.) in response to recent regulatory requests for these data.

For the surveyed banks, these approaches may just be short-term solutions for their interim financial reporting. Locally, we note that some banks have also applied the same or similar approaches for their own interim financials.

Cognizant that the massive and lingering effects of the pandemic will continue to impact the measurement of ECL moving forward, it is imperative that banks develop a comprehensive response to the additional complexities to proactively prepare for year-end reporting and beyond. The approach to this response must be agile considering the time left from now until the year-end. It is vital for banks to have risk modelling capabilities to address the limitations of the short-term solutions adopted in the interim while understanding the impact of the model changes under different assumptions in the long term.


As banks prepare for the year-end and beyond, they will need to consider some urgent priorities in the immediate term when updating their ECL models, including:

Incorporating the impact of government relief measures into the ECL calculation The impact of the relief measures imposed by the government, such as payment holidays or moratoriums, must be assessed to determine how they affect the measurement of ECL. The assessment should consider whether these measures address short-term liquidity issues rather than signal a significant increase in the credit risk of borrowers.

Determining reasonable and supportable macro-economic scenarios These should include the integration of possible pandemic or crisis scenarios (including government relief measures over time) not envisioned previously and how these could have altered the other economic scenarios and their related probability weightings in the ECL measurement. In coming up with the scenarios, banks should also consider the expected duration of the pandemic and the recovery period of the economy. Due care must be exercised so that there is no double counting of the impact of the assumptions on the scenarios and on the other inputs to the ECL measurement.

Consideration of management overlays As there is no consensus on how to forecast future conditions, banks may have to rely on overlays and on their own expert judgment. They will also need to determine the reasonableness of these overlays.

Maintaining strong governance over the ECL process The operational impact of the changes to the ECL models on data, systems and controls must be considered. Increased governance around the significant judgments (including overlays) and assumptions, modelling changes and other changes to the PFRS 9 processes and controls must be in place to ensure the reasonableness of the ECL measurement. Also, they will need to consider the disruptions in operational processes within the bank that may lead to other constraints in running the calculation of ECL.

Disclosures Given the high level of uncertainty and the inherent sensitivity of estimates, it is critical for banks to be transparent in the disclosures of the key assumptions used and judgments made to update the ECL models.

However, as banks work on addressing the urgent priorities above, they must also look at long-term considerations in order to have more robust PFRS 9 ECL models. In addition to being able to model and simulate possible pandemic or crisis-specific data (e.g., default and recovery data) once historical data regarding the pandemic becomes more available, they should also consider approaches to credit management. There must be stronger integration between credit risk management (risk appetite framework, credit approvals, limits, etc.) and the PFRS 9 ECL triggers given current challenges and learnings. The current credit risk assessment process should also be strengthened by incorporating robust sensitivity analysis and stress testing in light of the use of significant judgments and management overlays in the ECL measurement.


While current conditions are indeed unprecedented, banks should see this not just as a challenge in measuring ECL, but as a unique opportunity to advance their credit risk modelling capabilities. Doing so may help them become more resilient and even gain a competitive advantage in these unpredictable and uncertain times.

This article is for general information only and is not a substitute for professional advice where the facts and circumstances warrant. The views and opinions expressed above are those of the author and do not necessarily represent the views of SGV & Co.

Redgienald G. Radam is a Partner from the Financial Services Organization service line of SGV & Co.

Leading the way in business

Other SGV News and Publications