In the Spotlight - Changing criteria for a significant increase in credit risk - Banking
At a glance
As time elapses, there is an increasing likelihood that banks will want to update and change their criteria for determining when a significant increase in credit risk (SICR) has occurred, with the consequent impact on which loans move from 12-month expected credit losses (ECL) to lifetime ECL. This could be for a variety of reasons. However, such changes could attract scrutiny from analysts, regulators and other stakeholders, who might be concerned about the risk of ‘moving the goalposts’ to avoid the adverse profit or loss impact of more loans moving to stage 2.
This publication considers the relevant IFRS requirements, drivers for change, factors to consider when making such changes to SICR criteria, and the need for good disclosures and governance. The key messages are:
- IFRS 9 contains no specific guidance on whether or when to revise SICR criteria. However, it requires regular review of the methodologies and assumptions used for estimating ECL. So, banks should have specific procedures for reviewing, monitoring and reassessing the SICR criteria on a regular basis, no different to other components of the ECL process such as probability of default (PD) and loss given default (LGD).
- Determining what is a ‘significant’ increase in credit risk is highly judgemental and is not defined in IFRS 9. In this context, small changes or changes that are not sustained over a period of time would not typically be expected to result in changes to the SICR criteria.
- Changes to SICR criteria are therefore generally not expected to occur frequently, and there should be a clear rationale when they are made. However, they might be more frequent in the period following initial adoption of IFRS 9, as banks gain valuable experience of how the criteria work in practice.
Financial statement disclosures should be clear and transparent about the reasons for any changes made to the SICR criteria during a period, what the changes were, and their impact.
The concept of ‘significant increase in credit risk’ is fundamental to IFRS 9 and banks’ reported profits, given that it determines whether or not lifetime ECL needs to be recognised on loans and other debt financial instruments measured at amortised cost or fair value through other comprehensive income (FVOCI). On applying IFRS 9, banks set criteria, typically on a portfolio basis, for determining whether a financial instrument has had a SICR since initial recognition. Those criteria typically consist of:
- quantitative thresholds (for example, a specified percentage increase in PD);
- qualitative factors (for example, put on a watchlist, or enter forbearance); and
- the 30 days past due (DPD) backstop.
As time elapses, in particular as more information and experience are gathered, banks might update and change their SICR criteria. This could be for a variety of reasons. However, to protect against the risk of inappropriate changes being made and to pre-empt challenge by regulators, analysts and other stakeholders, banks will need to have robust and rational reasons for any change in this highly judgemental area. In particular, banks need processes and controls to mitigate any risk of ‘manipulating’ ECL (for example, by unnecessarily making changes to SICR criteria that reduce loss allowances as economic conditions worsen or using changes to SICR criteria to ‘store up’ loss allowances in better economic times).
This publication considers the relevant IFRS requirements, drivers for change, factors to consider before doing so, and associated disclosures.
What does IFRS say?
When assessing a SICR since initial recognition, IFRS 9 requires an entity to compare the risk of a default occurring at the reporting date with that at the date of initial recognition, considering reasonable and supportable information that is available without undue cost or effort. As time goes on, a bank’s ability to collect reasonable and supportable information might change (for example, more historical data will be built up over time).
As noted in paragraph B5.5.18 of IFRS 9, an entity might use qualitative and non-statistical quantitative information, statistical models or credit-rating processes to assess SICR, depending on the facts and circumstances. The key building blocks of a typical SICR or ‘staging’ assessment by a bank can be represented diagrammatically as follows:
IFRS 9 contains no specific guidance on whether or when to revise SICR criteria. However, paragraph B5.5.52 of IFRS 9 notes that an entity should regularly review the methodology and assumptions used for estimating ECLs to reduce any differences between estimates and actual credit loss experience.
Practically, because SICR is an accounting concept only, it is at greater risk of falling outside the suite of model monitoring that is normally applied to other aspects of the ECL process, such as PDs, LGDs etc, which are more core to day-to-day credit risk management. However, banks should also have specific procedures for reviewing, monitoring and reassessing the SICR criteria, as well as credit losses. Such reviews and monitoring might result in revisions and refinements to methodologies and assumptions – in that respect, the SICR criteria are no different from other aspects of ECL measurement.
ECL is a highly judgemental estimate; so, as a change of a component within that estimate, any change in the SICR criteria is also a change in estimate. In accordance with paragraph 34 of IAS 8, an estimate might need revision if changes occur in the circumstances on which the estimate was based or as a result of new information or more experience. The impact of a change in SICR criteria will therefore be reported in the income statement in the period when the change occurs, with no restatement of previously reported amounts. See FAQ 45.24.6 How do changes in impairment methodology (for example, change in Definition of Default) impact the assessment of whether a significant increase in credit risk has occurred? in the Appendix for further considerations, in particular whether the credit risk assessment at initial recognition should also be adjusted to reflect the changes to methodology and assumptions.
Furthermore, paragraph 35G(c) of IFRS 7 requires disclosure of any changes in the estimation techniques or significant assumptions used to measure ECL, which includes SICR criteria.
Drivers for change
There are three main triggers that could identify the need to amend the SICR criteria:
- Standard ongoing monitoring of the outputs (namely, which financial instruments are in stage 2 under the existing criteria) identifies anomalous outcomes. This is a ‘detective’ control: that is, it will help detect an issue with the existing criteria.
- Changes are made to the inputs (along with their underlying methodologies and assumptions) that are used by the SICR criteria. This could result in ‘preventative’ updates: that is, to ensure that the SICR criteria remain aligned with how the entity manages and measures credit risk.
- Greater experience and new information, which indicates that the existing SICR criteria should be reassessed.
The three main drivers for change noted above are often interrelated.
Factors to consider
Changes made to SICR criteria should be supported and justified. It is therefore important to understand exactly what has changed, and why, to determine whether it might be appropriate to make a change to the SICR criteria.
In addition, factors such as the impact of changing SICR criteria (both at the current reporting date and in the future, at different points in the economic cycle) and the frequency of change should be considered before making any change. There is a balance to be struck between (i) more frequent and smaller refinements to SICR criteria, particularly as more information and experience becomes available, and (ii) the need for consistency and comparability from period to period.
Judging what is a ‘significant’ increase is not defined and is highly judgemental, and there will inevitably be an acceptable range of SICR criteria. In this context, small changes (for example, identified by regular monitoring, or changes that are not sustained over a period of time) would not typically be expected to result in changes to the SICR criteria. Rather, systemic factors over time would indicate a need to change SICR criteria.
Accordingly, good governance and appropriate oversight is needed to ensure that the criteria are changed appropriately and only for valid reasons.
In addition to these general considerations, factors to consider that are relevant to each of the drivers for change are set out below:
- Monitoring and review of outputs.
- Monitoring. It would generally be expected that the monitoring required by IFRS 9 is performed based on at least the criteria originally used to select the SICR criteria. This is consistent with the view of the UK banking regulator (the PRA) that clear validation criteria and thresholds should be set, against which the performance of SICR criteria is regularly monitored1. For example, a bank might have selected its SICR criteria to optimise key performance indicators (KPIs), such as:
- the proportion of assets moving to stage 2 due to a change in PD;
- the proportion of assets moving to stage 2 due solely to backstop or qualitative criteria;
- the proportion of assets in stage 3 that spent less than a small number of months in stage 2;
- the difference in PD for assets in stage 2 compared to that for assets in stage 1; or
- the difference in PD between assets in stage 2 and assets that are between 1 and 30 DPD or on a watchlist.
Ongoing performance would then be expected to be monitored against these same KPIs, unless there was sufficient justification for changing them.
In addition, some portfolios might be more susceptible to changes in SICR criteria and so be subject to closer monitoring. For instance, a bank might more closely monitor i) portfolios with less history and so smaller data sets on which to base the initial selection of SICR criteria, or ii) portfolios initially reliant on
proxies or approximations for the PD at initial recognition on the adoption of IFRS 9, whose behaviour might change as those assets mature and newer assets originated with more accurate initial PDs increasingly dominate the portfolio. As a further example, the PRA cites sensitivity analysis as an integral part of ongoing SICR validation, to identify portfolios that should be subject to closer monitoring.
1See paragraph 61 of https://www.bankofengland.co.uk/prudential-regulation/letter/2019/written-auditor-reporting-update-and-main-thematic-findings.
- Review. A review of actual outcomes against those expected might identify that the SICR criteria, including PD thresholds, are not giving the results expected. There should be a clear escalation process for when monitoring thresholds or KPIs are breached, including a process to determine when and how SICR criteria should be adjusted. However, judgement will often be required, because the SICR criteria have to balance including assets in stage 2 early enough (before they might become credit-impaired) against including too many assets in stage 2, such that they move back and forth between stages 1 and 2 without a significant increase in credit risk.
Such monitoring might identify that the specified increase in PD needs revising, or that the SICR criteria do not take into account all forward-looking information. Any apparent anomalies identified from the monitoring should be analysed, to make sure that only the SICR criteria for the relevant sub-portfolios are changed and to avoid taking too ‘broad brush’ an approach.
- Qualitative criteria. Monitoring should include qualitative as well as quantitative criteria. Such monitoring should consider qualitative indicators that, on initial setting of the SICR criteria, were considered for inclusion but which at that time were not sufficiently discriminating. If that changes and they become relevant, they should be added to the SICR criteria. For example, such a review might identify that some qualitative identifiers (such as pre-watchlist items that are not being proactively managed) that were not previously considered to be key identifiers (, might need to be included in the SICR criteria.
- Changes to the methodology and basis for inputs. This could arise from a variety of reasons, including the review and monitoring process discussed above and from changes in credit risk management practices and the metrics used.
- Change in PD methodology. For example, an entity might change its definition of default to align with regulatory changes. If the KPIs first used to select the quantitative SICR criteria involved default (such as relative increases in PDs), the KPIs should be re-run when the definition of default changes. Hence, changes in PD methodologies could result in a need to recalibrate the relevant SICR criteria to result in broadly the same financial assets being captured as stage 2 as immediately before the change (assuming that no other factors come into play). However, care should be taken, because, if origination PDs are also updated for such a change, little or no change might be necessary to the SICR criteria.
- Changes to credit risk management. Changes in credit risk management could be due to many different reasons, for example:
- changes in credit risk characteristics of portfolios;
- changes to the economic environment;
- new information;
- changes in bank risk appetite.
It is necessary to understand the reason for the changes to credit risk management and the impact that those changes have on the quantitative and qualitative inputs to the SICR assessment. For example, the wholesale loan watchlist process might be updated to focus on higher-risk loans and exclude lower-risk loans that were previously included. If this is due to a deterioration in the economic environment or because the bank’s risk appetite has changed, it might be that more pre-watchlist items should be included in the SICR criteria to maintain a similar SICR threshold as previously, if the bank still judges that to be an appropriate measure of a significant increase in credit risk. That is, if the fundamental view of what is a SICR has not changed, but changes are made to the credit risk management process leveraged by the SICR assessment, corresponding changes might need to be made to the SICR process to compensate.
An inherent risk arises when different departments within a bank are involved – for example, if the credit risk department starts using a new qualitative factor in its day-to-day credit risk management, but the finance department is responsible for the SICR criteria and is not aware of this, so it does not consider whether the new factor needs to be added to the SICR criteria. Continued effective communication between risk and finance will therefore be an important part of ensuring that SICR criteria remain appropriate.
- New information and learning experience. This is particularly relevant as more experience and data are accumulated in the initial years of applying IFRS 9 or following the origination of new types of loan or loans to different types of borrower. For example:
- A bank might compare its SICR thresholds or the outcomes of its SICR assessment with peers and identify significant differences, for example, in the percentage of loans in stage 2. In considering the appropriateness of its own SICR criteria, the bank would need to understand the specific facts and circumstances as to why this might be the case, because there might be valid reasons for differences, such as a different mix of portfolios or different underlying risk profiles. Where the portfolios are similar and have the same underlying risk profiles, a bank might consider this peer comparison as part of its wider determination of whether a change to its SICR criteria is warranted.
- A bank might have built up more historical data with which to better calibrate its SICR criteria than previously. SICR criteria should be appropriate throughout the economic cycle. However, for example, a bank might not initially have had enough historical data for certain portfolios to identify key indicators in an economic downturn, which could necessitate revising SICR criteria as more data becomes available.
- A bank might change the terms of the suite of products that it offers to customers and whether or how it moves customers between those products, depending on their circumstances. As the bank gains experience of the impacts that this has on customer behaviour and risk, it might need to evolve what it considers indicates a SICR.
- A bank could gain new insight into how its SICR criteria behave in stressed economic conditions from the results of regulatory stress-testing exercises incorporating IFRS 9. This might show that SICR criteria do not have the initially intended results in more extreme downside scenarios.
The financial statements should be transparent about any changes made to SICR criteria during the period. Disclosure should include:
- the new criteria used at the reporting date [IFRS 7 para 35G(a)(ii)];
- an explanation of:
- the changes to the estimation techniques or significant assumptions made to determine whether the credit risk of financial assets has increased significantly since initial recognition; and
- the reasons for those changes [IFRS 7 para 35G(c)]; and
- as a change in estimate, the amount of the impact of the change in SICR criteria on profit in the period of change, and also its expected effect on future periods (if practical to estimate), as required by paragraph 39 of IAS 8.
For example, the change in SICR criteria might not have a significant impact in the current economic environment, but it could have in future periods if there is an economic downturn. In practice, it might be impossible to determine the expected effect in future periods. However, there could be a link with any sensitivity disclosures presented (for example, changes in sensitivities of ECL to more adverse macro-economic conditions), in which case a bank could consider explaining this.
Where do I get more details?
For more information, please contact Sandra Thompson (firstname.lastname@example.org) or Mark Randall (email@example.com).
FAQ 45.24.6 – How do changes in impairment methodology (for example, change in Definition of Default) impact the assessment of whether a significant increase in credit risk has occurred?
How do changes in impairment methodology, such as a change in definition of default (DoD), impact the assessment of whether a significant increase in credit risk (SICR) has occurred?
An entity changes the DoD that it uses in estimating IFRS 9 expected credit losses (ECL). This change could occur for a number of different reasons (for example, changes in regulatory requirements or in internal risk management policies and procedures).
The entity uses probability of default (PD), loss given default (LGD) and exposure at default (EAD) as inputs to its model used to estimate ECL. The DoD is one of the key inputs to this model. Moreover, the entity uses an increase in an instrument’s lifetime probability of default as one of the criteria to determine whether an SICR has occurred.
Does the change meet the definition of a change in estimate or change in policy as defined under IAS 8?
How should the entity consider the change in DoD for the purpose of assessing whether an SICR has occurred?
NB This FAQ assumes that the change is not the correction of a prior period error under IAS 8.
Question 1: Change in policy versus change in estimate
A change in ECL resulting from using an updated DoD, which is an input into the ECL estimation methodology, is a change in an estimate under the definition in IAS 8 and not a change in accounting policy.
Paragraph 5 of IAS 8 defines a change in accounting estimate as “an adjustment of the carrying amount of an asset or a liability … that results from new information or new developments and, accordingly, are not corrections of errors”.
In contrast, accounting policies are the specific principles, bases, conventions, rules and practices applied by an entity in preparing and presenting financial statements.
The DoD is an element of the ECL estimation methodology and the updated DoD reflects a change in the circumstances on which the estimate is based. In other words, it is a component of the measurement technique used to calculate ECL. It follows that changes to the ECL estimate that result from revising the DoD are accounted for as changes in accounting estimates under IAS 8. This is similar to revisions resulting from a change in a valuation technique or its application that IFRS 13 requires to be accounted for as a change in accounting estimate. [IFRS 13 para 66].
The effect of a change in an accounting estimate does not result in the restatement of prior year comparative amounts. [IAS 8 para 34]. [IAS 8 para 36].
Question 2: How to consider the change in estimate when assessing whether an SICR has occurred
The assessment of whether an SICR has occurred under paragraph 5.5.9 of IFRS 9 is a relative test that compares the risk of a default occurring as at the reporting date to the risk of a default occurring as at the date of initial recognition of the financial instrument. However, IFRS 9 has no explicit guidance on how to implement such a relative test when the DoD is changed, and in particular when the entity could continue to measure the risk of default at the date of initial recognition using the previous DoD and when it would need to recalculate the risk of default at the date of initial recognition using the changed DoD. Therefore, judgement will be required, considering the specific facts and circumstances and bearing in mind the overall objective of the SICR assessment as a relative test.
When making this judgement, relevant factors to consider include the following:
- Impact on the impairment methodology
Some changes to the impairment methodology will be more pervasive than others. The greater the impact that a specific change has on the overall methodology, the more likely it might have a material impact on the outcome of the SICR assessment without adjusting the origination PD on a ‘like for like’ basis.
- Impact on current and future reporting periods
An entity should consider the impact of the approach taken on the stage allocation of financial assets at both the date of implementing the change in DoD and the potential impact on future reporting periods, taking into consideration the requirement of paragraph 5.5.17 of IFRS 9 that ECL should be measured in a way that is unbiased. In particular, an entity should consider the risk that, if the staging assessment does not compare ‘like for like’ information, an instrument could change staging without there being a significant increase (or decrease, in the case of a change from stage 2 to stage 1) in the borrower’s default risk since initial recognition. Depending on the remaining expected life of the financial assets, there could be different periods of time over which the selected approach could have an impact on the financial statements, which should all be taken into consideration.
Paragraph 5.5.9 of IFRS 9 states that an entity should consider reasonable and supportable information, that is available without undue cost or effort, that is indicative of significant increases in credit risk since initial recognition. Therefore, if determining the origination lifetime PDs using the new modelling methodology would require undue cost or effort, this would not be required. For example, this might be the case for origination PDs determined on transition to IFRS 9 that, as permitted by paragraph B7.2.2, only approximated the credit risk at initial recognition. However, depending on the impact of the change, an entity should nonetheless explore other simpler approaches that would not require undue cost and effort (for example, by adjusting the SICR threshold when performing the relative test).
If using the new DoD at initial recognition would require information that was unavailable at the time of initial recognition, an entity should consider the principle set out in IAS 8 regarding not using hindsight. Judgement will be needed to assess the impact on it of any hindsight, when considering whether this precludes a recalculation of origination lifetime PDs.