In our second post 'building blocks of Impairment Modeling', we had highlighted that IFRS 9 uses a 'three stage model' for measurement of ECL, and one of the major challenges of implementing this model was tracking and determining whether there has been a significant increase in risk of a credit exposure since origination. This blog post delves into the intricacies related to the three stage model, and some nuances that need to be considered for a bank looking to implement IFRS 9.

The intention behind the stage assessment is primarily two-fold:

• To prevent front loading of income for those accounts which have deteriorated in quality since inception, as the interest rate may no longer appropriately cover the credit risk premium
• To recognize and provide for potential losses at an early stage, instead of waiting till the accounts become 90 days past due

It is important to note that the criteria for stage assessment is 'significant increase in credit risk' and the interpretation for this gives rise to some situations that goes against the grain of how bankers have traditionally considered asset quality. Two assets rated the same may no longer fall in the same stage, because it is not just the absolute rating grades that matter but the change in rating grades (or deterioration in credit quality) from the date of origination of the assets. In effect, IFRS 9's three stage model is not an absolute representation of portfolio quality as on a reporting date but a relative reflection of portfolio quality deterioration since inception.

As credit risk managers are beginning to wrap their heads around this concept, another complication arises. The comparison that needs to be done is not just between the rating grades and their associated PDs (of 12 month horizon , typically), but also on the difference between Lifetime PDs of the assets. The reason for considering lifetime PD is primarily because, for financial assets having a maturity of more than 12 months, the true risk of default is not being captured by the 12 month PD, especially if one were to consider adverse macroeconomic scenarios. BCBS too has endorsed and emphasized the use of Lifetime PD as one of the key factors for stage assessment.

This brings into focus the need to identify term structure of PDs and myriad other modelling and data issues. Even if banks are able to identify the lifetime PD as on reporting date for an account, the feasibility of measurement of the lifetime PD at the time of origination remains an area of concern. What this effectively means is that while all of the Watchlist accounts may fall into Stage 2, there could be further additions to Stage 2 classification, based on either change in Lifetime PD or on the basis the Credit Deterioration triggers.

Banks have traditionally been relying on the days-past-due based credit monitoring framework. However, the other criteria that also need to be considered for this Stage assessment are the Credit Deterioration triggers. Very similar to the 'Unlikely to pay' scenario in the Basel II definition of default, banks need to build a framework that identifies scenarios that would necessitate movement of assets from stage 1 to stage 2. We anticipate that Early warning frameworks would get a breath of fresh air with this requirement, as banks are likely to leverage their Early warning frameworks for this purpose. The only way to monitor such credit deteriorating triggers is to have a certain degree of automation along the lines of an Early warning framework.

IFRS9 allows banks to choose their own indicators, define their own cut-off levels and policy for stage migration. Eg. a bank may select 10 relevant triggers for a particular portfolio and draft a policy to migrate the facility to next stage only if 3 or 4 indicators get triggered. As a result, banks would need to check the IFRS 9 indicators that are included in their existing credit risk rating models and existing credit monitoring frameworks for all portfolios.

It is common practice to use behavioral scorecards for retail customers but the wholesale models primarily rely on financial statements, which are often outdated and hence are lagging indicators of credit risk. IFRS 9 standards have also set out in paragraphs B5.5.17, sixteen classes of indicators which should be considered by the banks in stage assessment.

For most of the credit quality indicators mentioned above, the key word to be emphasized is 'expected'. Hence, the triggers are not only based on historical information or facts, but also need to account for forward looking scenarios.

IFRS 9 also provides a number of practical expedients, intended to ease the implementation challenges for various banks. The first expedient is use of '30 days past due rebuttable presumption' as a primary indicator for movement of assets from Stage 1 to Stage 2.

The next expedient is 'low credit risk' exposures, providing banks an option for not assessing whether the credit risk has increased significantly since initial recognition. However, attributing low credit risk for financial instrument solely based on the value of underlying collateral is not recommended.

However, as discussed in our previous post (BCBS advocates high quality implementation of IFRS 9 standards), it is evident that BCBS does not promote the use of practical expedients, and expects that their use be limited.

Worldwide, it has been proven that stage assessment requires significant amount of information, data, analysis and use of experienced credit judgment. Therefore, it is advisable for implementation of systems and frameworks (if not established already) to assess whether a particular lending exposure or a group of lending exposures have witnessed significant increase in credit risk.

In the coming week, we will be touching upon the Impairment Modelling and various options available with the Bank for ECL estimations, as not all banks may consider calculation of separate PD and LGD for measuring ECL.