Financial instruments – Impairment
Feedback from Impairment Summit and Financial Instruments Working Group
During August 2011 the IASB and FASB staff and certain Board members held an Impairment Summit with credit risk managers from banks across multiple jurisdictions and a meeting of the IASB's Financial Instruments Working Group (FIWG). The purpose of both meetings was to obtain initial feedback on the Board's direction with the three bucket impairment approach.
The preferred approach of the Impairment Summit would align the impairment buckets with the credit quality of financial assets so that each bucket contains assets of similar credit quality regardless of whether the assets were originated or acquired. The participants of the Impairment Summit felt that requiring initial recognition of all financial assets in bucket one would result in a 'tracking issue' having to track the credit migration throughout the life of the loan. However, those participants who originate financial assets of lower credit quality would only support the above approach if the point of transfer between bucket one and bucket two is low enough so that most assets would be categorised in bucket one upon initial recognition.
The FIWG was conceptually opposed to recognition of day 1 lifetime expected losses for assets originated on market terms and that those loans should start in bucket one and move to lower buckets as they deteriorate in credit quality.
Participants of both the Impairment Summit and the FIWG believed that differentiation between buckets should be based on a principle rather than a brightline. Both groups also preferred a 12-month expected loss allowance rather than a 24-month expected loss allowance for assets in bucket one although the FIWG noted that both approaches are arbitrary. Both groups also supported a single impairment model for types of financial instruments.
Originated/purchased assets of lower credit quality
The Boards then began a discussion on how to treat originated or purchased assets of lower credit quality on initial recognition. The staff highlighted that the Boards had previously tentatively decided that loans acquired at a discount due to expected credit losses would calculate the effective interest rate considering those credit losses. Therefore, those loans would need to be separately considered for integration into the impairment model being developed, but those assets were not the assets of lower credit quality being discussed today. Additionally, the staff highlighted that the following topic would focus on where to develop boundaries for each of the buckets so this conversation would not focus on where the lines between buckets would be drawn.
During the July 2011 joint Board meetings, the Boards had stated their preference for a 'relative' credit risk approach (basing the bucket classification on the deterioration of credit quality) rather than an 'absolute' credit risk approach (basing the bucket classification on the credit risk existing at a point in time). Under the 'relative' approach, all assets would be categorised in bucket one upon origination or purchase and then would migrate to buckets two or three based on subsequent deterioration in credit quality (or moving back from buckets two or three to bucket one based on subsequent improvements in credit quality). Under the 'absolute' approach, assets would be categorised based on the credit quality existing at a point in time so that all assets of similar credit quality would be classified in the same bucket. However, this approach could also result in assets being originally classified directly into bucket two or three based on the credit quality of the asset.
The staff provided the Boards with a summary of the outreach performed with financial institutions in Europe, Asia, Australia, Africa and North America. The feedback noted that to apply the 'relative' credit risk approach, entities using the Basel II – Advanced Internal Ratings Based (A-IRB) approach would have less operational difficulties than other institutions; however, even those entities had concerns over the manual intensive nature of the process necessary to gather the relevant data. For other entities, the operational concerns of the 'relative' credit risk approach included 1) currently not maintaining ratings history for existing assets, 2) systems do not monitor deterioration of credit quality to the level of detail required, and 3) historical loss expectation data is not always available. The staff highlighted that consistent feedback has indicated that current credit risk management systems are not built to track deterioration in credit quality over the life of the asset, rather they are built to manage assets of similar credit qualities as of a point in time.
Based on the feedback received, the staff asked the Boards for further direction in the development of the impairment model. The three alternatives proposed to the Boards were 1) to continue investigating the operational challenges posed by the 'relative' credit risk approach, 2) move to an 'absolute' credit risk approach and accept full day-1 lifetime losses for assets originated/purchased into bucket two or three, or 3) explore ways to address the day-1 lifetime loss issue for assets originated/purchased into bucket two such as alternative measurements for expected losses of assets within bucket two or providing an option of applying the 'relative' credit risk approach.
The IASB Chair started the discussions by saying that he was surprised that financial institutions did not track credit migration of assets but has been convinced based on the consistent feedback the staff has received that the 'relative' approach would not be feasible. Another IASB member echoed the Chair's comments saying he too has been convinced after initially believing institutions would have access to such data, but that the cost to implement the necessary systems would not justify the incremental benefit to financial reporting. Another IASB member agreed and reminded the Boards that they often refer to how the model would be applied by the largest financial institutions but that they are building a model for all entities including non-financial institutions.
However, a few IASB members were less convinced of the feedback provided by the staff. One Board member mentioned that he thought institutions would be willing to implement the systems if the resulting accounting answer was preferable. Another IASB member mentioned that he had sympathy for the operational concerns the institutions had expressed, but they could not have it both ways, either they invest in making the systems enhancements are they accept the day-1 loss recognition for assets originated/purchased into bucket two. One IASB member also expressed his scepticism that the systems issue could not be more easily overcome.
One of the FASB members asked the staff about the scope of the model being developed and its application to all financial assets subject to impairment as the discussions of this iteration of the impairment model have focused strictly on commercial loans but not retail loans or debt securities. The staff responded that the same base model should be able to be applied to all assets but it may require accommodations based on asset type. That FASB member also mentioned that there had been conceptual differences across jurisdictions in the approach to impairment but that the operational issue seemed to be a cross-jurisdictional issue. The staff confirmed that consistent messages have been received across jurisdictions regarding the operationality of the 'relative' credit risk approach but that the conceptual differences across jurisdictions may still exist for the treatment of assets originated/purchased directly into bucket 2 and the ensuing recognition of day-1 lifetime expected losses. The FASB Chair inquired how their constituents are currently complying with their rollforward disclosure requirements as she envisioned it would have similar tracking issues, but the staff mentioned that was looking at changes from period to period rather than lifetime credit migration.
The Boards seemed to generally support moving off the 'relative' credit risk approach towards the 'absolute' credit risk approach. Their discussion then shifted to discussing the issue of day-1 losses for assets originated/purchased directly into bucket two.
One IASB member mentioned the difficulty he had with the day-1 loss notion for assets originated/acquired on market terms and suggested the Boards consider the measurement of impairment for bucket two assets. One IASB member mentioned that it was hard to have the conversation without first having the discussion on where the bucket boundaries would be set, but his feeling the number of assets being directly originated/purchased into bucket two would be fairly small. He supported an approach that would permit an option for tracking assets using the 'relative' approach if entities were inclined to pursue such a methodology. Another IASB member expressed support for the alternative of investigating other solutions such as an option to apply the 'relative' approach. Another IASB expressed support for addressing the day-1 loss issue by looking at the notion of bucket two.
One IASB member raised the question of what would be included in the Boards went with alternative two (accepting day-1 losses for assets originated/purchased into bucket two). He expressed concern over purchased assets in a business combination that would be originally measured at fair value (with an imputed discount for expected credit losses) but then could be placed into bucket two and then have full lifetime expected credit losses recognised again. One of the FASB members mentioned that they may need to look at interaction with business combination accounting as part of this project to allow for recognition of an impairment allowance as part of purchase accounting. The IASB member expressed his view that the day-1 losses did not reflect the underlying economics and supported recognition of impairment losses through the effective yield (eg the original IASB approach). Another IASB member had a view that the Boards should go back and reconsider whether three buckets are necessary and perhaps look at simply revising IAS 39.
The IASB Chair attempted to move the discussion forward by summarising that the Boards seemed to generally support a move to an 'absolute' credit risk approach but that two issues had been raised that needed to be further explored. The first issue relates to the acquisition of loans that are initially measured at fair value and then placed into one of the impairment buckets where an additional allowance is calculated. The second issue relates to 'markets' where the origination of assets of lower credit quality is the norm rather than an exception and the impact of initially categorising those assets directly into bucket two under the 'absolute' approach. The second issue prompted discussion on the term 'market' and whether that referred to a geographic location or a product line. Some Board members felt it was a geographic issue, while others thought it was a business model issue.
The IASB unanimously supported the IASB Chair's summary and a proposal that the staffs investigate further how to address the two issues identified. The FASB also agreed to have the staffs further investigate these issues, but expressed significant reservations in addressing the issue through permitting an option of the 'relative' credit risk approach for assets of lower credit quality or for recognising less than lifetime expected losses for assets in bucket two.
Principle of transfers between Buckets
The Boards then began a discussion on how to develop a principle of when to transfer financial assets between bucket one and bucket two (ie when it becomes necessary to recognise lifetime expected credit losses). The staff highlighted their consideration of three alternatives for the transfer principle: 1) based on the extent of the deterioration in credit quality, 2) based on 'any' deterioration in credit quality, or 3) based on deterioration in credit quality to a particular level. The staff has also considered three alternatives in the development of a transfer principle by utilising 1) external rating definitions and regulatory classifications, 2) credit risk management objectives, and/or 3) indicators.
The staff provided the Boards with the ratings grades and descriptions from two rating agencies. The Boards began a granular discussion of where to draw the line between bucket one and bucket two using the credit rating agency grades and descriptions. Some felt the line should be the investment/non-investment grade differentiator while others had other cut off levels. However, the Boards quickly took a higher level approach discussing how to develop a principle around the rating descriptions. Several Board members felt that by utilising language similar to that used by the descriptions of the credit rating agencies they could develop a workable principle. One IASB member mentioned he wanted to identify bucket two as the point when an entity begins to manage their assets differently.
While the Boards made no formal decisions, they did express support for the staffs to further develop a principle for transfer based on deterioration of credit quality to a particular level by utilising the concepts and definitions of rating classifications along with concepts of regulatory guidance and other credit risk characteristics. The staff will also develop guidance on information to be considered and examples based on real life fact patterns.