The focus on impairment is intensifying as the IASB and FASB unveiled their latest proposals for recognition and measurement of impairment in March. The IASB and FASB have been working on a new model for some time now, addressing criticisms that the current mixed measurement incurred loss model recognised provisions for impairment too little too late.

Having agreed in principle to shift to an expected loss model, pinning down a precise model has been somewhat trickier. The project has been on-going since 2008 with different proposals tabled. The basic challenge each model has tried to address is when and how to recognise impairment losses in profit or loss.

Are we there yet?

The IASB first proposed in its exposure draft issued in November 2009, a model where the provision for expected loan losses would build up overtime by accruing interest income at a rate that includes expected losses (i.e. a lower rate than the contractual rate). This was referred to as an 'integrated' effective interest rate approach as it integrated the recognition of interest income and provisions for credit losses. This approach was criticised by many preparers for being too difficult to implement operationally given the integration of data from credit systems into other systems used for interest income recognition. The model was also criticised by some banking supervisors for not building up provisions soon enough for assets with early loss patterns.

This led to a revised proposal, this time developed jointly with the FASB, based on a 'decoupled' approach, separating interest income recognition from the build-up of provisions for expected losses (i.e. the effective interest rate would remain the same as is currently required by IAS 39). This approach introduced a 'good book'/'bad book' approach, where the bad book represents loans that are managed differently due to credit deterioration which would be fully provided for.

The good book would capture the remaining loans which would have a provision for expected losses built up over time (a 'time-proportional' basis) but with a minimum provision to provide for losses in the 'foreseeable future'. Although this model was designed to fix some of the issues raised with the first proposal, it did not gain widespread acceptance due to new complexities of distinguishing between the good book and bad book and the determination of the provision for the good book.

Since then, the IASB and FASB have been working on a new model. The FASB is ahead in publishing its proposals having issued its exposure draft in December 2012 with an extended comment period to 31 May 2013. The IASB issued its exposure draft on 7 March 2013 with a 120 day comment period, hence there is some overlap of comment periods for constituents to consider both models and make comparisons. The two proposals have significant common ground but also have some important differences. Below is a high-level comparison of the two models.

What does the model apply to?

The impairment model has primarily been designed to apply to loan assets measured at amortised cost (which is the main focus of the analysis below). However, both the FASB's and IASB's classification and measurement models propose a classification of fair value through other comprehensive income (FVTOCI) for eligible loan assets. Despite the carrying value for FVTOCI assets being fair value, both Boards propose applying impairment based on the amortised cost value of the asset, with any difference between amortised cost and fair value residing in equity until disposal. Hence, this model will also be relevant for loan assets measured at FVTOCI.

The model will also apply to impairments of lease receivables and written loan commitments not at fair value through profit or loss. There is a difference in scope of the FASB and IASB models in that the IASB's model would also apply to allowances for losses on written financial guarantee contracts and the FASB's model would also apply to reinsurance receivables.

How many buckets?

The recent discussions about the impairment models have been focused on the number of buckets the model has as the impairment allowance for each bucket is different. The model that the Boards were jointly discussing during 2012 was described as a three-bucket model. Under this model, bucket one would represent loans that have exhibited no, or only insignificant, credit deterioration since initial recognition (excluding purchased distressed debt discussed below). The impairment allowance for this bucket would be those credit losses arising from credit events expected to occur in the next 12 months. Bucket two would generally represent portfolios of loans that have exhibited credit deterioration since initial recognition.

The allowance for these loans would be for lifetime expected credit losses. Bucket three would then capture individually identified loans that have exhibited credit deterioration since initial recognition and would also have an allowance for lifetime expected credit losses. Given that both bucket two and three have an allowance for lifetime expected credit losses, this model is increasingly referred to as a two-bucket model (i.e. 12 month loss bucket and life-time loss bucket).

Given the two different buckets and their different allowance requirements, it is important to be able to distinguish which bucket a loan, or portfolio of loans, sits within. As assets generally start in bucket one, it is the criteria that causes an asset to migrate from bucket one to bucket two which becomes important. Under the two-bucket approach, a transfer from bucket one to bucket two would arise when there has been a "more than insignificant deterioration in credit risk since initial recognition".

It is in part due to this transfer threshold that the FASB's and IASB's proposals have diverged. U.S. constituents' feedback to the FASB has indicated that it would be operationally too difficult to implement a model with such a transfer requirement. Some felt that providing for only 12 month losses for some loans would not lead to a sufficient amount of provisions. As a result, the FASB has proposed a single bucket model where an allowance for lifetime expected losses is established for all loans from initial recognition. The IASB, however, has maintained a two-bucket model in its exposure draft.

Measurement of expected losses

The measurement of expected losses is broadly the same under both Boards' model. Essentially it is a current estimate of expected losses (lifetime or 12 month). There could be different methodologies for achieving this. The IASB's discussions have focused on a probability weighted assessment of expected contractual cash flows not expected to be recovered. Any approach would be based on expected values rather than the most likely outcome and would reflect the time value of money.

Purchased distressed debt

Purchased distressed debt has always been regarded as a different beast and it is no surprise that the proposed accounting for it is different to other loans. Under both Boards' models, purchased distressed debt would initially be recognised at fair value and the net effective interest rate would include credit losses expected as at initial recognition (i.e. a loan loss allowance for expected losses would build up over time). Any subsequent change in lifetime expected losses would result in a catch up adjustment recognised in profit or loss. It should be noted that although the net accounting result is the same, the mechanics by which the result is achieved is different under each model.

Interest

There are some noteworthy differences in interest recognition for certain loans under the two different models. Considering the IASB's model first, interest is recognised by applying the effective interest to the gross carrying amount (i.e. excluding any allowance) unless the loan is regarded as credit impaired, in which case the effective interest rate is applied to the net amount. An asset is regarded as credit impaired when there are incurred credit losses (determined in the same way as under the IAS 39).

The FASB's model also initially recognises interest income on a gross basis. However, if it becomes probable that substantially all principal or substantially all interest will not be received, then interest ceases to be accrued. This non-accrual status for certain impaired loans already exists under U.S. GAAP and is an existing difference to IFRS.

What to expect next

It can be seen from the above that the two Boards have moved significantly in the same direction from an incurred loss model to an expected loss model, yet there are some important differences in the detail. The impact of these differences would vary from loan book to loan book. For example, a loan book consisting of shorter term loans of 12 months or less would have an allowance equivalent to lifetime losses under either the FASB's or IASB's model. However, for international banks with varied lending the difference between applying a one-bucket model versus a two-bucket model could be substantial, hence, many would prefer a converged solution to be reached. As both Boards will be sifting through responses over the same period, depending on views received, there is still some possibility of a single model. Watch this space.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.