Financial Instruments: Impairment
The Boards continued their discussions on development of the three-bucket expected credit loss impairment model.
Clarification of the measurement objective
The Boards revisited prior discussions on the measurement objective on an expected loss model as the staffs have received questions and concerns from constituents. Specifically, the Boards discussed the objectives of expected credit losses and the bucket one impairment allowance.
Objective of expected credit losses
The Boards have previously tentatively decided that:
- The measurement of expected losses should reflect shortfalls in cash flows (both principal and interest) on a discounted basis;
- An entity should use all reasonable and supportable information (historical, current and forecasts) to estimate expected losses;
- Expected losses should be estimated with the objective of an expected value which identifies possible outcomes (or a representative sample of the possible outcomes), estimates the likelihood of each outcome, and calculates a probability-weighted average. The Boards acknowledged that other appropriate methods could be used as a reasonable way to achieve the expected value objective. An example of a suitable method would be a loss rate method and the use of probabilities of default, loss given default and exposure at default data. In performing this calculation, an entity must not ignore observations and possibilities that are known.
However, constituents continue to raise concerns over the use of the term ‘expected value’ feeling it requires the use of complex statistical approaches. The staffs believe that the Boards intention with the use of the term ‘expected value’ could be clarified by saying “an estimate of expected credit losses shall reflect the following: 1) all reasonable and supportable information considered relevant in making the forward-looking estimate, 2) a range of possible outcomes that considers the likelihood and reasonableness of those outcomes (that is, it is not merely an estimate of the ‘most likely outcome’), and 3) the time value of money.”
One IASB member said he supported the staff’s clarification but questioned whether that was a measurement objective or whether it was simply guidance on how to achieve the objective. Another IASB member expressed some concern with the description of ‘all reasonable and supportable information considered relevant’ saying he felt they were trying to over-describe similar concepts which can cause confusion when translating the standard to other languages. Another IASB member agreed with this concern. Another IASB member questioned if language in the fair value standard could be leveraged to describe the information to be considered. The staff noted that similar-type language is already included in IAS 36.
The Boards tentatively decided that the measurement objective for expected credit losses should reflect all reasonable and supportable information considered relevant in making the forward-looking estimate, a range of possible outcomes that considers the likelihood and reasonableness of those outcomes (that is, it is not merely an estimate of the ‘most likely outcome’), and the time value of money.
Objective of the bucket one impairment allowance
During the December 2011 joint Board meetings, the Boards tentatively decided that the objective and measurement in bucket one would be to capture twelve months’ expected losses. Those expected losses are not only cash shortfalls over the next twelve months but also the lifetime expected losses on the portion of financial assets on which a loss event is expected over the next twelve months. However, since the Boards made that decision, the staffs have received numerous questions on both the objective of the bucket one allowance and how it should be calculated.
As a result, the staffs brought to the Boards three potential approaches on how to articulate the bucket one measurement objective.
The first approach would describe the objective as “expected losses for the portion of financial assets on which a loss event is expected over the next twelve months”. This approach would first identify the portion of the portfolio upon which a ‘loss event’ is expected to occur in the next twelve months and then measuring expected credit losses that will be ultimately realised as a result of those loss events occurring. However, under this approach a ‘loss event’ would not be defined so entities may utilise events such as transferring from bucket one, payment default, reaching a certain number of days past due, or default as defined by a regulatory framework.
The second approach simply describes the objective as “twelve months’ expected credit losses”. This approach would include application guidance illustrating examples of acceptable techniques for measuring twelve months’ expected credit losses such as a 12-month probability of default approach or an annual loss rate approach.
The third approach describes the objective as “expected losses for those financial assets on which a loss event is expected in the next twelve months”. This approach would focus on measuring all cash shortfalls expected over the full lifetime that are associated with the probability of a loss event in the next twelve months. This approach would acknowledge that various approaches could be used to estimate the bucket one allowance so long as the approach is consistent with the objective.
One of the IASB members stated that it is fundamental that the standard capture a specific comment from the agenda paper, that being “that losses being measured are not only the cash shortfalls over the next twelve months but also the lifetime expected losses on the portion of financial assets on which a loss event is expected over the next twelve months”. He felt that any of the three approaches when supplemented with this statement would work but stated a preference for the third approach. Another IASB member agreed and noted that the concept of twelve months of expected losses confuses people and it is imperative that the word lifetime be included in the description.
One FASB member noted that both the first and third approaches include the word ‘loss event’ without defining the term and had concerns that this would raise additional issues of what constitutes a loss event but also agreed with the inclusion of the above comment in the guidance. However, another FASB member noted that the term ‘loss event’ is included in that comment above and therefore the Boards should address what a loss event is suggesting ‘credit deterioration’.
Another FASB member said the Boards should not be mandating whether a one or two step approach is required in measuring these expected losses as that would be over-engineering the process. He also noted that it was important to clarify that it is a lifetime consideration and to permit flexibility as to how entities define a loss event.
An IASB member questioned how consistent a loss rate approach was with the bucket one objective. The staff responded that a probability of default (PD) and a loss given default (LGD) approach and a loss rate essentially provided the same results, it was just that a PD and LGD approach consciously had two separate components while the loss rate approach was a single calculation. Constituents from the US in particular had questioned whether a loss rate approach was consistent with the Boards tentative decision for the bucket one impairment allowance. The staff noted that the third alternative includes a statement that one doesn’t need an explicit twelve month PD portion of the calculation in order to acknowledge that it can be a one step process.
An IASB member questioned the estimate of lifetime losses for loss events occurring in the next twelve months. The staff responded that they would expect granular estimates over the short term and broader estimates over the longer term.
The Boards tentatively decided that the bucket one measurement objective is expected losses for those financial assets on which a loss event is expected in the next twelve months. The standard will clarify that expected losses are all cash shortfalls expected over the lifetime that are associated with the likelihood of a loss event in the next twelve months but that detailed estimates projecting longer term periods would not be required in estimating lifetime expected losses. The guidance will also acknowledge that various approaches can be used to estimate the bucket one allowance including approaches that do not include an explicit twelve month probability of a loss event as an input.
Application of the expected loss model to trade receivables
During the February 2012 joint Board meetings, the Boards discussed application of the three bucket impairment model to trade receivables. The Boards asked the staff to further evaluate whether an expected loss model would be operational for trade receivables without a significant financing component. During this meeting, the Boards agreed that regardless of whether an expected or incurred loss model is used for trade receivables, that those receivables would be initially be recognised in buckets two or three upon initial recognition.
The staffs conducted additional outreach and noted that many entities reporting under both IFRS and US GAAP currently use a provision matrix to estimate their incurred credit losses on portfolios of trade receivables. The provision matrix is based on incurred loss triggers and therefore trade receivables do not have an identified incurred loss until they become past due. However, for specifically identified receivables, a separate estimation process is used, typically because entities have better information for these receivables.
The Boards considered whether to apply the expected loss approach to trade receivables but permit a practical expedient such that a provision matrix can be used or whether to retain an incurred loss impairment model for trade receivables.
One IASB member mentioned that he previously had concerns with applying an expected loss model to trade receivables but has changed his mind based on the analysis provided by the staffs. He also noted his hesitation for having two separate impairment models. However he cautioned that the Boards needed to avoid unnecessary confusion and clarify that the intention is not to change current practice.
However another IASB member expressed concern with applying an expected loss model to trade receivables noting that receivables were not a problem during the financial crises and were not the intention of the impairment project. He suggested staying with the incurred loss model because changing the guidance will inevitably lead to complexity and confusion. The first IASB member responded that the current model is called an incurred loss model but in reality it is an expected loss model because of its reliance on incurred but not reported losses.
The Boards tentatively decided that trade receivables without a significant financing component would apply an expected loss model; however the standard will provide a practical expedient that will permit entities to utilise a provision matrix in estimating expected losses.
The staffs then discussed the remaining steps of the project noting the Boards still need to discuss off-balance sheet items such as loan commitments, transition provisions, application to debt securities (and the interaction with the classification and measurement project) and any other follow up items.