1. At the meeting of the Transition Resource Group for Credit Losses (TRG) on June 12, 2017, TRG members discussed two views about how to determine the “estimated life” of a credit card receivable ¹ for purposes of assessing credit losses under the guidance in Accounting Standards Update No. 2016- 13, Financial Instruments-Credit Losses (Topic 326): Measurement of Credit Losses on Financial Instruments.

2. The purpose of this memo is to provide the staff’s analysis and recommendations regarding the following separate question that was raised at the TRG meeting:

3. Estimating the remaining life of a credit card receivable balance depends on estimating the amount and timing of the payments expected to be collected on the receivable balance. When estimating the expected payments over the remaining life of a closed-end loan receivable, such as a mortgage loan or a student loan, there is no question that all remaining principal payments from the borrower are associated with the measurement date balance of the closed-end loan. A similar evaluation for an actively used credit card account with a revolving balance is significantly more complex because in addition to future expected payments there also will be new borrowing activity over the remaining life of the measurement date balance.

4. The original TRG submission on this topic dealt only with the question of payment allocation and assumed that all expected payments would be considered under either view. In the June 12, 2017 meeting, two views were presented:

5. TRG members generally agreed that entities could either consider future credit card receivable balances or not consider future credit card receivable balances when determining how to allocate future payments for estimating the life of the credit card receivable balance so long as losses were not recorded for unconditionally cancellable amounts. View 1 and View 2 would result in different outcomes when future credit card receivable balances are expected to have components that carry higher interest rates than the components of the measurement date receivable (in which case, if future credit card receivable balances are considered, payments would be allocated to the highest interest rate portion of the balance first consistent with the CARD Act, even if it is expected to occur in the future). In other cases in which the different components of the balance carry the same interest rate, the views would result in similar outcomes (generally, a FIFO methodology).

6. During the June 12, 2017 TRG meeting, a separate question was raised about how to determine estimated expected future payments; specifically, whether the amount of expected future payments estimated under one or both views should be either of the following:

7. This question arose because some stakeholders believe that it would be inappropriate to take all cash collections received on the account in the future and apply them to the measurement date balance to determine the life of a credit card receivable because some of those payments relate to future draws rather than the measurement date balance. For example, assume the measurement date balance of a credit card receivable is $100 and the entity estimates credit losses based on a model that indicates the next two monthly payments are expected to be $10 per month. In the third month, the model indicates an expected drawdown of $200 and from that point forward payments are expected to be $30 per month. These stakeholders believe that in this case a portion of the higher payment amounts beginning in month three are attributable to the draw activity in month three; therefore, consistent with View B, only a portion of the $30 payments should be considered when estimating credit losses on the reporting date balance.

8. Proponents of View B believe this view would not necessarily need to be applied to an entity’s entire credit card portfolio. Rather, the entity would need to consider expected payment activity for various portfolio segments and determine if inconsistency in expected payment levels may affect any specific segments. If so, the entity may need to apply View B when determining the allowance for credit losses related to the affected segments.

9. Other stakeholders who support View A note that a credit card business typically is managed by maintaining long-standing relationships with cardholders where both receivable balances and payment amounts fluctuate over time, potentially independent of one another. Card issuers forecast future drawdown and payment activity for a variety of purposes (pricing, budgeting, stress testing, etc.), and disaggregating expected payment amounts is not standard practice for any of these other forecasting processes. These stakeholders explained that this is because disaggregating expected payments would require evaluating the measurement date receivable balance as if it were a closed-end loan, which is inconsistent with how the business is managed or how credit risk is assessed for credit card receivables. Rather, card issuers view the measurement date receivable balance as one data point in a rolling relationship with the borrower for which expected draws and expected payments potentially are independent of one another and are related to a variety of factors.

10. Additionally, proponents of View A believe that View B should be prohibited because it may result in a portion of the allowance for credit losses being attributable to unconditionally cancellable credit exposures which would be inconsistent with 326-20-30-11 ³ These stakeholders only consider an expected credit loss to have occurred when the measurement date receivable balance exceeds the subsequent payments expected to be collected from the borrower (consistent with their view of a “rolling relationship” noted in paragraph 9). They note that the allocation of expected payments in View B could result in an allowance for credit losses being recorded before a shortfall in expected payments exists, which in their view would represent an allowance for credit losses on an unconditionally cancellable credit exposure.

11. Regarding the point raised in paragraph 10 on whether View B should be prohibited, the staff acknowledges that under View B forecasted activity on unconditionally cancellable credit exposures may affect the calculation of expected credit losses. However, the staff views this as an indirect effect on the calculation of expected losses for the reporting date receivable balance because the amount of expected payments is only one factor among many that are used to estimate the life of the receivable, which is a consideration in itself when measuring expected credit losses. Additionally, the staff observes that under either view the measurement of expected credit losses is determined in terms of the measurement date receivable balance, not the measurement date balance plus some amount of unconditionally cancellable credit exposure. The staff believes there is a difference between an estimate of credit losses on an existing receivable for which the future payments are influenced by forecasted borrowing activity and a direct estimate of credit losses made on unconditionally cancellable credit exposures that are not recognized at the reporting date. The staff believes only the latter is explicitly prohibited by the guidance in paragraph 326-20-30-11. Because View A and View B only potentially result in the former, the staff believes neither view is prohibited by the guidance.

12. The staff notes that View A is aligned with how credit card issuers view credit card receivables from a business and credit risk management perspective. As noted in paragraph 10, some credit card issuers only consider an expected credit loss to have occurred when the measurement date receivable balance exceeds the subsequent payments expected to be collected from the borrower. The staff notes that this view acknowledges the long-term nature of these revolving credit lines and the fact that the measurement date balance represents cumulative borrowing and payment activity from many periods, which cannot be disaggregated and assigned to specific periods with an appropriate level of precision. Therefore, the staff understands that disaggregating expected future payments and assigning them between the measurement date balance and expected future draws would be inconsistent with current business practices.

13. Additionally, credit card issuers do not currently split expected future payments in their models, so applying View B would represent new modeling that would be unique to estimating credit losses. ⁴ In addition to building the model capabilities, entities potentially would need to develop new assumptions and data sources around payment determination to apply View B. The staff understands that this would represent significant incremental implementation costs that may not be justified because while it may result in a more precise estimated life (which may or may not translate to a more precise measurement of expected credit losses), various methods are allowed by the new guidance and a range of outcomes is acceptable. ⁵

14. However, the staff believes there is conceptual merit to View B for certain types of credit card receivables for which there is inconsistency in expected payment levels. In these cases, higher expected payment amounts may be a consequence of an entity forecasting an increase in borrowing levels. Therefore, an entity that determines the estimated life of a receivable based on the full future payment amounts would reach a shorter life of the measurement date receivable due to forecasted activity from its model. The staff notes that a shorter estimated life may lead to a delay in recognition of credit losses, and in that case, the basis for the delay would be an entity’s own estimate of future borrowing activity. On the other hand, View B would consider only the portion of the expected payment that is related to the measurement date receivable balance, which may result in a more precise estimated life. ⁶

15. One of the bank regulatory agencies proposed that a possible way to operationalize View B may be to use a “pay rate” estimate that could be incorporated into models. Under this method, an entity would consider required minimum payments, historical data on payments above the required minimum, and qualitative adjustments to incorporate current conditions and forecasts of future conditions to determine a normalized expected payment based on the measurement date receivable balance. The entity would then attribute any expected payment amount exceeding the normalized expected payment to future draw activity when determining the estimated life of the measurement date receivable balance. The staff notes that while this may be an operable method to apply View B, it would be a new component of the estimate of credit losses that credit card issuers would need to develop, which may increase implementation costs.

16. The staff also notes that if View B were to be required, it may affect flexibility and scalability for smaller institutions that may not currently have the resources to apply the approach under View B. As noted in paragraph 8, proponents of View B believe that the approach should not be required for all entities or all segments of a portfolio, but that it should be considered in certain circumstances. The staff agrees that View B may be necessary only for segments of a credit card portfolio for which payment activity is expected to be inconsistent and for entities that utilize more sophisticated estimation methods.

17. Overall, the staff believes that, similar to the original question presented to the TRG on allocation, the difference between View A and View B has to do with the precision of the measurement of a receivable’s estimated life, which affects the measurement of expected credit losses. The new guidance on credit losses allows various approaches to be used to determine management’s best estimate that is consistent with a credit risk management perspective. Therefore, depending on an entity’s facts and circumstances, the staff believes either view would be appropriate.

18. A question also was raised on whether the method used to determine payments should influence the method discussed in the TRG meeting on how to allocate payments (or vice versa). The staff views the question of payment determination as a separate issue from payment allocation, but believes that in both cases the issues deal with how precise the estimate needs to be based on an entity’s own facts and circumstances. While it may become a best practice to consider how the methodology decisions on payment allocation and payment determination may interact with one another, the staff does not believe that any combination of the views should be prohibited. There are likely to be portfolios for which neither issue (payment determination or payment allocation) would create a significant difference in the measurement of credit losses ⁷ , and, in these circumstances, it may be appropriate not to consider future receivable balances for either issue. The staff also observes that there may be additional views on either the payment allocation question or payment determination question that may be developed over time, which may also be acceptable under the new guidance.

19. The staff believes either view should be permitted because two fundamental aspects of the new guidance are (a) flexibility in the methods allowed and (b) reporting an estimate of credit losses that is management’s best estimate and that is aligned with a credit risk management perspective. The staff believes that this issue relates to the precision of the estimate of credit losses and does not believe that either of the views are prohibited by the new guidance. Furthermore, the staff views this issue as separate from the allocation issue discussed at the TRG meeting; therefore, any combination of views from that issue and the issue described in this memo would be acceptable.

20. The Board agreed (7-0) with the staff’s recommendation that an entity may determine payment amounts, for the purposes of estimating the life of a credit card receivable, as either all or a portion of expected future payments. This decision effectively permits any combination of payment allocation methodologies discussed during the June 12, 2017 TRG meeting and payment amount determination methods discussed at the October 4, 2017 Board meeting. The Board observed that this decision is consistent with the amendments in Update 2016-13, which allow various approaches to be used to determine management’s best estimate consistent with its credit risk management perspective.

21. The Board noted that decisions about credit card payment determination and allocation should adhere to the rest of the model created by Update 2016-13. Specifically, chosen methodologies should be consistent over time and faithfully estimate expected credit losses for financial assets. The Board also noted that entities are not limited to the payment determination and allocation methodologies discussed in the June 12, 2017 TRG meeting and October 4, 2017 Board meeting if other appropriate means of estimating the expected life of a credit card receivable are available.