Reflections on the development of the FASB’s and IASB’s expected-loss methods of accounting for credit losses^†

† This paper is based on a presentation at an Accounting and Business Research symposium on The Role of Accounting Information in Debt Markets at the European Accounting Association Congress, Valencia (May 2017). The paper is informed by work carried out within a project for which some results are reported in an ICAEW research briefing (O'Hanlon et al. 2018). The authors gratefully acknowledge the financial support of the ICAEW's Charitable Trusts for that project. This paper has benefited from the comments of Kees Camfferman on a preliminary version and from the detailed comments and suggestions of an anonymous reviewer.

Abstract

After the financial and banking crisis of the late 2000s, the FASB and the IASB aimed to develop methods of accounting for credit losses that would give more timely recognition of those losses. The IASB (in 2009) and the FASB (in 2010) each initially issued its own exposure draft proposing separate approaches to achieving this. They then attempted to agree a converged expected-loss-based method for accounting for credit losses, but failed to achieve convergence. They then each issued an accounting standard that included its own expected-loss method, with effective dates of 2018 for the IASB and 2020/21 for the FASB. This paper provides an overview of the development of proposals and standards in relation to accounting for credit losses issued by the standard setters from 2009 to 2016. It then offers reflections on difficulties that the standard setters faced in this area and on problems that might arise after the new standards become effective. It raises the question of whether a route based on ‘expected loss’, which in relation to credit losses is a concept that originally became prominent for the purpose of setting banks’ capital requirements, was helpful to the process of improving the accounting for credit losses.

Keywords:

• financial accounting
• impairment
• expected loss
• incurred loss
• convergence

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

† This paper is based on a presentation at an Accounting and Business Research symposium on The Role of Accounting Information in Debt Markets at the European Accounting Association Congress, Valencia (May 2017). The paper is informed by work carried out within a project for which some results are reported in an ICAEW research briefing (O'Hanlon et al. 2018). The authors gratefully acknowledge the financial support of the ICAEW's Charitable Trusts for that project. This paper has benefited from the comments of Kees Camfferman on a preliminary version and from the detailed comments and suggestions of an anonymous reviewer.

1. See the FASB/IASB press release dated 20 October 2008, available from: http://www.fasb.org/news/nr102008.shtml [Accessed 18 August 2017].

2. For example, Hashim et al. (2016) report that loans typically account for 60% to 70% of banks’ total assets. Ryan (2017) observes that ‘the allowance for loan losses (which reduces loans outstanding) and the provision for loan losses (which reduces net income) are by far the most important and judgmental accrual estimates for most banks’.

3. Throughout this paper, references to a method being perceived as conceptually strong denote that the method is perceived as faithfully representing the economic substance of activities. This is consistent with the way in which the term ‘conceptual’ is used in many of the comment letters examined by O'Hanlon et al. (2018) in their study of responses to the various proposals on accounting for credit losses issued by the IASB and/or the FASB between 2009 and 2013.

4. See transcripts of remarks made by the SEC's Chief Accountant, Michael Sutton, in November 1997 (available from: https://www.sec.gov/news/speech/speecharchive/1997/spch195.txt) and by the SEC’s Chairman, Arthur Levitt, in September 1998 (available from: https://www.sec.gov/news/speech/speecharchive/1998/spch220.txt) [Both accessed 10 April 2018].

5. SAB 102 is available from: https://www.sec.gov/interps/account/sab102.htm [Accessed 10 April 2018].

6. In a discussion of Beck and Narayanmoorthy (2013), Ryan and Keeley (2013) argue that SAB 102 did not express any preference for historical loss information in making allowance estimates, but acknowledge that the SAB requirements for the application of consistency and discipline may have delayed banks’ recognition of losses on heterogeneous loans.

7. Available from: https://www.occ.treas.gov/news-issuances/speeches/2009/pub-speech-2009-16.pdf [Accessed 10 April 2018].

8. Prior to and during the deliberations leading up to the 2003 revision of IAS 39, the treatment of expected losses under the internal-ratings-based (IRB) approach to be introduced in Basel 2 (BCBS 2004b) was being considered by BCBS (BCBS 2001a,b, 2003, 2004a).

9. Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

10. Available from: https://www.aba.com/Advocacy/Issues/Documents/CECL-backgrounder.pdf [Accessed 24 April 2018]. As noted in IAS 39 (IASB 2003, paragraph AG90), IBNR denotes loss events of which a lender is unaware as at a reporting date but that can be assumed to have occurred prior to that reporting date, but not loss events expected to occur after that reporting date.

11. Hodder and Hopkins (2014) refer to the requirement that regulatory accounting principles must be consistent with or no less stringent than GAAP. IASB (2014a, paragraph BC5.116) notes that ‘the interaction between the role of prudential regulation and loss allowances is historically stronger in the U.S.’. A speaker at a European Parliament hearing on IFRS 9 commented that ‘ … the whole history of U.S. GAAP accounting for loan losses is steeped in what the prudential regulators in the U.S. have required […]’ (Mike Ashley, European Parliament, 1 December 2015).

12. A document containing the contribution by this speaker is available from: http://www.europarl.europa.eu/cmsdata/93535/Statement%20Ashley%20EN.pdf [Accessed 18 August 2017].

13. Ng and Roychowdhury (2014) examine the association between loan-loss allowances included in Tier 2 capital and bank failure. Their examination suggests that this element of Tier 2 capital does not behave like an element of capital that acts as a buffer against failure risk.

14. Some banks that used the IRB approach also used the standardised approach for some business units, and their Tier 2 regulatory capital therefore continued to include some general loan-loss allowances.

15. Practice in the U.S. is to treat the whole loss allowance as a general allowance for the purpose of calculating Tier 2 capital under the Basel standardised approach. For reference to this, see OCC (1998, p. 4).

16. U.S. IRB banks are required to satisfy the capital requirements of both the IRB approach and the standardised approach. (For a reference to this, see the Federal Register, Friday October 11, 2013, p. 62,021. This is available from: https://www.gpo.gov/fdsys/pkg/FR-2013-10-11/pdf/FR-2013-10-11.pdf [Accessed 20 April 2018].)

17. Issues related to the trade-off between restricting earnings management and permitting discretion are sometimes considered in relation to a trade-off between ‘relevance’ and ‘reliability’. In light of the fact that ‘reliability’ has now been removed as a qualitative characteristic from the IASB and FASB conceptual frameworks, we do not structure our review around this trade-off. ‘Relevance’ and ‘reliability’ were included as qualitative characteristics in the pre-2010 conceptual frameworks of the IASB and the FASB (FASB 1980, IASB 2001). The FASB and IASB 2010 conceptual frameworks removed the term ‘reliability’ from their qualitative characteristics on the ground that the term was ambiguous in the context of conceptual frameworks (FASB 2010b paragraphs BC3.20-BC3.24, IASB 2010, paragraphs BC3.20-BC3.24). This ambiguity had been noted by Ball (2006, p. 9). It is also noted in the IASB's 2018 Conceptual Framework (IASB 2018, paragraph BC2.29).

18. The directional effects of loan-loss expense on earnings and regulatory capital are not necessarily the same. Loan-loss expense that decreases net-of-tax earnings and decreases shareholders’ equity can increase regulatory capital if gross-of-tax loss allowances are included in regulatory capital.

19. Prior to implementation in the U.S. in 1989 of the Basel 1 accord, loss allowances were included in primary capital. Subsequent to the implementation of Basel 1, which saw the introduction of a split of capital into Tier 1 and Tier 2, general loss allowances up to 1.25% of risk-weighted assets could be included in Tier 2 capital. The Basel 1 accord, issued in 1988, was implemented by the U.S. bank regulatory and supervisory agencies in 1989. (For a reference to this, see the Federal Register, Friday December 7, 2007, p. 69,289, footnote 3. This is available from: https://www.gpo.gov/fdsys/pkg/FR-2007-12-07/pdf/FR-2007-12-07.pdf [Accessed 12 April 2018].)

20. O'Hanlon (2013) finds no evidence that accounting for credit losses by U.K. banks became less timely after the move from the evidence requirements of pre-existing U.K. GAAP to those of IAS 39. However, it should be noted that accounting for credit losses under U.K. GAAP was already in accordance with incurred-loss and that the effect of the adoption of IAS 39 on accounting for credit losses in the U.K. was therefore likely to be less pronounced than in some other countries.

21. The concept of ‘prudence’ featured in the pre-2010 conceptual frameworks of the FASB and the IASB, with the term ‘conservatism’ being used interchangeably with ‘prudence’ in the FASB framework. In both cases, it was emphasised that ‘prudence’ denoted the exercise of caution under conditions of uncertainty and should not lead to the systematic understatement of net asset values. The 2010 FASB and IASB conceptual frameworks, the term ‘prudence’ was removed because of the possibility that it might be incorrectly interpreted as encouraging downward bias in net asset values, which would be inconsistent with neutrality (FASB 2010b, paragraphs BC3.27-BC3.28, IASB 2010, paragraphs BC3.27-BC3.28). In 2018, the IASB issued a revised conceptual framework. This reinstated the concept of ‘prudence’, again stating that the term is intended to denote the exercise of caution under conditions of uncertainty and not the systematic understatement of net asset values (IASB 2018, paragraph 2.16).

22. For fuller discussion of conditional and unconditional conservatism in the general financial-reporting literature, see Watts (2003a), Watts (2003b), Beaver and Ryan (2005) and Ryan (2006).

23. Evidence in Lee and Rose (2010) for U.S. bank holding companies for 2000–2009, from which Ryan (2017) quotes some figures, indicates that the average ratios of loan-loss allowance to equity, equity to total assets and loan-loss allowance to total assets are in the region of 10%, 10% and 1%, respectively. These magnitudes are similar to those reported by Hashim et al. (2016, pp. 255–256) for 2007–2014 for U.S. banks and for banks in developed markets outside the U.S.

24. The early stages of this activity took place against the background of proposals under which most financial instruments would be measured at fair value (IASB 2008, FASB 2010a). However, there was strong opposition to the proposed significant increase in the use of fair value, particularly for loans. Subsequent stages of development of new impairment methods took place in the expectation that a large proportion of financial assets, including loans, would continue to be measured at amortised cost.

25. The European Parliament paper by O'Hanlon et al. (2015) was written before the FASB standard (FASB 2016) was finalised.

26. See, for example, the letter from the Cigna Corporation in respect of IASB (2009): ‘This proposed model that reports level investment returns implies consistent, level investment risk, which does not reflect actual performance or events. Credit risk changes over time for individual instruments and financial statement reporting should not inadvertently mask such changes.’ Comment letters on IASB (2009) are available from: http://archive.ifrs.org/Current-Projects/IASB-Projects/Financial-Instruments-A-Replacement-of-IAS-39-Financial-Instruments-Recognitio/Impairment/ED/Comment-Letters/Pages/Comment-letters.aspx [Accessed 24 April 2018].

27. Some found it surprising that the FASB (2010a) proposals on credit-loss impairment differed so much from the IASB (2009) proposals, particularly in light of the aim at the time that the FASB and the IASB should work together on accounting for financial instruments. See the comment letter from the American Insurance Association (page 1). Comment letters on FASB (2010a) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=1810-100 [Accessed 18 August 2017]. In relation to this point, IASB (2009) had noted that ‘It is not uncommon for the boards to deliberate separately on joint projects and then subsequently to reconcile any differences in their technical decisions’ (IASB 2009, paragraph IN13).

28. The TPA allowance would be equal to all credit losses expected for the remaining portfolio life multiplied by the portfolio's age as a proportion of its expected life.

29. IASB (2013a) noted that the TPA spreading method differs from the IASB (2009) spreading method in that it spreads across time both the recognition of initially-expected losses and the recognition of subsequent changes in expectations about losses: ‘The SD (supplementary document) attempted to reflect the relationship between expected credit losses and interest revenue using the TPA. The TPA reflects this relationship through the allocation of expected credit losses over time, “adjusting” the contractual interest. However, it does this through a short-cut, and therefore the result does not represent the economics as faithfully as the 2009 ED (exposure draft) did. Because the TPA allocates over time both the initial expected credit losses and the subsequent changes in lifetime expected credit losses, the measurement results in an understatement of changes in expected credit losses until the entity recognises lifetime expected credit losses.’ (IASB 2013a, paragraph BC35).

30. O'Hanlon et al. (2018) categorised a total of 183 respondents to FASB/IASB (2011a) as being either non-U.S. respondents that could be assumed to be IASB constituents or potential constituents (111 (61%)) or U.S. respondents that could be assumed to be FASB constituents (72 (39%)). Some respondents (international accounting firms, international associations, international bank-regulatory organisations and other international regulatory organisations) were included in neither of these categories for the purpose of the statistics reported here.

31. See the comment by the Chair of the IASB at a meeting of December 2012, which we quote in our subsection 3.5.

32. After August 2012, the IASB and the FASB continued to work towards convergence in some areas. For example, they issued a joint proposal on Leasing in 2013 (FASB 2013, IASB 2013c), although the FASB and the IASB eventually produced different standards in this area, and they issued a converged standard on Revenue Recognition in 2014 (FASB 2014, IASB 2014c). It therefore appears likely that, although the FASB's disengagement from the three-bucket deliberations may have been related to a broader move away from convergence in general, this in itself would not have stopped the attempts to achieve a converged common solution on credit-loss impairment. It appears likely that the dominant factor was the boards’ inability to agree on the substance of the proposals.

33. Recording available from: http://media.ifrs.org/2011/IASB/July/Impairment_session2.mp3 [Accessed 18 August 2017].

34. Recording available from: http://media.ifrs.org/AP4Impairment18102012.mp3 [Accessed 18 August 2017].

35. See paragraph 32 in agenda paper 5A, available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2012/October/Impairment-1012-05A.pdf [Accessed 18 August 2017].

36. Recording available from: http://media.ifrs.org/2013/IASB/July/Impairment_AR5_PM.mp3 [Accessed 18 August 2017].

37. Because the magnitude of loss allowances is affected by practice with regard to charge-offs as well as by the timeliness with which credit-loss expense is recognised, more timely recognition of credit losses does not necessarily lead to higher loss allowances. See Liu and Ryan (2006) for evidence that charge-offs may be used to obscure the effect on loss allowances of the management of credit-loss expense. For a given charge-off practice and other things equal, more timely recognition of credit losses would give rise to larger allowances.

38. Recording available from: http://media.ifrs.org/2012/IASBMeetings/December/ImpairmentAR5141212.mp3 [Accessed 18 August 2017].

39. Recording available from: http://media.ifrs.org/2012/IASBMeetings/December/ImpairmentAR5141212.mp3 [Accessed 18 August 2017].

40. IASB (2013a) defined 12-month expected credit losses for a loan for which credit risk has not increased significantly since initial recognition to be the product of (i) the gross carrying amount of the loan, (ii) the proportion of the gross carrying amount of the loan expected to be lost if the loan defaults (loss given default) and (iii) the probability of default over the next 12 months (IASB 2013a, paragraphs IE2-IE3).

41. Recording available from: http://media.ifrs.org/2013/IASB/September/Impairment_AR5_Session1.mp3 [Accessed 18 August 2017].

42. The three-stage terminology from IASB (2013a) is not used in the standard itself, but is used in an IASB summary document (IASB 2014b, p. 18).

43. See, for example, paragraph 6 in the document available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2013/July/5D-Impairment.pdf [Accessed 18 August 2017].

44. This point was stressed in communications with stakeholders, including at the FASB roundtable meeting of 4 February 2016 attended by representatives of Community Banks at which the representatives expressed serious concern about the forecasting requirements of CECL.

45. For references to such activities, see:

• IASB March 2010 – available from: http://archive.ifrs.org/Current-Projects/IASB-Projects/Financial-Instruments-A-Replacement-of-IAS-39-Financial-Instruments-Recognitio/Impairment/Meeting-Summaries/Documents/FI_0310b13Aobs.pdf (paragraphs 9 and 10) [Accessed 1 December 2017];

• FASB/IASB April 2011 – available from: http://archive.ifrs.org/Current-Projects/IASB-Projects/Financial-Instruments-A-Replacement-of-IAS-39-Financial-Instruments-Recognitio/Impairment/Meeting-Summaries/Documents/FI0411b04Eobs.pdf (paragraphs 2 and 52–54) [Accessed 1 December 2017];

• IASB October 2012 – available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2012/October/Impairment-1012-05A.pdf (paragraph 32)) [Accessed 1 December 2017];

• IASB July 2013 – available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2013/July/05A-Impairment.pdf (paragraphs 33–37), [Accessed 1 December 2017].

46. Available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2013/July/05A-Impairment.pdf (paragraph 35) [Accessed 1 December 2017].

47. Of the 1,588 comment letters examined by O'Hanlon et al. (2018), 47 (3%) were from financial and banking regulators.

48. Of the total of 510 letters in response to FASB (2012) and IASB (2013a) examined by O'Hanlon et al. (2018), 14 (3%) were from financial and/or banking regulatory bodies.

49. See question 23 in a document published by the American Bankers’ Association, available from: https://www.aba.com/Advocacy/Issues/Documents/CECL-backgrounder.pdf [Accessed 26 April 2018].

50. It should be noted that the situation described here differs from one in which a loan might be made on terms favourable to the borrower such that the fair value of the expected recoverable amount is less than the amount lent. In such a situation, it would be uncontroversial that the loan should be written down to fair value at day 1.

51. For conceptual framework documents, see FASB (2010b) and IASB (2010).

52. See, for example, the letter from the Illinois CPA Society commenting on FASB (2012): ‘We believe the fundamental premise on which the Proposed ASU rests is at odds with the Conceptual Framework. […] By requiring banks to recognize a provision for current expected credit losses for at-market loans at inception the Proposed ASU would require banks to recognize expected credit losses twice – once through the initial pricing of the loan and again through the separate day one loss allowance. This results in financial statements that do not faithfully represent the economic phenomenon of the issuance of an at-market loan, whereby neither party has realized an economic gain or loss.’ Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

53. A discounted-cash-flow (DCF) approach is only one of various methods by which loss allowances may be determined under the FASB's CECL method.

54. See, for example, the letter from Duff and Phelps LLC commenting on FASB (2012): ‘One theoretical flaw in the model is that the discount rate applied to the expected credit losses (effective interest rate) already takes into account the credit risk priced in by the lender at origination and would have the effect of double-counting the initial estimate of credit losses.’ Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

55. See, for example, the letter from UDVA commenting on FASB/IASB (2011a). With respect to the proposal for day-1 recognition of foreseeable-future-period (FFP) losses, this respondent says that this treatment is ‘[…] inconsistent with initial recognition of financial instruments at fair value as the transaction price (fair value in an arm's length transaction) already reflects the expected losses.’ Comment letters on FASB/IASB (2011a) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2011-150 [Accessed 18 August 2017].

56. See, for example, the letter from the Canadian Bankers Association commenting on FASB (2012): ‘We do not believe that recognizing all interest earned over the life of the loan and total lifetime expected credit losses at inception is an appropriate reflection of the economics of the loan transaction.’ Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

57. See, for example, the letter from Cisco Systems, Inc. commenting on FASB (2012): ‘Recognition of ECL (expected credit losses) at inception would contribute to significant income statement volatility and provides an avenue for potential earnings management. For example, under the Proposal, we would record a credit loss at the time of purchase of a financial instrument that is not driven by an economic event. Subsequently, if we sold the same financial instrument in a different reporting period, we may record a gain even though there is not a change in credit risk.’ Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

58. See paragraph B21 in the document available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2013/July/5D-Impairment.pdf [Accessed 18 August 2017].

59. See, for example, the letter from the Federal Financial Institution Regulatory Agencies in respect of FASB (2012): ‘We have observed that credit losses typically emerge relatively early in the life of loans and do not occur ratably over a financial instrument's life.’ Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

60. See, for example, the letter from Carlson Capital LP in respect of FASB (2010a): ‘Our long-held view is that there is an inherent risk of loss from the moment every dollar is lent.’ Comment letters on FASB (2010a) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=1810-100 [Accessed 18 August 2017].

61. This passage is from the response to question 3 in a FASB ‘Frequently Asked Questions’ document available from: http://www.fasb.org/cs/BlobServer?blobkey=id&blobwhere=1175826417092&blobheader=application/pdf&blobcol=urldata&blobtable=MungoBlobs [Accessed 18 August 2017].

62. Available from: https://www.aba.com/Advocacy/LetterstoCongress/Documents/RussellGolden-FASB-011316.pdf [Accessed 26 April 2018].

63. IASB (2014a, paragraph BCE.93 (d)) notes the benefit of the broadened information set in making credit-loss impairment more forward-looking.

64. See paragraph A6 in the document available from: http://archive.ifrs.org/Meetings/MeetingDocs/IASB/2013/July/5D-Impairment.pdf [Accessed 18 August 2017].

65. Comment letters on IASB (2013a) are available from: http://archive.ifrs.org/Current-Projects/IASB-Projects/Financial-Instruments-A-Replacement-of-IAS-39-Financial-Instruments-Recognitio/Impairment/Exposure-Draft-March-2013/Comment-letters/Pages/Comment-letters.aspx [Accessed 18 August 2017].

66. Comment letters on FASB (2012) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2012-260 [Accessed 18 August 2017].

67. Comment letters on FASB/IASB (2011a) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2011-15011-150 [Accessed 18 August 2017].

68. O'Hanlon et al. (2018, p. 23) report that a substantial number of US respondents to the 2012 FASB exposure draft and some non-U.S. respondents to the 2013 IASB exposure draft expressed preference for a modified and less restrictive IL-based approach over an EL-based approach.

69. See, for example, the letter from BDO in respect of IASB (2009): ‘We are not convinced that an expected loss model would have done anything to identify the onset of the global financial crisis at an earlier stage, or reduce the extent of the impairment losses that were reported after the onset of the global financial crisis.’ Comment letters on IASB (2009) are available from: http://archive.ifrs.org/Current-Projects/IASB-Projects/Financial-Instruments-A-Replacement-of-IAS-39-Financial-Instruments-Recognitio/Impairment/ED/Comment-Letters/Pages/Comment-letters.aspx [Accessed 24 April 2018].

70. We did not see any evidence that the standard setters themselves saw any contradiction in the fact that the crisis had started in the U.S., where loss recognition tended to occur earlier than elsewhere.

71. Comment letters on FASB/IASB (2011a) are available from: http://www.fasb.org/jsp/FASB/CommentLetter_C/CommentLetterPage&cid=1218220137090&project_id=2011-150 [Accessed 18 August 2017].

Additional information

Funding

This work was supported by ICAEW Charitable Trusts [grant number 5-453].