Why is there inconsistency in accounting for liabilities in IFRS? An analysis of recognition, measurement, estimation and conservatism

Abstract

We report that International Financial Reporting Standards (IFRS) are inconsistent with respect to the recognition and measurement of liabilities, both in the conceptual framework for financial reporting and in accounting standards themselves. We demonstrate that this arises in part because the International Accounting Standards Board (IASB) does not make a conceptual distinction between the process of measurement, which requires a currently observable measurement attribute, and the process of estimation, which is inherently subjective. The IASB employs only the logic and language of measurement, while actually requiring entities to report both measurements and estimates in financial statements. Our contribution is to identify and interpret this conceptual conflict, to demonstrate that this has particular relevance to accounting for liabilities, and to draw implications for accounting research and policy with respect to recognition, measurement and conservatism.

Keywords:

• conceptual framework
• conservatism
• IFRS
• liabilities
• measurement
• recognition

View correction statement:

Erratum

Acknowledgements

The authors gratefully acknowledge input and support from the editorial and review process at Accounting and Business Research; seminar/conference participants at the Universities of Aberdeen, Cambridge, Edinburgh and Oxford, and at the FRC's Academic Panel, ICAS and the EAA (Ljubljana); workshop funding from the ICAEW Charitable Trusts, EFRAG and the University of Aberdeen.

Notes

1. See Section 5 and Table 2.

2. Of course, expectations of the future do currently exist. The extent to which they should be used as the basis recognition in the financial statements is discussed later.

3. As noted above attributes may or may not have a currently observable measure. If not, they are not measurable, and hence are not strictly ‘measurement attributes’. However, the literature (both academic and professional) often refers to all attributes considered for recognition as measurement attributes. To make easy reference to that literature, we also use measurement attribute in that general manner, but distinguish in our analysis between attributes with currently observable measures and those that are estimates.

4. While our focus is IFRS, our analysis is equally applicable to US GAAP (Benston et al. 2007). The FASB's conceptual framework provided the foundation and logical structure for the IASB's Framework (Storey and Storey 1998), and the 2010 revisions to both frameworks were carried out jointly by both IASB and FASB, and adopted simultaneously by both, consistent with the 2002 ‘Norwalk Agreement’ to achieve convergence (FASB 2006).

5. A more detailed summary is given in Table 2, which we discuss later in the paper.

6. This holds notwithstanding that the challenge in practice is reduced if there is, first, a larger sample of underlying events determining the value of the liability and, second, greater relevant past experience; both factors improve confidence in forecasting, as well as the information content of subsequent re-measurements. See later discussion on ‘reliable estimation’.

7. The definitions in this section are those, still extant, from the 1989 version of the Framework.

8. See EFRAG et al. (2013a) for a discussion of these alternative approaches.

9. In practice, an interpretation such as ‘more likely than not’ is consistent with both the Framework (Botosan et al. 2005) and with the binary nature of recognition (an item is either recognised or it is not).

10. See EFRAG et al. (2013c) for a discussion of this issue in the context of the current IASB project to revise the Framework.

11. This assumes that the attribute in question is determined by a single observable measure, for example, cost given by the price in an actual transaction. It is the case that sometimes the definition of an attribute may involve a choice between two or more observable measures. For example, determining fair value may require a judgement to be made about which market to observe the transaction price. Two attributes, a retail market price and a wholesale market price, may both be observable, but determining the attribute of fair value as defined by the IASB still requires a judgement, and hence leaves open the possibility of lack of consensus. We do not discuss in this paper whether in such cases the attribute of fair value is currently observable or not. The point is that the IASB has extended the notion of verifiability beyond that in the academic literature.

12. The 2010 revision to the Framework explicitly removed the word ‘measurement’ from the text.

13. The IASB's approach to faithful representation is consistent with its own (1989) definition of measurement, but this is only because that definition is similarly loose. The Framework states that ‘Measurement is the process of determining the monetary amounts at which the elements of the financial statements are to be recognized and carried in the balance sheet and income statement’ (para 99). Under this definition, anything that ends up being recognised in the financial statements has been through some (undefined) process of measurement. Equally, a ‘numerical depiction’ can be deemed to be faithfully representational even if it cannot in principle be verifiably complete, neutral or free from error. In short, these definitions broaden, and so weaken, the definition of measurement to the extent that it loses any effective meaning. Such an approach opens the door to an alternative interpretation of the role of measurement in accounting. Power (2004) notes that ‘the contemporary performance measurement imperative does not depend on a view that measurement reveals things as they really are’. A ‘measure’ in this context can be socially constructed, functioning as a measure without actually being one.

14. The leases ED does, however, propose that, for the purposes of estimating the cash flows for a lease liability, the term of a lease should be determined as the longest lease term that is more likely than not to occur. This approach is described as a recognition approach to determining the cash flows.

15. In this context, the IASB has stated ‘The IASB regards aligning IAS 37 with other IFRSs – so that all liabilities are recognised – as more important than preserving consistency with all aspects of the existing 20-year-old Framework’ (IASB 2010).

16. IAS 19 requires an entity to recognise a liability for profit-sharing and bonus payments only if, inter alia, a reliable estimate of the obligation can be made, and it gives specific conditions for this to be the case. IAS 19 also requires an entity to account for its share of a multi-employer defined benefit plan as if it were a defined contribution plan (i.e. not to recognise a liability for any deficit in the plan) if it cannot identify its share of the underlying financial position and performance of the plan with sufficient reliability for accounting purposes. For other defined benefit plans, IAS 19 explicitly assumes that an entity can use actuarial techniques to make a reliable estimate of the benefit that employees have earned.

17. The attributes described in the Framework are historical cost, current cost, settlement value and present value.

18. The requirement in IFRS for the use of unobservable fair values should not be over-stated. For example, Gebhardt (2012) reports, for non-financial companies in the STOXX Europe Large 200 Index, that only 6.4% of financial liabilities are carried at fair value and that 90.7% of these are based upon observable data (Level 1 or Level 2).

19. Some measurements allowed under IFRS 4 may not be discounted.

20. The reason given in the basis for conclusions is that this approach is consistent with a cost-based measurement. The leasing project seems to be the outlier in this respect. The discount rate is not updated, even when the cash flow estimates are.

21. The ED notes that the effect of credit risk in such transactions should be minimal, because policyholders do not take out insurance with companies that they expect to default.

22. For a review of empirical contributions, see Botosan et al. (2005).

23. In practice, of course, asset and liability values may differ for loans and other financial instruments also (e.g. Gray 2003).

24. The cost of release is captured by the amortised cost method if there is an absence of cancellation charges or similar contractual clauses (This need not hold universally – for example, if initial recognition in an entity's accounts does not coincide with initial inception of the loan, or where there are re-measurements.) In the case of the fixed-rate bank loan, this (entity-specific) amount will be higher than fair value if there is an increase in either the rate of interest or the entity's credit risk (Lipe 2002).

25. Insurance is a hybrid between pensions and deferred consideration: the liability is an expected but uncertain outflow of cash, not products or services.

26. Entity-specific measurement limitations can be assumed away in the standard-setting literature, as, for example, illustrated in Leisenring et al. (2012).

27. By provisions we mean liabilities under IAS 37. We do not discuss adjustments to asset measurements such as impairments of accounts receivable, which are not liabilities.

28. This approach of indirect, proxy measurement can be argued to provide the prevailing logic of historical cost accounting (Paton 1922, Sterling 1970). The converse case to that given above can be true also: the example of an asset generated by (some) share-based compensation illustrates the case of an exit value that is observable with an entry value that is not observable.

29. The carrying amounts of assets can also require estimation, on initial recognition for assets held under a finance lease, and on re-measurement for assets that have become impaired. Setting standards on impairment throws up similar issues to those discussed above in relation to estimating liabilities, precisely because it involves a process of estimation rather than measurement (O'Hanlon 2013). Finance lease assets, although not having an observable measure, can be estimated within a relatively narrow range if there is sufficient evidence from other sources of an appropriate cost of finance.

30. In this respect, the 2010 version of the Framework suppresses the discussion of the 1989 version, which had explicitly justified conservatism in terms of greater verifiability requirements for assets than for liabilities, based upon a presumption of opportunistic incentives arising from an agency relationship. See EFRAG et al. (2013b) for a discussion of these issues in the context of the current IASB project to revise the Framework.

31. In the light of the earlier discussion of measurability, it is ironic that the IASB is in agreement with Sterling (1970) in rejecting conservatism.

32. For listed debt, changes in market value are currently observable and so absence of measurability would not be a justification for non-recognition (Archer and Peasnell 1985). This observability does not hold in general, however, and conservatism could justifiably be positioned as the basis for consistency in accounting practice.