Hicksian Income in the Conceptual Framework
We are grateful to Richard P. Brief for correspondence and to Mrs Leena Baxter for access to the library of the late Professor Will Baxter and to relevant correspondence. Our thanks are also due to David Gwilliam, Mike Jones, Richard Slack, Ananda Ganguly and Min Qi, as well as to participants in the Accounting Standards Board's Academic Panel meeting, London, 4 November 2005; in the LSE Economics of Accounting workshop, 13 December 2007; in the BAA FARSIG Colloquium on The Future of Financial Reporting, London, 11 January 2008; in the EAA Annual Congress, Rotterdam, May 2008; in the AAA Annual Meeting, Anaheim, August 2008; and in the University of Sydney Accounting Research Seminar, 12 December 2008, for their constructive comments on earlier versions. We are also grateful for the advice of this journal's editor and referees and for the accompanying commentary by Frank Clarke. The remaining faults are ours.
Abstract
In seeking to replace accounting conventions by concepts in the pursuit of principles-based standards, the FASB/IASB joint project on the conceptual framework has grounded its approach on a well-known definition of income by Hicks. We welcome the use of theories by accounting standard setters and practitioners, if theories are considered in their entirety. Cherry-picking parts of a theory to serve the immediate aims of standard setters risks distortion. Misunderstanding and misinterpretation of the selected elements of a theory increase the distortion even more. We argue that the Boards have selectively picked from, misquoted, misunderstood and misapplied Hicksian concepts of income. We explore some alternative approaches to income suggested by Hicks and by other writers, and their relevance to current debates over the Boards' conceptual framework and standards. Our conclusions about how accounting concepts and conventions should be related differ from those of the Boards. Executive stock options (ESOs) provide an illustrative case study.
The Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) have undertaken a joint project to converge and improve their conceptual frameworks for financial accounting and reporting. The overall approach was outlined in an important paper, Revisiting the Concepts, in May 2005 (FASB/IASB, 2005) which emphasized that ‘to be principles-based, standards cannot be a collection of conventions but rather must be rooted in fundamental concepts’.1 At the time of issue this was presented as an authoritative manifesto of how the two Boards intended jointly to undertake this convergence and improvement, based on and building on their existing frameworks, even though there had not been any prior exposure to allow public comment as to whether some more radical approach would be appropriate (cf. Bromwich, 2001; Dean and Clarke, 2003; Wells, 2003; Potter, 2005; Dennis, 2006, 2008; Rayman, 2006; AAA, 2007, Penno, 2008).
Here we focus primarily on what appears to be—and was predicted in that 2005 paper to continue to be—the bedrock of the Boards' ongoing development of the converged framework, namely the conceptual ‘primacy of assets’ (p. 9) as elements of financial statements. We question the claim in the 2005 ‘manifesto’ that this primacy is derived from Professor Sir John Hicks' (1946) definition of income (pp. 7, 182). We support the use of accounting theory by standard setters and practitioners, provided that the theories they choose are considered in their entirety, and do not have their elements cherry-picked opportunistically to suit standard setters' immediate objectives. Moreover, theory is best understood as a whole, in spirit and nuance, instead of taking short quotations and interpreting them out of context. The main objective of this article is to indicate the dangers of the inappropriate use of theory and thereby to assist accounting policy makers to avoid reaching unsupported conclusions about measurement of income. We provide an analytical and critical case study of the use of income theory in accounting policy making.
1: THE FOUNDATIONS
Relevance of a conceptual enquiry into financial accounting and reporting depends on the assumed objective(s). We confine our analysis to the value perspective (where the financial statements report wealth and income [cf. Van Cauwenberge and De Beelde, 2007]), as this is also the objective embraced by FASB/IASB (2005).3 Starting with the overriding objective of ensuring the usefulness of accounting figures in making economic decisions, including assessment of cash flow prospects, the frameworks of the Boards focus on ‘enterprise resources, claims to those resources, and changes in them’ (p. 3). This leads to definitions of the elements of financial statements, beginning with assets. It has been tentatively decided for the joint framework that, as a working definition, ‘an asset of an entity is a present economic resource to which the entity has a right or other access that others do not have’.4
The definition of assets, which is in substance similar to the definitions in the existing individual FASB and IASB frameworks, is offered in contrast to ‘earlier efforts that included deferred debits among assets’ (FASB/IASB, 2005, p. 6). It is then argued that all other elements in financial statements can be derived from the definition of assets, which gives assets ‘conceptual primacy’ and leads to the ‘asset/liability’ view of income measurement, ‘in which income is a measure of the increase in the net resources of the enterprise during a period, defined primarily in terms of increases in assets and decreases in liabilities’ (p. 7).5
The Boards' objection to ‘deferred debits’ (Sprouse's ‘what-you-may-call-its’—Storey, 2003, p. 44) is that they are allegedly indefinable without circularity, being simply the result of the revenue and expense, or matching, approach to measuring income. However, at least in principle, the assertion that traditional accounting conventions allow into the balance sheet items that would not meet the FASB/IASB asset definition appears to be false. A traditional U.K. professional textbook such as Cropper (1930) explains in relation to items of deferred revenue expenditure that these ‘must be carefully reviewed, and . . . . may be “held up” as an asset legitimately, if written off over a reasonable period. It is assumed in such cases that benefits will accrue in succeeding years from the expenditure, and so these years should bear their proportion of the burden’ (p. 94). So the deferred debits resulting from matching, as traditionally understood, must also represent probable future economic benefits in the form of estimated future cash flows.6
It is at this point that the Boards' framework purports to be ‘grounded in a theory prevalent in economics: that an entity's income can be objectively determined from the change in its wealth plus what it consumed during a period’ (FASB/IASB, 2005, p. 7). What Hicks (1946, pp. 178–9) called Income No. 1 is cited in support. He defined this on p. 173 as ‘the maximum amount which can be spent during [a period] if there is to be an expectation of maintaining intact the capital value of prospective receipts (in money terms)’. The Boards' assertion of the conceptual primacy of assets, and of the superiority of the asset/liability view over the revenue and expense perspective in measuring a business's income, is based on this foundation (FASB/IASB, 2005, p. 7).
Although it does not specifically cite Hicks, the SEC's (2003) staff report on the Adoption . . . of a Principles-Based Accounting System supports the FASB's analysis, observing that
from an economic perspective, income represents a flow of, or change in, wealth during a period. Without first having an understanding of the wealth at the beginning of the period, it is not possible to determine the change in wealth during the period. The accounting equivalent to identifying ‘wealth’ is identifying the assets and liabilities related to the class of transactions. This identification of wealth acts as a conceptual anchor to determining revenues and expenses that result from the flow of wealth during the period. Historical experience suggests that without this conceptual anchor the revenue/expense approach can become ad hoc and incoherent. (s. III.B)
The Boards' attempt to ground their converged framework for accounting principles on a sound economics foundation is to be welcomed.7 Unfortunately, their chosen foundation will not support the particular structure that the Boards propose to erect. Although Hicks (1946) was concerned only with an individual's income, his definition of Income No. 1 can be reformulated for a company as equal to ‘the maximum amount that could be distributed to the equity shareholders in a period and leave intact the capital value of the company's prospective receipts as at the beginning of the period’ (e.g., Solomons, 1961). Ex ante this will be based on what is expected about cash flows and interest rates at the beginning of the period, and ex post on what actually occurs during the period and on revised expectations about the future at the end of the period.
These ideas can be made precise and usefully defined for later using symbols. Adapting the notation in Bromwich (1992, chaps 3, 4): cash flow for period 1 as estimated at the beginning of period 1 (at time 0) = C1t0; and as realized during period 1 and known at time 1 = C1t1. The value of prospective cash flows arising in periods 2 and onwards, as foreseen at time 0 = V1t0; and as foreseen at time 1 in the light of up-to-date information and revision of expectations during period 1 = V1t1. Given an unchanged discount rate r, the value of prospective cash flows in periods 1 and onwards, as foreseen at time 0 = V0t0= (C1t0+ V1t0) (1+r)−1; and as re-estimated with hindsight at time 1 in the light of new knowledge and revision of expectations during period 1 = V0t1= (C1t1+ V1t1) (1+r)−1. Income No. 1 ex ante for period 1 = (C1t0+ V1t0) − V0t0= rV0t0 and (Hicks') Income No. 1 ex post for period 1 = (C1t1+ V1t1) − V0t0 . If the rate of interest r is expected to remain constant, and all income and only income is distributed/consumed, future periods' income ex ante will also remain constant (at rV0t0), that is, it is the ‘permanent income’ (e.g., Beaver, 1998).
FASB/IASB (2005, p. 18) quote Hicks' observation that Income No. 1 ex post possesses ‘one supremely important property . . . [That kind of income]ex post is not a subjective affair, like other kinds of income; it is almost completely objective.’ But the quoted words are selected to exclude a critical qualifying condition. The relevant full sentence reads (Hicks, 1946, pp. 178–9, with the missing part italicized for emphasis by us): ‘So long as we confine our attention to income from property, and leave out of account any increment or decrement in the value of prospects due to changes in people's own earning power (accumulation or decumulation of “Human Capital”), Income No. 1 ex post is not a subjective affair, like other kinds of income; it is almost completely objective’.
What does the critical omitted qualifying clause (which was worded slightly differently in the 1939 first edition—Clarke, 1988, p. 423) imply? The concept of income as ‘current cash flow plus increase (decrease) in the net present value of the entity's expected future cash flows’ (excluding transactions with owners), or ‘cash flow plus actual capital accumulation’, has long been advanced by many leading academic accounting authors in writing for practitioners (e.g., by Edwards, 1938, in the U.K.). However, this concept of income is fully determinable and objective only in the presence of complete and perfect markets (i.e., when every resource and claim on future cash flows has been commoditized into fully exchangeable assets and where everyone faces the same prices, including the discount rate (Beaver and Demski, 1979). Such fully exchangeable assets are what Hicks calls property (referring presumably to land, traded investments, agricultural and mineral commodities, etc.). Under these circumstances, ‘the capital value of the individual's property at the beginning of the [period] is an assessable figure; so is the capital value of his property at the end of the [period]; thus, if we assume we can measure his consumption, his income ex post can be directly calculated' (1946, p. 179). In this situation, there is indeed no doubt about the magnitude of wealth and therefore about the magnitude of changes in it.8 But then the reporting of income is redundant; it adds nothing to the knowledge about wealth.
Since markets are rarely perfect or complete, the value of most if not all business enterprises includes significant cash flow prospects that are not reliably captured in the observable market prices of their net assets, no matter how liquid many individual markets are.9 This internal goodwill component of value (i.e., the value of super profits over and above the normal rate of return on net assets) depends inter alia on the skill with which management and the workforce exploit an enterprise's resources and its markets, and its business, social and political opportunities—what Hicks labels human capital.
In general therefore an objective version of Hicks' No. 1 ex post concept of the business income of a listed enterprise is more likely to be found in the measure of its shareholder return used in financial economics (dividend plus change in share price), that is, the change in its capital value on the stock market, than in the change in the enterprise's net assets.10 As discussed further in the next section, this view was later articulated by Hicks himself (1979). But if firms are merely to report their stock price return (plus dividends) as their income, their accounts are again redundant, at least for valuation and investment decisions.
2: ‘FIRM’ OR ‘NET ASSETS’?
In the real world of incomplete and imperfect markets, there is no justification for the FASB/IASB (2005) paper's rendering (at p. 18) of Hicks' capital value as ‘in accounting terms, its assets and liabilities’.11 There is of course an extensive academic literature exploring how far concepts and measures of asset and liability value that are consistent with (while not generally capturing all of) Hicks' underlying model of capital value may be developed (including the literature on deprival value, e.g., Edey, 1974; Baxter, 1984; on current exit value, e.g., Chambers, 1966; Clarke and Dean, 2007; and more recently on ‘fair value’, e.g., Benston et al., 2006; Bromwich, 2007; Hitz, 2007; Sunder, 2008; and Dean et al., 2010); and how changes in such net asset values may be related to Hicks' notion of Income No. 1 (e.g., Introduction to First Edition in Parker et al., 1986; a classic treatment is Edwards and Bell, 1961). Any such links require further substantial restrictive assumptions to handle inter alia what are identified in the FASB/IASB paper (2005, pp. 15–16) as the cross-cutting issues of uncertainty, unit of account and management intentions.
FASB/IASB (2005) cite Hicks (1946) who analyses income of individual persons, making no reference to firms. However, Hicks (1979) revisits the earlier analysis and begins by commenting that an early nineteenth-century mill-owner, in trying to estimate the profitability of his business, would be seeking to ascertain ‘the maximum that could be safely taken out of the business . . . without damaging the prospects of the business. But that, it is clear, would be a matter of judgement.’ He argues that, with the advent of the income tax, and of the joint-stock corporation, there are other parties now interested in knowing the business's profitability and ‘at this point the accountant enters’.12
Like Ijiri (1975), Hicks observes that the accountant's approach needs to be as objective as possible to minimize disagreements, not measuring profit in the way the mill-owner himself might do. According to Hicks, the accountant's approach naturally draws on the mercantile tradition long-familiar to accountancy, whereby the problem of determining income from sales and trades that overlap across accounting periods can be solved relatively simply by carrying forward the inventory at cost.13 Industrialization required the determination of the periodic cost of using long-lived assets such as machinery in the form of depreciation. To Hicks:
It is just the same problem as the allocation of overheads, and to that, as is well known, there is no firm economic solution. Neither has the accountant found a solution—only a name and a set of, essentially arbitrary, rules . . . There is thus no reason why there should be any simple rule which would cause the profits that are calculated by its use to have any correspondence with the income that would be assessed by the criterion with which we began—the maximum that can be safely taken out of the business. (pp. 4–5)
Hicks then turns to exploring what the depreciation for a period would have to be to satisfy this criterion. He is only able to do so by postulating a purchase of the whole business at time 0 and a sale of the whole business at time 1, to obtain an objective measure of (using our notation) V0t0 and V1t1.
He notes (in our notation, not his terms) that:
There can, I think, be little doubt that an accountant, who was asked to do the accounts of a business with this peculiar history, would refuse to do them in terms of V0t0 and V1t1; he would insist in doing them in terms of . . . the values which ‘stand in the books’. The economist, however, would find V0t0 and V1t1 much more interesting . . . it would be these market values which he would want to take as representing the initial and final capital. (p. 6)
So we can see that what Hicks here recognizes as an objective ex post measure of a firm's income (the change in the firm's market value) does not provide the foundation for a measure based on the change in a firm's net assets sought by FASB/IASB (2005).
Hicks' (1979) argument and analysis finally led him to regard as current profit that defined by Lindahl (1933): that is, (C1t1+ V1t1) − V0t1= rV0t1.14‘This is effectively what Friedman would call the permanent income derived from the business’ (Hicks, 1979, p. 11, assuming constant r). And although Hicks has been ‘looking for a definition of current profit which, so far as possible, should register the performance of the business within the year, excluding what has happened before and what is to come after . . . V1t1 . . . would appear to have a large part, even, in many cases, the dominant part, in determining the current profit’ (p. 10).15 Moreover, it is the income of the proprietors, rather than of the business (p. 11).16 (Emphases are in the original.) Thus Hicks accepts that his desired measure of income has to include major subjective elements.
In short, Hicks does not find a satisfactory, practical way of defining a business firm's income that could be used in accounting, whether ex ante or ex post.
3: HOW USEFUL IS INCOME NO. 1 EX POST?
There is, however, an even greater problem with the FASB/IASB's (2005) reliance on Hicks' concept as the bedrock of its approach to the conceptual framework. This applies independently of the relation of accounting asset/liability measures to Hicks' ‘capital value’, and independently of whether Income No. 1 ex post can be objective. In a subsequent paragraph, Hicks (1946) goes on to say: ‘Ex post calculations . . . have no significance for conduct . . . On the general principle of “bygones are bygones”, it can have no relevance to present decisions.’ This undermines the FASB's/IASB's attempt to use Hicksian income as the foundation for their asset-liability view to serve the ‘overriding objective of decision usefulness’. Their structure is built on sand, as it is only the overall wealth available at the end of each period, not the ex post income of the period, that is relevant for decision making about future investment and consumption, etc. So the only relevant decision-orientated aggregated information that can be provided by financial reports is information about the endowment of wealth available to the firm (e.g., Bromwich and Wells, 1983). That is, income figures cannot facilitate any decision making incremental to that which could be made from being told only the endowment at the end of the period—unless income can be shown to generate some information about the future that is not already contained in the endowment.
Hicks does concede some role for his Income No. 1 ex post: such calculations ‘have their place in economic and statistical history; they are a useful measuring rod for economic progress; but . . . they have no significance for conduct’ (1946, p. 109).
However, it may be argued that one cannot expect to be able to predict the future and income ex ante without some knowledge based in past experience (e.g., as hypothesized in Friedman, 1957). Hicks discussed further the role of accounting in this regard in a book review for the Economic Journal (Hicks, 1948). As explained by Brief (1982), supported by extensive quotations, Hicks here endorsed what he thought to be an important argument (buried in the compilation of miscellaneous articles, etc., constituting the book being reviewed), namely the importance of the underlying objectivity of the ‘statistics’ that the accounts record: hence the justification for historical cost and the dubious value of introducing subjective adjustments (e.g., to the lower of historical cost and market value). So, for example, the bias introduced by historical cost in inflationary times is a matter for correction by users in their interpretation of the accounting numbers.
However, Hicks added his own observation that bare statistics are never sufficient: so what is to be done for external shareholders?17 The accountant ‘has thus some public obligation to pack into his figures the maximum of information even if he can only do this, within the limits prescribed, by some sacrifice of objectivity. How ought this difficulty to be got over? Should it be laid down that companies must publish an audited report as well as audited accounts? Or would this make the accountant, more than ever, master of the destinies of us all?’ (1948, p. 564, emphasis added).
So the main issue with Income ex post is ‘how much of the future is it useful to bring into accounts of the past if they are to be helpful in forming expectations about future Income ex ante?’ (cf. Barth, 2006). A considerable amount is inevitable, even in traditional accrual accounting that attempts to match revenues and expenses (Edey, 1970). How much more is useful must be primarily an empirical question, to which the answer may vary according to how far permanent and transitory elements can be distinguished. This varies according to different types of business activity (e.g., Penman, 2007; Bezold, 2009); while also being subject to different users' needs and trade-offs of relevance and reliability (cf. Sundem, 2007). There is no necessary merit in simply tracking Hicks' Income No. 1 ex post (Sunder, 1997, p. 79; cf. Schipper and Vincent, 2003) even if this were possible using accounting numbers—except perhaps for comparing with previous internal estimates of income ex ante in order to improve future estimations (e.g., Edwards and Bell, 1961; Bromwich, 1974; Goford, 1985). This is an information content approach to the conceptual framework and the usefulness of income measures (e.g., Christensen, 2010; cf. Macve, 2010).
In short our fundamental objection to the FASB/IASB (2005) paper as a statement of the conceptual foundation that should underpin its framework is that, on Hicks' own assessment, ex post income, whether more or less subjective, is largely irrelevant to the Boards' decision usefulness objective for financial accounting and reporting.
4: A ROLE FOR INCOME NO. 1 EX ANTE?
Some authors (e.g., Black, 1993) have argued that the primary focus of accounts (not just of their users) should be on estimating standard stream income. Given that stream and a (constant) discount rate one can directly derive the value of the firm by capitalization (e.g., Whittington, 1983, p. 33).18
The FASB/IASB (2005, p. 7) say that a concept of income founded ultimately on the definition of assets is necessary because, among the proponents of the alternative (the revenue and expense) view, ‘none could meet the challenge’ of defining ‘income directly, without reference to assets or liabilities or recourse to highly subjective terminology like proper matching’ (emphasis in the original).19
Hicks himself could ‘meet this challenge’. Dissatisfied with the adequacy of his ‘No. 1’ version when interest rates change, he offered ‘Income No. 2’, defined as the amount that an entity can consume in a period and still expect to be able to consume the same amount in each ensuing period (1946, p. 174). In the case of a joint stock company this translates as ‘the maximum dividend the company could pay this period to its current equity shareholders and expect to be able to pay them the same dividend in all future periods’, which is equivalent to what financial analysts call its maintainable (or permanent) income. That is, Income No. 2 is the sustainable perpetuity based on the existing information set. Within Hicks' (1946) framework of analysis, Income No. 2 is, as he notes on p. 174, the same thing as Income No. 1 only when there is no expected (or actual) change over time in the rate of interest at which future cash flows are discounted to obtain the capital value (i.e. the yield-curve is flat).20
If the company held only fixed interest irredeemable government securities, this maintainable income would be objective.21 Otherwise uncertainty about future changes in the yield-curve, and about risk premia required on corporate bonds, equities and real property, would mean that maintainable income, even from frequently traded assets, would inevitably be subjective (Macve, 1984; Draper et al., 1993).
It was this No. 2 concept of income that underlay the proposals in the U.K.'s Sandilands Report (1975) for current cost accounting (an ‘entry value’ approach—e.g., Clarke, 1988). Sandilands (1975, p. 47, para. 166) said: ‘ no accounting system can predict a company's future prospects. However, an accounting system can at least ensure that the profit figure reported is such that, if the profit for the year were fully distributed, it would not prejudice the ability of the company to continue to generate the same profit in future years.’ Whereas Sandilands rejected general price level adjustments (i.e., for inflation), Scott (1984, p. 205) argues for the importance of assisting users to estimate (real) standard stream income (alongside gain or change in value) and, while critical of much of the methodology proposed by Sandilands, suggests ways in which accounts can be best adapted under changing prices to achieve this, many of which must inevitably be subjective. Scott concludes (drawing on his own experience as an investment trustee for a charity): ‘First . . . there is a strong practical need for estimates of standard stream income and, second . . . useful estimates can be provided—but not, so far, or perhaps ever, by accountants qua accountants’ (p. 240). This conclusion appears close to Hicks' (1948) view that adjustment to the basic historical cost accounting records should be made, as far as possible, by those using and interpreting the accounts, rather than within the accounts themselves.
Given maintainable income one can, under further restrictive assumptions, also derive definitions and measures of assets and liabilities that would be consistent with this concept, but they are to be derived from income, not income from them. Thus, Ohlson (2006) argues that investors like to have a natural starting point in the income statement as they try to forecast subsequent periods' sustainable earnings.22 This concept of sustainable earnings is again consistent with Hicks' (1946) No. 2 Income.23 Ohlson therefore argues that reporting such maintainable earnings would require that assets and liabilities be derived from income and not vice versa.24
Finally, given the conceptual tension between Hicks' Income No. 1 (expressed in terms of capital value) and Income No. 2 (expressed in terms of maintainable income), there are also grounds for believing that the most relevant income concept for users and their economic decisions will often vary with their individual circumstances and conditions (Paish, 1940). Thus, someone facing a major expenditure (e.g., of a family wedding or an unexpected and uninsured illness) would be concerned more about its effect on their wealth (Income No. 1), while someone facing retirement might be more concerned with how much maintainable pension they are entitled to, or can obtain from their investments (Income No. 2).
This insight can help relate to each other the underlying motivations of those who identify with the asset/liability approach to accounting income (more like Income No. 1) and those who identify with the revenue/expense (or matching) view (generally involving more smoothing and thereby closer to Income No. 2). The two are at once complementary (in the sense that each provides a different but useful perspective on the firm) and opposed (with regard to methods of measuring enterprise income). Neither approach should therefore necessarily be preferred in principle over the other as the basis for accounting standard setting. In each case the relevant approach should be chosen on its merits in that context (consistent with the argument in Penman, 2007; cf. Chisman in Jones and Slack, 2008). Since neither perspective can perfectly measure either of the two underlying income concepts, it is important to recognize that in some cases stocks (e.g., of assets) are more readily measurable than flows (of revenues and expenses) while in other cases the converse holds. So the two approaches complement each other and accountants must to learn to live with this duality.25
In short, the FASB/IASB (2005) paper, in focusing solely on Hicks' Income No. 1 ex post, ignores the conceptual and practical importance of Hicks' Income No. 2 ex ante for decision making. (In the Appendix, we illustrate the differences between the two income approaches by exploring their distinction and the consequences for the accounting for stock option expense controversy.)
Since FASB/IASB (2005, p. 7) claim that income is not definable ‘directly, without reference to assets or liabilities or recourse to highly subjective terminology like proper matching’, it follows that neither can be its components such as revenue and expense. This has led the Boards into some difficulty, for example, in relation to depreciation expense (IASB, 2009, Basis for Conclusions BC54), and to revenue recognition (FASB/IASB, 2008b), where the Boards are stated (at para. 5.20) to be uncomfortable with the potential implication of valuing contract assets and liabilities at inception in that it could lead to recognition of Day 1 revenue and income ‘before the entity transfers to the customer any of the goods and services that are promised in the contract’. This discomfort illustrates the inevitable continuing power of conventions—in this case of matching—at the heart of conceptual debates, that we discuss further in the next section.
To summarize these arguments, Table 1 indicates briefly the differences between what Hicks actually says about income and what the Boards claim in their 2005 paper about his position. These differences, and the discussion above, indicate clearly the dangers of cherry-picking selected fragments of theory, particularly when the fragments are themselves misunderstood or misinterpreted in isolation from the whole. The chosen fragments of theory simply do not support the structure the Boards seek to erect on them.
5: CONVENTIONS VERSUS CONCEPTUAL PRINCIPLES
FASB/IASB (2005) see the conceptual framework project as a crusade against conventions: ‘To be principles-based, standards cannot be a collection of conventions but rather must be rooted in fundamental concepts’. Economists writing about accounting have generally been very respectful of accounting conventions.26 As noted above, Hicks argued that the accountant's solution to the depreciation problem was a natural development from merchandise accounting. He also credited the accountant's view of capital as a fund with a profound influence on English classical economists,27 and noted that Marshall seemed content with the accountant's approach to depreciation (Hicks, 1974, p. 313). While twentieth-century inflationary pressures and tax policy changes put extant conventions under great strain (1974, p. 312), Hicks appeared to believe that the necessary adjustments could best be made by those using and interpreting the accounts rather than by expecting reform of the accounts themselves. Indeed this could interfere with the underlying, objective statistical record (Hicks, 1948; Brief, 1982).28
Kaldor too (1955, p. 123) noted that ‘[T]he accountant is rightly in search therefore of a concept of income ex post which is as near a counterpart as can be found to the investor's income ex ante. In the light of the foregoing analysis it is not surprising that the accountant's definition of income ex post is based, as it can only be based, on a series of admittedly arbitrary conventions whose value depends, to a large extent, on their status as time-honoured conventions—i.e., on their steady and consistent application.’29
There may of course also be value in sticking to agreed rules30 for purposes of contractual and other settling up such as taxation, bonuses, partners' profit shares, loan covenants, etc.; Lindahl's ‘restatement with hindsight’ (as adopted by Solomons, 1989, in his Guidelines) would never allow closure between contractual parties.
That is not to say accounting conventions cannot be improved: the quoted economists had, perhaps, an overindulgent view of accounting's achievements which may not be surprising given that they were generally writing before the advent of U.K. standards and the wider understanding of just how inconsistent many accounting practices are (although Clarke, 2010—following 1988, p. 416—notes Hicks' involvement in the late 1940s in a joint committee with ICAEW). But it is naïve of FASB/IASB to overlook the power of conventions, and their surrounding expectations, in maintaining the ordinary fabric of social structure and interaction.31 The important questions to ask are: Does analysis of how conventions and social norms operate suggest that it is time to modify them? If so, how? Will the benefits outweigh the cost?
Solomons (1961) predicted the twilight of income measurement within twenty-five years, yet in 1989 he was still writing his Guidelines for the U.K.'s Accounting Standards Board (ASB) on how best to report income. Similarly, Ohlson (1987), in his commentary on Beaver and Demski (1979), argued that the reporting of income is too embedded in accounting tradition to be abandoned, despite the inescapable conceptual limitations.
The logic derived from an imaginary perfect world is frequently insightful, but it cannot be applied to our imperfect world without adaptation. Even if we could agree on one of the twelve Hicksian, or any other, concepts of income, we know that the current accounting conventions cannot measure them precisely. Instead of seeking to replace conventions with concepts, the Boards could seek a better understanding of how and why accounting conventions work, and which of them could be adapted to the current financial reporting environment in the light of relevant conceptual considerations.32
To argue, as FASB/IASB (2005) do, that income based on accounting conventions cannot measure Hicksian income does not give superiority to income based on the concepts of assets and liabilities. Since net asset values do not sum to equity value in incomplete and imperfect markets of our world, income based on net asset values does not equate to Hicksian income. Moreover the Boards' related attempts to pin down the definitions of the elements of financial statements in their conceptual framework project are unlikely to be helpful in this endeavour, or even to be achievable.33
While much of the conceptual discussion reviewed here, including the FASB/IASB (2005) paper itself, appears to recycle arguments from more than fifty years ago (e.g., Dean, 2008), there have been interesting recent practical developments in alternative ways of setting out income and value in accounting reports, given dissatisfaction with existing conventions. The most conspicuous of these at the present time are the developments in supplementary reporting of life insurance profitability according to a (market consistent) embedded value (MCEV) model. MCEV is a form of fair value accounting that was originally developed by actuaries for financial management and control of life insurance business, based on discounted present values. It has increasingly been adopted worldwide for supplementary reporting, to overcome the severe limitations of the traditional solvency approach to life insurance accounting in a new world, where there has been extensive restructuring of financial institutions together with changes in both their market opportunities and in their regulation. MCEV now uses available market prices as benchmarks, wherever feasible, to derive opening and closing ‘economic balance sheets’ for the in-force—that is, existing—business each period, and analyses the changes between them in terms of predicted return and variances from expectations. It thus bears structural similarity to a Hicks No. 1 ex ante–ex post cycle. However, so far it has been rejected by the IASB in their own project on insurance accounting. (For further discussion, see, e.g., Goford, 1985; Horton et al., 2007.)
ASB (2007) has recognized that this new MCEV approach has potentially major implications for profit measurement and reporting in other industries. Standard setters could usefully design more effective prescriptive standards on the basis of such experiments to evaluate current concepts, practices and their consequences. Such an experimental approach, with due allowance for our state of ignorance, may be better than the attempts at abstract definitional refinement of concepts such as income or assets. This abstract approach becomes especially risky when it is based on fragments of theory(ies) conveniently selected to serve the objectives of standard setters in violation of the integrity of the theory(ies).
The Boards could fruitfully reorient their efforts in this direction. As the example of executive stock option compensation in the Appendix illustrates, improvements in financial reporting practice and standards are often driven by a recognition that current conventions may have outlived their practical usefulness, rather than by the logical implications of any underlying conceptual framework.
6: CONCLUSIONS
We have presented reasons why Hicksian concepts of income cannot be invoked to support the asset-liability perspective promoted in the FASB/IASB (2005) manifesto for the Boards' joint conceptual framework project. Firstly (as discussed in section 1) firms do more than just earn a return on their identifiable net assets. These assets may or may not have a readily available market value. There is also normally the element of what Hicks calls human capital in how firms exploit their opportunities, so even if asset markets are in competitive equilibrium, if they are not complete this creates internal goodwill.34 Measurement of this inevitably requires subjective estimation, precluding the feasibility of objective measurement even ex post, contrary to the objectivity claimed in FASB/IASB (2005).
Secondly (as discussed in section 2), Hicks has difficulty in arriving at a practical measure of business income that could be reflected in accounts, as he finds it necessary to conduct the analysis at the level of the change in the value of the firm itself, not of its net assets, and this income is that of the proprietors rather than of the business. He finds that the measure of this income, even ex post, is largely driven by changes in expectations about the firm's future cash flows, rather than by the realized cash flows of the period just completed.
Thirdly (as discussed in section 3) our fundamental objection to FASB/IASB (2005) as a conceptual foundation for financial reporting is that Hicks' own assessment of any practical ex post measure of income, whether more or less subjective, is that it is irrelevant to decision making—and therefore it must be largely irrelevant to the Boards' decision usefulness objective for financial accounting and reporting.35 At best it can provide relevant statistics for prediction—but that may imply that adjusting the factual record about past transactions for changes in expectations about the future is best left to decision makers as users. Assistance from competing information intermediaries such as analysts, the press, and academic research based on information from within and without the firm may also help. Adjusting the financial statements themselves for this purpose may therefore be unnecessary and it is up to the Boards to demonstrate what comparative advantage accountants have in adding value through bringing ever more of management's and analysts' estimates of the future into audited financial statements and reports (e.g., Christensen, 2010).
Fourthly (as discussed in section 4), if the focus were to shift primarily to income ex ante, it may be argued that an equally important perspective on what the future holds is to consider not just the likely changes in future value (or gain), as captured by Hicks' No. 1 ex ante concept of income, but also the standard stream (No. 2 ex ante) view of income, as useful in helping to triangulate the amount to be reported as a firm's expected earnings. As Paish (1940) pointed out, there are legitimate economic motivations underlying interest in both views. Given the variety of user preferences and objectives, any choice between them can itself only be an accounting convention. Therefore, any measures to assist estimates along both these dimensions may usefully be reflected in general purpose financial reports. For example, as far as practicable, both the current value of net assets and changes in net assets may be reported, without requiring all the changes to be reported as earnings (e.g., Horton and Macve, 1996, 2000).
The conceptual framework project of FASB and IASB will not be able to eliminate either of the two income concepts; user preference may force them to retain both. In many situations the revenue/expense matching view of income/earnings is closer to the maintainable earnings concept than the asset/liability view (e.g., Bezold, 2009). It seems unlikely that the Boards' attempt to eliminate the revenue/expense view in favour of the asset/liability view can succeed. Indeed, it is already in the process of being deconstructed in their Revenue Recognition and Fair Value projects (FASB/IASB, 2008b; IASB, 2009).
After exploring the dangers of standard setters misapplying selected fragments of misunderstood theories, the final section examined the role of conventions in measuring income. The Boards' conceptual framework should seriously attend to the necessary interrelationship between concepts and conventions in practical affairs. Indeed, revisiting the concepts in this way will help the Boards as well as their constituents to understand why accounting practice has to include conventions and how those conventions, despite there being no clear framework for identifying what is optimal (e.g., Demski, 1973; Sunder, 1997; and Christensen and Demski, 2003) have become so powerful as calculations of performance, including business performance, in the modern world (e.g., Hoskin and Macve, 2000). We therefore suggest a revision of the key sentence we quoted at the beginning from p. 1 of FASB/IASB (2005) to read: ‘To be principles-based, standards have to be a collection of (socially) useful conventions, rooted in fundamental concepts’.
In summary, Hicks' (1946) analysis does not provide a conceptual basis for the FASB/IASB's exclusive focus on a balance sheet approach to the financial reporting. Nor does it help address the difficult problem of measuring and reporting business performance and identifying drivers of value creation.
We have argued that the Boards should try to understand the practical roles of conventions in financial reporting and how and when they might be modified to serve the legitimate interests of interested parties (e.g., by reducing apparent inconsistencies that no longer serve any purpose). However, the corporate structure of these Boards, designed for debating technical issues, may not necessarily equip them to address such challenges. The ultimately political nature of the social welfare issues may be better suited for broader social institutions reflecting social norms of the kind that the idea of generally accepted accounting principles was originally meant to encapsulate. How to construct useful, practicable, and broadly accepted financial reports may require evolution as well as design (e.g., Basu et al., 2009). Whether it is desirable for the Boards themselves to converge towards becoming one, monopolistic standard setter remains an open question (e.g., Bromwich, 1992; Dye and Sunder, 2001; Sunder, 2009, 2010). Clearly the Boards' unsuccessful appeal in FASB/IASB (2005) to the claimed objectivity of Hicksian income as an unambiguous foundation for financial accounting and reporting fails to resolve these issues.
Appendices
Appendix
EXECUTIVE STOCK OPTIONS—A CASE STUDY IN INCOME CONCEPTS AND THE ROLE OF CONVENTIONS
In this Appendix we explore through a highly simplified example how the arguments we have presented about the essential distinction between Hicks' No. 1 and No. 2 versions of income might play out in the case of a controversial example like accounting for executive stock option (ESO) expense. We also examine how changes in the role of accounting conventions such as matching appear to be more relevant in understanding how practice has changed than concepts grounded in the standard setters' asset/liability approach.36 We initially assume certainty (apart from the previously unannounced arrival of a new CEO) so there is generally no difference between income ex ante and income ex post, or between risks of investment. For simplicity we assume discrete compounding with cash flows arising at the end of each year throughout. Numbers are exaggerated to bring out the effects more clearly but it is assumed (unless otherwise stated) that all parties are price takers in perfectly competitive markets.
Under IFRS2 (IASB, 2004), which is essentially similar to the revised SFAS123 (SFAS123R) (FASB, 2004), when stock options are granted to executives they may no longer be accounted for simply at any difference between the exercise price and the current market price of the related shares (which may be zero), but must be recognized at their ‘fair value’ on the date of issue. In a perfect market the Black-Scholes option pricing model shows how an option will have both an intrinsic value and a time value (the latter reflecting the risk of how far the option may move into and out of the money as the underlying share price changes stochastically until the date of its exercise). However, in our simplified world of certainty there will be no such risk from share price volatility and the options will only have intrinsic value (assuming they can be exercised at any time). This is sufficient to illustrate the point at issue here about income.
Suppose Company A currently has 100 shareholders each owning an equal percentage of the 1,100 shares (11 shares each) traded on an exchange. Cash flow forecasts at time t0 are $2,200 p.a. If the rate of interest is expected to remain at 10% p.a. the value of the firm is $22,000, that is, $20 per share, with each shareholder having a holding of 11 shares worth $220 and an expectation of receiving dividends of $2 per share (equal to EPS) for the foreseeable future.37 Hicksian income ex ante (both No. 1 and No. 2) totals $2,200 p.a. ($22 per shareholder) and is also the permanent income. For simplicity, assume no dividend is paid before the end of period 1 (t1).
At time t0 company A unexpectedly hires a new CEO to start work immediately and incentivizes her with stock options allowing her to purchase shares at any time at a strike price of $0. The stock market (including all current shareholders and the CEO) estimates that the effect of this CEO's arrival will be to increase the value of all expected cash flows by $300 p.a. to $2,500 p.a. (an increase worth $3,000), but executive labour market conditions mean that the CEO cannot extract any quasi-rent and is only worth the going rate of $2,000. This raises the value of the firm to $25,000.38 The option grant must therefore be $2,000/25,000 * 1,100 = 88 shares. Existing shareholders retain 1,012 shares (and are better off by $1,000 as the current share price rises to [$25,000/1,100=] $22.73).39 The CEO receives 88 options worth $22.73 = $2,000.40 Each share will receive an extra $300/1100 dividend each year = $0.273, to give a new total dividend of $2.273 per share, equal to the new EPS.
If no dividend is paid from the windfall gain at t0, the revised Hicksian income No. 1 and No. 2 ex ante now totals $2,500 p.a. (to existing and potential shareholders) but as far as existing shareholders are concerned it is $2,300 p.a. (i.e., diluted by the effective transfer of 88 shares attracting a dividend of $2,500/1,100 = $2.273 p.a. each = $200 in total).
Under IFRS2 and SFAS123R, however, the additional internal goodwill and increase in the stock market value of the proprietors' shares will not be recognized; while the cost of the option grant of $2,000 will be treated as an expense, depressing year 1 reported income to $2,500 − $2,000 = $500.41 If investors were then to project this as Company A's permanent income, its stock market value would correspondingly fall from $22,000 to $5,000. While this would make the reported income ‘value relevant’ in the sense of association with stock-market prices, it is clearly not ‘value relevant’ in the sense of adequately informing stock-market prices (e.g., Macve, 1998).42 If the intangible asset were to be recognized then there would be a windfall gain totalling $3,000 (of which $1,000 accrues to existing shareholders), so that Hicksian Income No. 1 for the firm becomes $2,500 + 3,000 = $5,500 and for its existing shareholders $2,300 + 1,000 = $3,300.43
Aboody et al. (2004a) and Landsman et al. (2006) document evidence that the stock market recognizes both the expense of executive stock-option compensation and the intangible asset of the additional future earnings to be generated by the CEO's arrival. So if the accounts were to be made fully value relevant (i.e., tracked the Hicksian No. 1 income of the firm) what would be the effect of interest rate changes? Expected changes will be sufficient to demonstrate the point.
Now suppose that interest rates were expected to be 10% p.a. during the first year and then to rise to 20% p.a. thereafter.44 At t0 the value of Company A would have been $12,000 before the unanticipated arrival of the new CEO and $13,636 afterwards.45 At t1 the corresponding values (of all future cash flows) would be expected to be $11,000 and $12,500 respectively.46 This gives No. 1 incomes for year 1 of $1,200 and $1,364 respectively (both equal to interest at 10% on the revised opening value).47 Under IFRS2 and SFAS123R (expensing the stock options but not recognizing the intangible asset, i.e., the value of the anticipated future increase in annual earnings) income would fall by $1,091, together presumably with the impairment to the book value of existing net assets of $10,000 caused by the rise in interest rates.48
Under Hicksian Income No. 2, the only economic change is the expectation of the increase in annual cash flows of $300 following the CEO's arrival, so income rises from $2,200 p.a. to $2,500 p.a. (of which existing shareholders will get 1,012 * $2.273 p.a. = $2,300 p.a. [a net increase of $100 p.a.], and the new CEO 88 * $2.273 p.a. = $200 p.a. as before).49
Which is the more useful measure of income?50 There are two interrelated problems here: the unrecorded internal goodwill and the effect of changing interest rates. Given the FASB/IASB's favoured asset/liability approach, recognizing only the impact of the latter on values (Hicks No. 1) increases the accounting asymmetry here. And even without this, the only partial recognition of the ESO impact (i.e., the expense without the intangible for the benefit) means that evaluation of any accounting choice, or of change in accounting standard, already faces the economic problem of the second best (Lipsey and Lancaster, 1956), that is, that fixing only one element of the problem may make the overall situation worse (e.g., Landsman et al., 2006).
Paradoxically there is actually no overall change in recognized net assets under IFRS2/SFAS123R as option expense is offset by increase in paid-in capital.51 So there appears to be some much more conventional notion of proper matching providing the justification for this treatment. As Warren Buffet famously said (see, e.g., Macve, 1998): ‘If options aren't a form of compensation, what are they? If compensation isn't an expense, what is it? And, if expenses shouldn't go into the calculation of earnings, where in the world should they go?’
It is clear that the definitions of income, assets and other such fundamental elements can serve as signposts but cannot provide definitive answers to practical questions. The opportunity for the IASB and the FASB finally to succeed in 2004 in requiring expensing of stock options probably had more to do with changes in attitudes to business transparency following the Enron debacle (e.g., Gwilliam and Jackson, 2008). As the summary of SFAS 123R noted: ‘Over the last few years, approximately 750 public companies have voluntarily adopted or announced their intention to adopt Statement 123's fair-value-based method of accounting for share-based payment transactions with employees’.
The cost (in lower reported earnings) to companies of adopting option-expensing could thus be interpreted as a signal that companies' accounting numbers were more credible overall (e.g., Morris, 1987). Of course, this also created new incentives for different kinds of firms to underreport that expense either as free-riders or because the immediate crisis of public confidence had abated before long (cf. Aboody et al., 2004b, 2006).
For our own conclusions, there would appear to have been changes in societal expectations of business legitimacy that made the new convention more useful and acceptable. The resulting political forces52 were probably more important than the conceptual niceties, which had been insufficient to resolve the controversy during the period leading to the issue of FASB's previous version of SFAS123 in 1998 (e.g., Zeff, 1997). That is not to say that the conceptual considerations are irrelevant; clearly the anomaly of the asymmetric recognition of the cost of the grant versus its anticipated future benefits (Macve, 1998) has added yet another factor (alongside other cases such as research and development) that undermines the consistency of the Boards' Conceptual Framework as asset/liability based.
The Boards' 2005 manifesto wrongly claims that their proposed revision of their Conceptual Framework is supported by Hicks' analysis of income. This example of ESOs illustrates that the recognition of this mistake might be a useful first step towards making real progress.
