From compromise to concept? – a review of ‘other comprehensive income’

Abstract

This paper reviews how ‘other comprehensive income’ (OCI) entered financial reporting by tracing major Financial Accounting Standard Board (FASB) and International Accounting Standards Board (IASB) projects that required direct entries to equity and describing recent efforts to make sense of the practice. It was the fixation on net income that brought about departures from all-inclusive income, which were repeatedly made over the years without decidedly devoting attention to developing a conceptual basis. OCI was used as a compromise to incorporate current values in the balance sheet, while retaining historical cost principles in the income statement. When the practice was labeled as OCI, it became institutionalized without a clear meaning. A sense-making of the practice then replaced the debates on the adequacy of using OCI and standard setters have realized that additional layers of theory became necessary to explain, for example, reclassification adjustments. Yet, the IASB has made clear in its recent Exposure Draft of a revised conceptual framework that it does not intend to pursue a fresh start in performance reporting that appears to be needed conceptually. Instead, practical considerations, primarily on International Financial Reporting Standards adoption in Japan, seem to lead to another ex-post rationalization of OCI, this time around a conceptually vacuous use of the relevance characteristic.

Keywords:

• other comprehensive income
• IASB
• FASB
• standard setting
• accounting history

1. Introduction

Under both US Generally Accepted Accounting Principles (GAAP) and International Financial Reporting Standards (IFRS), comprehensive income represents the change in an entity's equity during a period that results from transactions and other events from non-owner sources (Statement of Financial Accounting Concepts [SFAC] No. 6, para. 70 and International Accounting Standard [IAS] 1.7). Its principle component is net income, profit or loss under IFRS, which is seen as the primary indicator of a company's financial performance. The second component, other comprehensive income (OCI), has been more difficult to explain and uncertainty abounds about what it represents, given the inability ‘to identify attributes possessed by OCI items under current US GAAP that are not also possessed by other items included in earnings’ (Rees and Shane 2012, p. 811). Up to now, OCI was defined only as the elements of comprehensive income not included in profit or loss, being a repository of items that for one reason or another were excluded from profit or loss.

This study sheds light on the US and international standard setting history surrounding the emergence of OCI, seeking to make the following two points: First, it was the fixation on net income as the primary performance indicator that brought about what is now known as OCI. Attempts to improve the relevance of earnings have been debated already in the 1940s. Yet, it was only in the 1970s that the argument that random fluctuations are not part of an entity's performance gained support and the engineering of what goes into net income began. Once a decision to exclude items from income had been made, a precedent was set for a tailoring of net income, primarily to avoid earnings volatility. Over time, the reluctance to use direct entries to equity waned, as each decision made sense to move forward the respective project, promote agreement among Board members or attend to constituent concerns. That is, the practice was used in projects ‘to make conceptually appropriate changes to the balance sheet', that would have been impossible ‘without the safety valve of OCI'.¹ Taken together, these decisions not only appear ad-hoc, but also made OCI a de facto accepted reporting category. At the same time, they exacerbated the lacking conceptual foundation, as a definition or normative guidance on when to use OCI was never convincingly addressed. Each additional use may have been a twist, bend or adaptation of the previous instance to fit the compromise currently crafted such that the original meaning of the treatment, if there was one, became blurred. When OCI was addressed in 1997 in comprehensive income reporting, it was defined not as a concept, but as a practice and, by doing so, became institutionalized without a clear meaning. The practice advanced to further prominence afterwards and could be referred to by a simple label that specified a journal entry, but did not suggest a definition or provide conceptual guidance.

Second, the debates about the adequacy of direct entries to equity have been replaced by a sense-making of OCI in that its conceptually devoid use required additional layers of theory to design a conceptual basis inductively. These layers relate to whether earnings or comprehensive income measures an entity's performance, whether performance should be presented in one or two statements and whether reclassification adjustments (‘recycling’) are needed. Conceptually strong suggestions to revise the income statement have been made over the years (e.g. Barker 2003, 2004), yet they met with standard setters’ reluctance to embrace comprehensive income as an all-inclusive performance measure. Such a decision is seen to undermine the premise under which OCI was used, namely to exclude items from income that are not part of financial performance. In its current efforts toward a revised conceptual framework, the IASB considerations are exacerbated by the need to incorporate views from its Japanese constituency, where a strong attachment to net income as the main financial performance indicator exists. Hence, the Board has constructed a definition of OCI around current measures being relevant in the balance sheet, but another basis being relevant for determining profit and loss.² These competing notions of relevance seem as descriptive and devoid of conceptual guidance as previous attempts at defining OCI such that the anchoring to the qualitative characteristic appears as another practical ex-post rationalization to accommodate constituents’ objections to a single statement of performance.

In recounting the emergence of OCI, this paper adds to previous attempts to synthesize attributes of the practice based on its current usage (e.g. Barker 2004, Van Cauwenberge and De Beelde 2007, Rees and Shane 2012). Focusing primarily on US standard setting, it begins in Section 2 with the Committee on Accounting Procedure's (CAP) discussions on excluding items from income. Subsequently, three major projects that resulted in OCI accounting are reviewed – financial instruments (Section 3), foreign currency translation (Section 4) and pensions (Section 5). These projects involved conceptual discussions about direct entries to equity and are most pertinent to describe how OCI became institutionalized. As Table 1 shows, OCI was also used in hedge accounting, which began as a practical expedient, but was further developed into today's hedge accounting in 1998.³ The paper then examines the FASB 1997 project on reporting comprehensive income in Section 6, before reviewing the more recent discussions on financial statement presentation, performance reporting and the revised conceptual framework (Section 7). Giving due recognition to the many instances in which OCI requirements under US GAAP were carried forward by the international standard setter (see Table 1), Sections 3–6 also summarize the IAS/IFRS experience with OCI, based primarily on Camfferman and Zeff (2007, 2015). The main exception to the taking over of US requirements related to the revaluation reserve of IAS 16 and IAS 38, which rather had its basis in the accounting practice of some European countries. As the current discussion on OCI is led by the IASB, Section 7 focuses more explicitly on the IASB's deliberations. Section 8 concludes with a discussion of the paper's main arguments.

Table 1. Overview of OCI items under US GAAP and IFRS.

Project	US GAAP	Initial reason for OCI	IAS/IFRS	Influenced by
Financial instruments	SFAS No. 12 (1975)	Industry practice	IAS 39 (1998)	US thinking
	SFAS No. 115 (1993)	–	IFRS 9 (2014)	–
Currency translation	SFAS No. 52 (1981)	Volatility reduction	IAS 21 (1983)	US thinking
Hedge accounting	SFAS No. 80 (1984)	Practical expedient^a	IAS 39 (1998)	US thinking
	SFAS No. 133 (1998)	–	IFRS 9 (2014)	–
Pension accounting	SFAS No. 87 (1985)	Volatility reduction	IAS 19 (1983	US thinking
	SFAS No. 106 (1990)	–	Revised 2004	UK thinking
	SFAS No. 158 (2006)	–	Revised 2011)	IASB thinking
PPE, intangibles	–	–	IAS 16 (1982)	European practice
	–	–	IAS 38 (1998)

^aSee note 3.

2. The CAP and income reporting

The first controversy about excluding items from income occurred in the 1940s, when the CAP, the first US body to establish accounting principles, clashed with the Chief Accountant of the Securities and Exchange Commission (SEC). By then, academics had provided accounting guidance to the SEC, first in ‘A Tentative Statement of Accounting Principles’ (American Accounting Association 1936) and, again, in Paton and Littleton (1940). Both documents advocated an all-inclusive income statement, in which all revenues, expenses, gains, and losses were recorded.

SEC Chief Accountant William W. Werntz, generally supportive of this view, noted that the diversity of accounting practice could not be explained by accounting principles and that accountants ‘have been willing to certify [income] statements, either on the basis that they conform to their own view of the particular matter or on the ground that in the absence of effective criteria they are unable to object’ (Werntz 1945, p. 9). In fact, companies had been charging material non-recurring expenses to surplus, that is, equity, while crediting gains of a similar nature to income (Carey 1948; see also Littleton 1940). As this practice was increasingly frowned upon, Werntz encouraged the CAP to confirm that an income statement was all-inclusive. The Committee, however, hesitated to endorse such a format, but issued Accounting Research Bulletin (ARB) No. 8 in February 1941, in which it called an all-inclusive statement a ‘theoretical ideal’ because charges against earned surplus were ‘from time to time [ … ] a necessary though perhaps debatable feature of accounts'. The CAP did not define acceptable charges against surplus, but discouraged them and, to increase transparency, advocated a combined statement of income and earned surplus in ARB No. 8. Subsequently, income statements were said to increasingly assume an all-inclusive format that awarded full disclosure of all transactions in one place (Carey 1948).⁴

A few years later, the CAP's views must have matured, as the Committee then preferred the income statement to present an entity's operating performance or earnings capacity. This attitude suggested ‘a sharp determination of income’ by direct entries to equity of items unrelated to operating results (Bailey 1948, p. 11). ARB No. 32, issued in December 1947, resulted after long discussions among Committee members and within the audit profession about the nature and form of income (see, for example, 1947 midyear report of the CAP). It included the presumption that all items of profit and loss were included in net income, but, using a non-operating rationale, suggested to exclude ‘items which in the aggregate are materially significant in relation to the company's net income and are clearly not identifiable with or do not result from the usual or typical business operations of the period’ (para. 11).⁵

The contentious nature of ARB No. 32 became apparent by the Bulletin presenting arguments for the two approaches to income reporting. The all-inclusive group argued that annual income statements, added together, should represent total net income. An all-inclusive statement was also easy to prepare compared to the alternative, which was considered open to manipulation both in terms of presenting a larger earnings capacity and using distortion as an argument for income smoothing. ARB No. 32 provoked dissents along these lines from three Committee members, who suggested a two-section format that distinguished an operating and a non-operating section, the sum of which would be reported as ‘net income for the year'. This view was close to the ‘definite position’ taken in Paton and Littleton (1940, p. 102) that ‘all determinants of income in the broadest sense – including unusual and irregular factors – should be reported in the income statement', albeit in two sections with the latter section reporting non-recurring and extraordinary items.

The Committee members in favor of distinguishing operating performance wanted to help users identify unusual and extraordinary items, a task for which companies were better qualified than readers of financial statements. Arguing against an over-reliance on a single income figure, this group pointed to the importance of comparisons between companies and over years, and argued that income should show as precisely as possible the result of what occurred in a specific year. The only thing that appeared to unite the two camps was the need for ‘full disclosure of all material charges or credits of an unusual character’ (ARB No. 32, para. 10).

ARB No. 32 was published in the January 1948 issue of the Journal of Accountancy, which contained two additional articles of interest. The first one was written by George D. Bailey, Chairman of the CAP and senior partner of Touche, Niven, Bailey & Smart. Bailey (1948) acknowledged the advantages of an all-inclusive income statement, but made a case for presenting net income ‘as sharply as possible’ on two grounds: First, operating performance was more useful for judging the future, both when reporting on stewardship to current shareholders and when reporting to prospective investors. Second, such an income statement was more useful to external parties, ‘non-insiders', who were increasingly important as addressees of financial statements.⁶ Since an operating performance statement implied judgment regarding items to be included in income, Bailey (1948) envisioned managers and auditors to assume responsibility alike, albeit with the support of a growing body of accounting bulletins.

The second article of interest was written by SEC Chief Accountant Earle C. King, who shared his predecessor's position that operating performance statements facilitated abuse. He ‘authorized the [SEC] staff to take exception to financial statements which appear to be misleading, even though they reflect the application of Accounting Research Bulletin No. 32’ (King 1948, p. 25). In the journal's editorial, Carey (1948, p. 1) aimed to reconcile the views, praising Bailey's reasoning as ‘simply and forcefully, as a consistent philosophy of financial reporting, clearly related to current needs'. He also expressed sympathy for the SEC's ‘wait and see’ attitude, but made clear his belief that ‘the doubts and fears will prove to be unfounded’ (p. 2). He suggested that there was a trend in accounting that financial reporting increasingly focused on ‘non-insiders'. Companies realized the disadvantage of an all-inclusive statement because investors decided themselves whether income items were recurring or extraordinary. While this judgment may always be exercised by investors, Carey (1948) considered it a substantial task and viewed positively the CAP's decision to help readers of financial statements judge transactions via operating performance. He suggested that the CAP's approach to require full disclosure was enough of a safeguard against abuse.

This positive note was offset in 1950, when the SEC proposed amendments to Regulation S-X to make all-inclusive income mandatory. Following protests from the CAP, a special committee was set up and a compromise was crafted: Special items could be excluded from net income, but had to be shown at the bottom of the income statement, followed by the figure ‘net income or loss and special items'.⁷ This presentation awarded items excluded from income much transparency, but did not solve the problem of determining which non-recurring gains and losses could be called ‘special items’ and excluded from income. Thus, the controversy shifted to the question where in the income statement certain items should be presented, and wide divergence existed in the reporting of these items (Barr 1965). The income statement itself, reflecting the application of amended Regulation S-X, increasingly took an all-inclusive format that separated ‘net income before extraordinary items’ from ‘net income for the year’ (Grady 1965). It was thus only a small step for the CAP's successor, the Accounting Principles Board (APB), to embrace all-inclusive income in December 1966, when it issued Opinion 9, excluding only prior period adjustments from income. Extraordinary items could still be presented separately, but became part of ‘net income'. This approach was reaffirmed by the APB in Opinion 20 of 1971 and Opinion 23 of 1973, and carried forward by the FASB in Statement of Financial Accounting Standards (SFAS) No. 16 of 1977.

In summary, this early period demonstrates that, absent pre-existing bulletins, opinions or standards, a promising conceptual debate on the determination and presentation of income was held. This debate was led in modern terms, revolving around the usefulness of financial statements to external parties, to be judged in terms of the stewardship of current and prospective investors. Accordingly, it seemed only a matter of time for the income statement to follow an operating performance approach. Accountants and auditors would have had substantial discretion in determining what to include in income. Proponents of operating performance considered this discretion beneficial to serve users’ needs, whereas the SEC's Chief Accountants Werntz and King saw it as facilitating accounting abuse. Likely, their attitude resulted from the widespread and diverse use of direct entries to equity in practice, often exploited to show a more favorable income figure. In the end, the SEC intervened in 1950 and enacted the more rigid all-inclusive income statement. The operating performance approach was thus never made operational, not the least because the controversy about which items to exclude from income loomed too large. What remained unclear throughout the debate was whether items should be excluded due to their extraordinary nature or because they were not recurring, and how often they could return before being labeled ‘non-recurring’ or, in the APB's terms, ‘special items'.⁸ It may also be of interest that the episode considerably antedates the current debate on a single statement versus two statement presentation of income, including the discussion of having a page break between the income statement and the statement of OCI.

3. The FASB and marketable securities

The first item excluded from income by an accounting pronouncement related to marketable securities, bringing about a direct entry to equity in 1975. In the pre-FASB era, ARB No. 43 (Chapter 3A, para. 9) provided some guidance for the diverse practice on marketable securities and required lower-of-cost-or-market valuation for securities classified as current assets if the decline was not temporary. Value decreases were included in income, while write-ups of previous write-downs were not addressed.

In 1973 and 1974, stock prices declined sharply and companies wrote down their securities. When the market recovered, carrying amounts were well below cost or market value and the Screening Committee on Emerging Problems asked the FASB to provide guidance ‘as soon as possible’ (FASB Status Report, No. 28, 9 September 1975, p. 2). To complete the project swiftly, the Board did not issue a Discussion Memorandum and kept the project narrow in scope, only defining when marketable securities should be written down and whether write-ups of previous write-downs should be allowed.

Following a 6 November 1975 Exposure Draft and a public hearing in December, SFAS No. 12 was issued in December 1975 with a 5–2 vote.⁹ The statement required companies to assess the value of their marketable securities on a portfolio basis with the negative difference between aggregate cost and market value representing a valuation allowance. For current assets, changes in the valuation allowance were included in income, whereas changes relating to the non-current portfolio were recorded in equity (para. 11). In light of the project's urgency, the Board adopted an unprincipled solution and, for the same reason, refrained from a conceptual clarification of the recognition of unrealized gains and losses (para. 30). While precedents for direct entries to equity existed in accounting practice, in particular of insurance companies (Kirk 1989), the Board had created a standard setting precedent for excluding items from income. Grinnell and Norgaard (1980) noted that the large majority of marketable securities covered by the statement were non-current and conjectured that the standard ‘was designed in anticipation of the adoption of current value as an integral part of the financial reporting system’ (p. 72).

It took another decade, however, until current values for marketable securities gained more acceptance. When a large number of thrift institutions went bankrupt in the second half of the 1980s, accounting was cited as a contributing factor to the severity of the ensuing savings and loan crisis. Soon, a wave of support developed for applying current values to investments in securities. The SEC, long a bastion of historical cost accounting, ‘encouraged, some might say dictated, the use of market-based measures to value certain debt securities’ (Wyatt 1991, p. 80). SEC Chairman Richard C. Breeden testified on the savings and loan crisis before the Senate Committee on Banking, Housing and Urban Affairs on 10 September 1990 and ‘urged [the FASB] in the strongest way to adopt a market value approach’ (Beresford 1993, p. 75). Breeden (1990) argued that financial institutions were in the business of actively managing their asset and liability portfolios, which involved a frequent trading of securities. In his view, historical cost accounting inappropriately depicted that environment such that he called for the use of ‘market-based measures of valuation at the earliest possible date’ (Breeden 1990, p. 32).¹⁰

The FASB moved quickly to consider the SEC's views in its project on financial instruments, added to the agenda in 1986. The Board, spending almost half of its time in 1991 on the project, soon intended to ‘require market-based measures for some financial assets and permit the option of using market-based measures for related liabilities’ (Johnson and Swieringa 1996a, p. 161). It was debated, however, which securities were to be measured at fair value and whether holding gains and losses should be recognized in income. At one point, it was agreed that some holding gains would be excluded from income, while, one month later, narrow support was found for including them in earnings.¹¹

This support waned in the first half of 1992, as Board members, under pressure from financial institutions, wanted to revisit conclusions reached earlier (Berton 1992, Zeff 2002). In July, the FASB's Director of Research and Technical Activities, Timothy Lucas, proposed a compromise solution that assigned marketable securities to one of three categories (Johnson and Swieringa 1996a). If an entity had the intent and ability to hold debt securities to maturity, the instruments would be carried at amortized cost in the ‘held for investment’ category. Without such an intent or ability, securities would be classified as ‘held for possible sale’ and carried at fair value with holding gains and losses excluded from earnings. The third category, ‘trading', would also be carried at fair value, but holding gains and losses would be recognized in income. Following this breakthrough, an Exposure Draft was published in September 1992, which nonetheless provoked strong criticism: 70% of the comment letters came from financial institutions, who worried that the standard would make capital more volatile (Johnson and Swieringa 1996a). Yet, the Board upheld its conclusions and, in May 1993, issued SFAS No. 115 by a 5–2 vote.¹²

In summary, the exclusion from income of the gains and losses resulting from the remeasurement of marketable securities was brought about by a larger emphasis on the balance sheet. The movement toward fair valuation had been portrayed as the outcome of an ‘evolutionary development’ (Wyatt 1991, p. 83) or a ‘sea change in finance’ that accounting had to keep up with (Johnson and Swieringa 1996a, p. 165). Despite this notion of inevitability, the Board faced considerable opposition throughout the project, being ‘caught in a crossfire’ between the SEC and other regulators (Van Riper 1994, p. 146; see also The Economist 1992, Worthy 1992). The preparer community was also divided: Kinney (1996, p. 182) reported that, ‘as financial statement users, some banks favored mark-to-market for other financial statement preparers, but not for the banks themselves as preparers'. The overall result of the project was a ‘patchwork solution’ (Johnson and Swieringa 1996a) that expanded considerably the recognition of items outside income, using a mixture of measurement and realization models as a theoretical justification, which seemed to have some appeal, but continued to lack a conceptual basis.

3.1. The International Accounting Standards Committee (IASC) experience with financial instruments

The IASC project on financial instruments began in 1988, quickly turning into the Committee's most challenging assignment (Camfferman and Zeff 2007, pp. 361–77). Also inspired by the expansion of finance in the 1980s, the IASC followed the FASB in devoting itself to a standard on financial instruments. After a slow start, the IASC's steering committee increased its pace, wanting to have an all-encompassing financial instrument standard by 1993 in an attempt to take a leadership role in the area. In its 1991 Exposure Draft E40, the Committee proposed a mixed-measurement model with historical cost as the benchmark treatment and financial instruments being assigned to categories based on management intent. This draft defined the broad outlines of the subsequent debate, where agreement on the general direction existed, but opinions differed on the details of a standard.

As the FASB issued its SFAS No. 115, voices from the US became stronger that the IASC should move more decidedly toward fair valuing and rely less on management intent. While other respondents to the draft also expressed criticism, the IASC intended to move ahead with E40. With a considerable number of amendments and increasing opposition, the Committee approved a revised Exposure Draft E48 in early 1994, which met with an even more negative reaction. After reluctantly agreeing to divide the project in several parts, the IASC approved its disclosure and presentation standard IAS 32 quickly afterwards in March 1995.

The remainder of the project was then re-organized with a new steering committee that shelved the approach of E48, intending to study the remaining issues fundamentally with a view to developing a discussion paper first. This March 1997 paper contained fair value as the basic measurement principle with recognition of unrealized gains and losses in income. While this approach gained support from the US, the IASC realized that it would bring strong criticism from preparers and that the Committee might be unable to complete the project prior to the International Organization of Securities Commissions’ (IOSCO) deadline for completing a set of ‘core standards’ by April 1998.

Given these considerations, the IASC abandoned work on its own financial instrument standard and dissolved the steering committee. Instead, Committee staff began to develop the remainder of the standard based on US requirements, which, barring some editorial changes, were used in a new draft standard. In spite of the public outrage at this change of course and the ensuing reputational damage, the IASC largely continued on this path in a revised proposal E62, which, in December 1998, made its way into the recognition and measurement standard IAS 39. Defying attempts to change key requirements, the proposal was supported only by the minimum number of 12 votes, with the decision fostered by IOSCO's revised deadline for the ‘core standards’, which had been moved to the end of the year. Hence, the financial instrument standard reflected some ‘messy compromises’ (Camfferman and Zeff 2007, p. 13), but took over key requirements of SFAS No. 115, including the recognition in equity of unrealized gains and losses on available-for-sale securities.¹³

4. The FASB and foreign currency translation

Interspersed with the FASB's experience with marketable securities was the project on foreign currency translation, which was first covered in SFAS No. 8, issued in 1975. Responding to strong criticism of this standard, the FASB reconsidered its approach and issued SFAS No. 52 in 1981, introducing a further disruption of all-inclusive income.¹⁴

Following the demise of the Bretton Woods system in 1971, companies had to cope with floating exchange rates in an increasingly international business environment. Accounting practice was very diverse such that the FASB sought to create uniform guidance for both foreign currency transactions and the translation of foreign currency financial statements. Its debates were firmly grounded in pre-existing conceptual premises, that is, a historical cost framework and an all-inclusive income statement, and largely followed a research study by the American Institute of Certified Public Accountants, which argued that the translation process should not change the measurement of an asset or a liability (Lorensen 1972). The proposed ‘temporal method’ suggested translation of assets measured at historical cost at a historical exchange rate and translation of items measured at current values at a current exchange rate. Translation adjustments would be recorded in the income statement.

The FASB adopted this approach in SFAS No. 8, which carried one dissenting vote that foresaw practical problems with the standard because exchange rate fluctuations would make net income more volatile. Indeed, practitioners and academics soon criticized the standard heavily, claiming that it encouraged uneconomic actions because of companies’ strong focus on the bottom line. Companies were said to pursue more aggressive risk management strategies to avoid earnings volatility (Evans et al. 1978).

In the face of vast criticism and repeated requests to change the standard (Van Riper 1994), the FASB eventually decided to reconsider SFAS No. 8 in 1979. It concluded quickly that translation adjustments should not be part of income from continuing operations, but reported either as an outside earnings or as a non-operating item (FASB Status Report, No. 89, 6 July 1979).

When the FASB issued SFAS No. 52 in 1981, it introduced the ‘functional currency method', which required current exchange rates for the translation of foreign currency financial statements and translation adjustments to be recorded in stockholders’ equity.¹⁵ The Board noted the most frequently made recommendations from its constituents and explained how it had responded to them – a clear sign that it was willing to listen to its critics and change its mind on foreign currency translation. This attitude also implied that, in the case of foreign currency translation, compromises were more important than a conceptually pure approach.

Still, there was considerable disagreement among Board members, as indicated by the narrow 4–3 vote on SFAS No. 52. The four assenters had two different views of translation adjustments: one group thought they reflected the economic effects of exchange rate changes, while the other considered them a mechanical byproduct of the translation process. The three dissenters, including the Board's Chairman, saw SFAS No. 52 as a rejection of pre-existing fundamental concepts that would reverberate beyond the foreign currency translation standard. They criticized the remeasurement of items while being held and the recording of translation adjustments in equity. Such a deferral of income recognition was said to stand in contrast to previously applied concepts of realization.¹⁶

The Board did not favor including the separate component of equity in income at some point, because the adjustments had been included in the then-emerging concept of comprehensive income (first group, see following paragraphs) or because they were seen as an adjustment to equity (second group). Yet, it was decided to use what is now called ‘recycling’ by including translation adjustments in income as part of the net gain or loss on sale or liquidation of the underlying investment (para. 14). While the information was thought ‘probably marginal', it was argued that, upon sale or liquidation, the unrealized translation adjustments would be realized. Reference was also made to the argument that non-owner transactions that change equity need to be recognized in income at some point (para. 119). The FASB considered its conclusion ‘desirable until the concepts of reporting all components of comprehensive income are further developed’ (para. 119). That is, pending decisions on the underlying income concept, the treatment of translation adjustments could be revised.

Board members’ ability to compromise was hence fostered by the concomitant conceptual framework project, during which the reporting of earnings and presentation of the income statement was also debated. Responding to user concerns that all-inclusive income contained too much noise and was not necessarily informative about earning power, the FASB explored the presentation of income in a multiple-step format to reduce the emphasis on net income. It hesitantly moved in that direction by introducing a new income concept, namely comprehensive income, in SFAC No. 3, issued in 1980. This concept would be all-inclusive, that is, it would represent the total change in equity during a period resulting from transactions and events from non-owner sources. While it was envisioned to display intermediate components as subtotals, comprehensive income did not emerge from SFAC No. 3 as a well-defined concept. Some Board members opposed the abandoning of earnings as a performance indicator and the FASB had been indecisive about its preferred method of capital maintenance.

Less than a month prior to SFAS No. 52, on 16 November 1981, the FASB issued an Exposure Draft on ‘Reporting Income, Cash Flows, and Financial Position of Business Enterprises', in which a revised income statement was discussed in broad terms. The Board again suggested that income could be presented in a layered format with subtotals for operating activities, discontinued operations, peripheral or incidental activities, and price changes.¹⁷ This presentation would shield ‘core earnings’ from the effect of remeasurements and, possibly, accommodate translation adjustments in the wider concept of income.

When the Board could not agree on later phases of the conceptual framework, in particular recognition and measurement, it shelved its project on reporting income, but incorporated some of the considerations in Concepts Statement No. 5, issued in 1984. It was suggested that perceptions about realizability and volatility explained why some items, such as translation adjustments, were excluded from income (SFAC No. 5, para. 46). Comprehensive income, including a statement of comprehensive income, was upheld in SFAC No. 5, but neither was put to use such that Anthony (1987, p. 76) commented: ‘Comprehensive income is a meaningless term. I have not seen it before and I doubt that I will see it again.’

In summary, SFAS No. 52 introduced remeasurements to the balance sheet by requiring the use of current exchange rates for the translation of foreign currency financial statements. Such an accounting would have made income vulnerable to exchange rate fluctuations. Protecting the income statement from volatility was hence a paramount concern of SFAS No. 52 and was made possible by the unprincipled solution of recording translation adjustments in equity. This undermining of net income as an all-inclusive measure could have been healed by the embracing of an alternative all-encompassing figure, namely comprehensive income. Yet, the conceptual framework project came to an early halt such that comprehensive income did not advance beyond an abstract concept that merely showed the Board's vision of a revised income statement. While SFAC No. 5 gave some conceptual cover to the recording of translation adjustments in equity, the compromise solution of SFAS No. 52 did have consequences beyond the foreign currency translation project, as the dissenters to Statement No. 52 predicted. In particular, reclassification adjustments had been upheld, although they had entered SFAS No. 52 in spite of Board members’ initial views and only pending a conclusion on comprehensive income. Indecision thus introduced an unprincipled practice that standard setters have debated time and again. Likewise, the lack of guidance on when to use entries to equity suggested that subsequent decisions on income would not have a conceptual basis either.

4.1. The IASC experience with foreign currency translation

In January 1974, the IASC set up a steering committee to address the translation of foreign accounts (Camfferman and Zeff 2007, pp. 119–23). It was decided quickly that transactions in foreign currency would be converted at closing rates with any resulting differences taken to income. Yet, the steering committee was unable to take a position on the translation of non-current balance sheet items and foreign subsidiaries’ financial statements, where it broadly considered similar methods used by the FASB in the US, that is, something close to the temporal method or translation at the current rate. As it was also unable to decide on the treatment of translation differences, the committee referred the topic back to the IASC, which continued the debate in vain between 1974 and 1977. The resulting Exposure Draft E11, approved in July 1977, left open key issues of foreign currency translation and provided only little guidance on how to resolve the choices allowed by the draft.

As E11, along with constituent responses, was considered by the steering committee, the FASB began to debate its SFAS No. 8, quickly focusing on the outlines of what became SFAS No. 52. The FASB's task force included representatives from the UK, Canada and the IASC, which seemed to give the steering committee an incentive to follow the project closely. The resulting revision of the exposure draft bore strong resemblance to the US approach and hence met with protests from France and Germany who perceived an Anglo-Saxon bias in the foreign currency translation standard, which they claimed to have emerged outside of the IASC's normal procedures. Yet, once these procedural complaints were resolved, the two countries also supported the content of the revised standard. The IASC waited until the FASB was about to issue SFAS No. 52, until it approved its own E23 in October 1981 and IAS 21 in March 1983, requiring translation differences to be reported in equity and thus proposing a treatment ‘largely in agreement with [SFAS No. 52]’ (Camfferman and Zeff 2007, p. 123).¹⁸

5. The FASB and pension accounting

Soon after direct entries to equity were used in foreign currency translation, another project excluded further items from net income: The pensions project ran for more than a decade during which companies repeatedly asked the Board to smooth the effect of a pensions standard on their financial statements (e.g. Miller and Redding 1992, Van Riper 1994, Miller et al. 1998). As will be shown below, the FASB gradually made compromises and, at the end of the struggle, issued SFAS No. 87 in 1985.¹⁹

Prior to the FASB, the APB had addressed itself to pension accounting and issued Opinion No. 8 in 1966, which focused on the measurement of periodic pension expense, but granted considerable discretion when determining this expense. In 1974, the Employee Retirement Income Security Act (ERISA) raised the public's awareness of pensions by making pension claims enforceable under law (e.g. Regan 1976). ERISA extended plan participants’ legal claims on a company's pension assets to up to 30% of the plan sponsor's net worth. Hence, the question arose whether a company's unfunded pension obligation should be recognized, particularly since the large amounts of unfunded plans had attracted the attention of newspapers and the public (Lucas and Hollowell 1981).

In light of the publicity awarded to pension benefits, the FASB added two projects to its agenda in 1974. The first concerned the financial statements of employee benefit plans and resulted in SFAS No. 35, issued in 1980.²⁰ The second project concerned employers’ accounting for pensions, which, it was hoped, could be informed by the conceptual framework.

The FASB published a background paper in March 1980 and a discussion memorandum in February 1981, analyzing the underlying issues of pension accounting and possible reporting requirements. In November 1982, the Board published its tentative conclusions as ‘Preliminary Views’ and a second discussion memorandum followed in April 1983. It became clear that the FASB advocated recognition of a net pension liability for defined benefit plans, measured as (i) the pension benefit obligation, based on the terms of the pension plan and current estimates about future salary levels; (ii) less the fair value of plan assets; (iii) plus or minus a measurement valuation allowance, a buffer item included in the net pension liability to absorb measurement changes in the obligation and plan assets. The allowance as well as its amortization over plan participants’ remaining years of service was to reduce measurement volatility resulting from the estimates of future events as well as the fair valuation of plan assets.²¹

The FASB received more than 500 responses on these documents, facing opposition from preparers and seven of the Big Eight accounting firms (Berton 1984). Besides the recognition of a liability, seen as detrimental to companies’ borrowing capacity, respondents criticized the use of future salary levels in the measurement of the obligation and the fair valuation of plan assets, both of which were seen as major sources of volatility. Companies were stirred into action, beginning an ‘extended probing about the sensitive project', exerting pressure on the Board and attacking the FASB's modus operandi (Van Riper 1994, pp. 119–20).²² To avoid the ‘great harm to the business community’ that the proposals were said to bring, it was repeatedly suggested that the Board change its course.

Against the background of a vocal constituency, the FASB published an Exposure Draft in March 1985 and, after considering more than 400 comment letters, it issued SFAS No. 87 in December 1985 by a 4–3 vote.²³ The standard showed that the Board had responded to the criticism by making considerable compromises over the years: The minimum pension liability was based on current, not future salary levels, thus representing an accumulated benefit obligation. Several procedures were installed to limit the volatility of earnings that the measurement process would bring about, such as using the expected, instead of the actual, rate of return on plan assets, and the corridor approach to amortize unexpected gains or losses on plan assets (e.g. Miller 1987). A further smoothing mechanism concerned the recognition of an additional pension liability, required if the recognized liability was smaller than the difference between the accumulated benefit obligation and the fair value of plan assets. To the extent that the additional liability did not exceed unrecognized prior service costs, an intangible asset was created. Any excess of the additional minimum liability over prior service costs was recorded in equity to protect the income statement against an additional pension charge.

The Board acknowledged that a conceptually appropriate solution would have followed a different path and that gains and losses should be recognized without delay, but ‘those approaches would be too great a change from past practice to be adopted at the present time’ (SFAS No. 87, para. 107). Accordingly, the Board

continues the evolutionary search for more meaningful and more useful pension accounting. The FASB believes that the conclusions it has reached are a worthwhile and significant step in that direction, but it also believes that those conclusions are not likely to be the final step in that evolution. Pension accounting in 1985 is still in a transitional stage. (para. 5)

Hence, the standard's measures to smooth income appeared as compromises, required to make progress and obtain agreement on the project's key objective, the recognition of a pension benefit obligation.²⁴

It was only in 2005 that the FASB took the next step in pension accounting. The Sarbanes-Oxley Act of 2002 had mandated the SEC to study off-balance sheet arrangements and, among other things, the Commission criticized pension accounting for allowing preparers to report fewer liabilities. The Securities and Exchange Commission (2005, pp. 107–8) requested a review of the accounting for defined benefit arrangements to require consolidation of pension plans, eliminate smoothing mechanisms and revisit the valuation of plan assets.

The FASB split its subsequent project into two phases: It would first address the balance sheet presentation of pension accounting, before reconsidering the topic fundamentally, possibly together with the IASB. The first phase was swiftly completed with an Exposure Draft in March 2006 and SFAS No. 158 in September 2006. The standard carried forward the recognition of a pension asset or liability on a net basis and did not change the determination of periodic pension cost. However, the balance sheet was to show the funded status of the pension plan, measured as the difference between the fair value of the plan assets and the benefit obligation. The latter included estimates of future salaries and was based on the projected benefit obligation, which in SFAS No. 87 had been considered conceptually preferable. Any unrecognized gains and losses and prior service costs were included in what by then was known as OCI, that is, the income and expense items excluded from net income (see Section 6). The OCI charges would then be reclassified to the income statement when recognized as part of periodic benefit cost.

SFAS No. 158 aligned pension accounting more closely with the element definitions of the conceptual framework and simplified requirements by abolishing the recognition of an intangible asset and a minimum pension liability. At the same time, it expanded the use of OCI because the FASB considered this form of recognition consistent with the earlier approach to minimum pension liability adjustments (SFAS No. 158, paras B36, B41). The project again demonstrated that a conceptually appropriate recognition of the pension plan in the balance sheet would make the income statement more volatile. Since these fluctuations may not reflect current entity performance, OCI was used once again to protect the income statement.

In summary, the FASB's early approach to pension accounting was based on the element definitions of the conceptual framework, with a preference to recognize the best estimate of the pension obligation (Lucas and Hollowell 1981). This conceptual approach drew strong opposition so that ‘the Board invented techniques to arbitrarily smooth out income and defer the effects of changing to the new methods’ (Miller et al. 1998, p. 136). In its Basis for Conclusions, the FASB explained at length how it had addressed concerns of volatility, admitting that ‘both the extent of volatility reduction and the mechanism adopted to effect it are essentially practical issues without conceptual basis’ (SFAS No. 87, para. 177). Once a decision was made to reduce volatility and delay recognition of some gains and losses, no second-best solution was said to exist (para. 186). Accordingly, reporting requirements seemed to arise incidentally, that is, in the course of finding mechanisms to reduce income volatility. Recording minimum pension liability adjustments in equity ‘was only the tail end of the compromises’ used in SFAS No. 87.²⁵ When the FASB revisited pension accounting in 2006, it expanded the use of direct entries to equity and required reclassification of the amounts recorded in equity. Its argument that this approach was consistent with SFAS No. 87 disregards that the initial use of OCI had come about in the absence of conceptual arguments. SFAS No. 158 thus expanded a treatment that had made its entrance to pension accounting as a compromise.

5.1. The IASC experience with pension accounting

The IASC began its project on retirement benefits in 1977, at a time when only the US had accounting requirements on pensions, albeit limited to calculating periodic pension expense (Camfferman and Zeff 2007, pp. 129–30). Cooperating with national bodies, the Committee approved IAS 19 in 1983. The standard distinguished between defined benefit and defined contribution plans, but focused on the pension expense, which was based on current and prior service costs and the employer's plan contribution, respectively. It did not require recognition of an unfunded pension obligation, nor did it provide guidance on actuarial assumptions, but conflicted with practice in many European countries by prohibiting commonly used methods to determine periodic pension costs.

The topic was brought to the fore again in 1994, as a revised IAS 19 was to be part of the IASC's set of ‘core standards’ and IOSCO encouraged modification of IAS 19 along the lines of SFAS No. 87 (Camfferman and Zeff 2007, pp. 395–7). This suggestion implied recognition of a liability for underfunded benefit plans and additional guidance on the defined benefit obligation. The revised version of IAS 19, approved in early 1998, indeed resembled the FASB's accounting rules closely. Yet, while the key elements of SFAS No. 87 were adopted, including the corridor approach, the revised standard did not require any OCI treatment.

This step came only in 2004, when the IASB converged its standard with recent UK requirements, where actuarial gains and losses had to be recognized immediately (Camfferman and Zeff 2015, pp. 135–7). As companies prepared for IFRS adoption, the UK Accounting Standards Board (ASB) feared that this key rule would be lost and it pressed the IASB to amend the pension standard. Accordingly, IAS 19 was revised to allow earlier recognition of actuarial gains and losses than required by the corridor approach, with the option of doing so in OCI.²⁶

Another change to pension accounting came in 2011, when the IASB concluded a further revision of IAS 19 as part of its Memorandum of Understanding with the FASB (Camfferman and Zeff 2015, pp. 569–73). During the project, the majority of Board members, aware of preparers’ opposition, favored recognition in OCI of the items resulting from the remeasurement of plan assets and the defined benefit liability. Without convincingly arguing for this approach, the Board suggested that the improved presentation of OCI items due to a revised IAS 1 (see Sections 6 and 7) justified recognition in OCI, while declaring that a revised conceptual framework could suggest changes to this presentation (BC96–7). This revision of IAS 19 hence abandoned the corridor approach and expanded recognition outside profit or loss without any recycling, and thus went against converging IFRS pension accounting with US requirements.

6. The FASB and reporting comprehensive income

After direct entries to equity had been introduced and expanded in the projects described above, calls were made for the FASB to make use of comprehensive income to accommodate items recognized outside net income and achieve a transparent display of earnings (Robinson 1991, Sutton and Johnson 1993). Users similarly encouraged the FASB to review the reporting of income:

We have profound misgivings about the increasing number of wealth changes that elude disclosure on the income statement. Yet, individual items may be interpreted differently. That calls for the display of comprehensive income that allows components of different character to be seen and evaluated separately. (Association for Investment Management and Research [AIMR] 1993, p. 63)

Hence, the Board should now ‘move comprehensive income from concept to application’ (AIMR 1993, p. 88).

Citing these concerns as cause for action, the FASB began to address comprehensive income reporting in September 1995. It first considered a broad scope for the project to provide the bypassing of income with a conceptual basis and impose discipline on future decisions (Johnson and Swieringa 1996b). Such a scope implied a revisiting of the most controversial issues in the conceptual framework, that is, recognition and measurement, such that the Board decided on a narrower scope, the reporting of comprehensive income.²⁷ The resulting standard would thus focus on presentation questions without providing a conceptual basis for current practice and future standard setting decisions. Put differently, the scope decision transformed a conceptual controversy into a presentation issue that could be addressed more readily, suggesting in turn that the question of whether OCI was a recognition and measurement issue or a mere presentational matter was ambiguous as well.

On 20 June 1996, the Board published an Exposure Draft which proposed three major issues: (i) a comprehensive earnings per share figure, (ii) reclassification of items recorded in equity, and (iii) a display of comprehensive income in one or two statements of financial performance, that is, an option to show net income as a subtotal in one income statement or to display two income statements and begin the second one with net income.²⁸ The Exposure Draft used the term ‘OCI’ for the first time, simply meaning all items of comprehensive income that were excluded from net income.

The proposals drew strong responses that explained what respondents believed OCI should not be:²⁹ Primarily, it was asserted that comprehensive income introduced another bottom line to financial statements. This second measure would confuse, or mislead, users as to the appropriate measure for enterprise performance, since the artificial volatility of comprehensive income was not indicative of a manager's ability to run a company. The Exposure Draft was also said to undermine existing OCI requirements because these standards had been issued under the premise that items would be excluded from income and, hence, any performance measure. Items included in OCI could not be considered income because the earnings process was not complete and the items were not yet realized or may never realize. Respondents also suggested that OCI was on display already because the items were reported directly in equity.

Addressing these comments, the FASB made the reporting of comprehensive earnings per share optional and added a third alternative to display OCI, namely in the statement of changes in equity. SFAS No. 130, issued in June 1997 by a vote of 5–2, also affirmed the practice of reclassification adjustments.³⁰ Attempting to use theory to rationalize OCI, it was argued that, without these adjustments, items would be double counted in comprehensive income: once when recorded in OCI and again when included in net income (para. 18). This reasoning showed that the Board did not apply comprehensive income coherently and that it had turned around the argument used in SFAS No. 52 regarding translation adjustments, that is, if items were recorded in comprehensive income via OCI, there was no need to include them in income again at a later point.

The two dissenters argued that items of OCI were conceptually not different from items reported in net income, which they saw as support for reporting both components with equal prominence. They believed that ‘the Board inappropriately failed to respond to the clear and unequivocal call from users of financial statements for the transparent presentation of all items of comprehensive income'. The Board had acknowledged ‘the conceptual superiority of displaying comprehensive income in a statement of performance', but had retreated from this position.

In summary, SFAS No. 130 provides further evidence for the pragmatic character of OCI. To make transparent an unconventional practice, the concept of comprehensive income was unearthed after more than a decade during which it had sunk into oblivion. It represented a missing piece that fell into place to harmonize accounting standards with the conceptual framework. The FASB's project managers admitted that ‘comprehensive income is no doubt a term that is unfamiliar to many’ (Johnson and Reither 1995), suggesting that its sole purpose was to give OCI more prominence. From a reporting perspective, the project achieved its objective and brought about transparency. However, by labeling items excluded from income as OCI, it left aside conceptual considerations and created a definitional dilemma, exacerbating the difficult sense-making of current accounting practice. That is, SFAS No. 130 defined OCI accounting, not as a concept, but as a practice and, by doing so, institutionalized it without a clear meaning.

6.1. The IASC experience with comprehensive income

The IASC's standard on financial statement presentation IAS 1 had been issued in January 1975, accepting a wide range of accounting practices in presenting an entity's financial position and the results of the period (Camfferman and Zeff 2007, pp. 390–3). IAS 1 was addressed again in 1993, at a time when both the FASB and the ASB were considering the topic. The IASC also perceived the expansion of direct entries to equity as an incentive to make this practice more transparent, perhaps in a statement of comprehensive income. Hence, it proposed such a second performance statement, termed ‘statement of non-owner movements in equity’, in E53, approved in June 1996. Facing strong opposition on these proposals, the IASC compromised by only requiring a statement of changes in equity in the 1997 version of IAS 1. As an alternative presentation, ‘items directly recognized in equity’, as elements of OCI were still known, could be shown in a statement of recognized gains and losses, which was inspired by the ASB's terminology. Hence, the IASC experienced similar opposition to a comprehensive income statement as had the FASB in the US. At the same time, indecision about the presentation of these items became clear during the revision of IAS 1 and studies on performance reporting were begun by the G4+1 standard setters. As will be shown in the following section, financial statement presentation re-appeared soon after the IASC had made the transformation into the IASB.

7. FASB and IASB: financial statement presentation and conceptual framework

In late 2001, both FASB and IASB decided to revisit performance reporting and added a project to their respective agendas such that each Board developed its own reporting model: The FASB wanted to retain comprehensive income, but proposed an ordering of the income statement based on a business, financing and a non-business/non-financing category, to be followed by OCI. In that sense, the FASB put less emphasis on net income as a subtotal via presentation of comprehensive income in a continuous income statement. The IASB, adopting the term ‘comprehensive income’ during the project, also subscribed to the idea of a single performance statement, but, unlike its counterpart, wanted to eliminate OCI and recycling.³¹ In cooperation with the ASB, it went on to suggest a matrix format of income presentation, a three column disaggregated format, where business and finance income would be shown on a historical cost basis and a separate column would depict remeasurements (Barker 2003, 2004). This approach brought forth both political and technical difficulties (Camfferman and Zeff 2015, Section 5.5.2): feedback from Japanese companies and the Accounting Standards Board of Japan (ASBJ) suggested strong disapproval of the proposed format, which they saw as an attempt to move away from net income. Likewise, there were unresolved technical questions regarding the classification of issues such as inventory write-downs.³²

Both Boards obtained further critical comments about their performance reporting models during field visits, and it was repeatedly suggested that they should coordinate their work. In 2004, the Boards began a joint project that was expanded into a fundamental reconsideration of financial statement presentation in three phases: The first phase would discuss a complete set of financial statements, the second phase would address financial statement presentation and phase three was to revisit interim financial reporting.

While the FASB decided later to consider the first two phases together, the IASB issued an Exposure Draft on Phase A in March 2006 and a revised IAS 1 in September 2007. Among other changes to bring IAS 1 in line with SFAS No. 130, the IASB made a distinction between owner and non-owner changes in equity. The former were reported in the statement of changes in equity and the latter in the statement of comprehensive income, presented in one or two statements. Initially, the IASB had preferred a continuous income statement, but it acceded to the strong opposition from its constituents who feared that such a presentation gave undue prominence to comprehensive income (BC52).³³ The revised IAS 1 also contained a definition of reclassification adjustments (para. 7) and required enhanced disclosure of reclassified amounts for each item of OCI either in the statement of comprehensive income or in the notes (para. 90).

The subsequent Phase B involved a fundamental rethinking of financial statement presentation, as became evident in a joint Discussion Paper of October 2008. The Boards proposed a management approach to financial statements, which arranged assets and liabilities according to an entity's business and financing activities (2.19). ‘Cohesiveness’ was to become the guiding principle of financial statement presentation, such that comprehensive income would also be structured in a business and financing section, albeit in one continuous display (3.24). For all items of OCI except some currency translation adjustments, an entity would indicate to which category the item related (3.25) and would reclassify these items into the same section as the asset or liability generating the underlying income or expense (3.26).

The Boards’ resolution to follow a ‘cohesiveness’ principle had earlier in the project led to a questioning of OCI. In an October 2006 meeting, the staff referred to the lack of a common conceptual basis for OCI items and proposed that ‘cohesiveness’ meant assigning these items to either the business or the finance section of the income statement, that is, to eliminate both OCI and recycling (Agenda Paper 10/2006, paper 6C). This conclusion implied that net income would not be presented as a subtotal and, thus, as a performance indicator. In spite of offering a principled solution to income reporting, the staff's recommendations were not pursued further. Japanese companies and the ASBJ had supported strongly the case for retaining net income throughout the projects on performance reporting and IASB member Tatsumi Yamada endorsed the concerns from his home country (Camfferman and Zeff 2015; also Whittington 2005). As a compromise, the IASB decided to make the elimination of OCI a long-term objective, but retain it for the time being. Yet, the 2008 Discussion Paper did not include a reference to this approach, likely in view of the imminent decision by Japan to adopt IFRS (Camfferman and Zeff 2015).

While the Discussion Paper was developed further in a July 2010 Staff Draft, the Boards published a separate Exposure Draft on OCI in May 2010.³⁴ This step was considered necessary due to the projects on financial instruments (jointly) and pension accounting (IASB only), which were expected to make further use of OCI. In their Drafts, the Boards made another attempt at presenting comprehensive income in a single statement and at giving reclassification adjustments more prominence. However, in their revised standards of June 2011, the Boards retreated again from a single statement because respondents had once more fretted over potential confusion about the bottom line, the lacking relation of OCI to core business activities and the lack of management control over OCI items. Nonetheless, the Boards considerably constrained the two-statement format by requiring consecutive presentation of income and OCI. This decision made the difference between the two formats minimal, essentially coming down to a page break.

Respondents had also asked the Boards to postpone changes to income presentation until there was a conceptual basis for OCI, that is, until they had made progress on a revised conceptual framework.³⁵ Initially begun as a joint project, the Boards continued their work on the conceptual framework project separately, with the FASB in deliberations in 2015 and the IASB having issued a Discussion Paper in July 2013 and an Exposure Draft in May 2015. In DP/2013/1, the IASB noted the criticism that OCI was ‘perceived as a “dumping ground” for anything controversial’ and that recycling remained vague (8.3). Asserting that all items of comprehensive income indicate an entity's performance, the IASB noted investors’ and managers’ preference for profit or loss as a performance indicator. In similar vein, it affirmed profit or loss as a total or subtotal (8.22) and decided to retain reclassification adjustments (8.26), without providing insights into its decision process (Barker et al. 2014).

Given its inclination to maintain the status quo, the IASB inductively assessed OCI to develop a reporting model for current practice. The Board noted, and rejected, common suggestions to distinguish between profit or loss and OCI, asserting that one single attribute cannot distinguish between profit or loss and OCI (8.38), which ‘comes close to saying that OCI cannot be defined’ (Barker et al. 2014, pp. 173–4). Instead, the Board suggested two descriptive approaches:³⁶ Following the first, profit or loss would be the primary performance indicator and include all income items, unless OCI increased the relevance of income. That is, OCI would be used, if the IASB ‘occasionally’ decided on different measurement approaches for the balance sheet and income statement (‘bridging items’), or if an item is an incomplete depiction of a remeasurement (‘mismatched remeasurements’). All items of OCI would be reclassified. The second approach followed similar principles, but took a broader view to allow a more selective use of recycling. Only this broader view would accommodate current reporting practice, but neither approach was considered convincing because neither described OCI in a mutually exclusive way.

In its May 2015 Exposure Draft, the IASB did not develop its previous proposals using the feedback from respondents. Instead, the Draft appears as an attempt to shut off the debate on OCI, using relevance as a conceptual bridge to descriptively make sense of the practice. All items of income and expense are presumed to be part of profit or loss, unless using OCI increases the relevance of profit or loss, which can be the case only for changes in current measures of assets and liabilities (7.24). Similarly, items of OCI are presumed to be recycled, unless doing otherwise increases the relevance of profit or loss (7.27).

The Draft hence continues the ambiguity of OCI in three respects. First, it suggests that OCI is the result of a dual measurement approach, where current values are deemed important for items of the balance sheet, but a different measurement basis is more relevant for determining profit or loss (6.76).³⁷ Yet, by further covering OCI in the chapter on presentation and disclosure, the IASB has increased the convolution, making it unclear whether OCI relates to the recognition of items directly in equity (at least historically), the measurement basis for determining profit or loss, or merely the presentation of income and expense items. Second, this indecision is closely connected to the ambiguity of ‘financial performance’, which seems to consist primarily of profit or loss, but also includes OCI. Leaving this key concept undefined means avoiding confrontation with constituents, especially from Japan, which might undermine the prospects of IFRS adoption there. Third, the IASB has decided that it is ‘not feasible or appropriate’ to define or describe OCI (BC7.36). Instead, it has used the relevance characteristic to provide ‘high-level guidance’, requiring unspecified ‘compelling reasons’ for using OCI and rejecting recycling. This paper has shown that OCI was never the result of relevance considerations and using this qualitative characteristic implies another ex-post rationalization that has the potential of further convoluting its meaning. Like the alternative views described in AV2–3, it is argued here that the proposed guidance for OCI does not entail a conceptual solution to performance reporting, but creates the risk of continuing the piecemeal approach of excluding items from profit or loss with the backing of a conceptual framework.

In summary, both FASB and IASB seem to be steering toward a layered presentation of comprehensive income, with the presentation in a single statement only a matter of time. To some extent, this approach is reminiscent of the format outlined by Paton and Littleton (1940), that is, an all-inclusive presentation of income in two sections, separated according to whether they are recurring. The remaining difference results from the strong attachment to income as a performance indicator and the retaining of reclassification adjustments. A striking feature of the Boards’ deliberations is the close relation between income reporting and conceptual issues and it has appeared difficult to discuss one without touching on the other. The IASB, having considered briefly to eliminate OCI, has decided to retain the practice at least in the short run. By doing so, it avoided not only confrontation with its constituents, but also a fundamental discussion of recognition and measurement concepts, which, in a conceptual framework project, appears surprising. Given the longevity of a conceptual framework, maintaining OCI now implies that the once discussed long-term objective of eliminating OCI remains a long way off.

8. Concluding discussion

Both preparers and users of financial statements focus on net income as the key measure of a company's financial performance. As one dissenter to the IASB's Exposure Draft of a revised conceptual framework explained (AV30), it is this fixation on net income that is ‘the root of the debate between historical cost and current value’ and the resulting controversy about how to determine financial performance. This paper has demonstrated that the herculean efforts to shape net income by way of engineering what goes into profit or loss led to the emergence of what is now known as OCI.

From the start, OCI was a compromise solution, needed to foster agreement among Board members. As Johnson and Reither (1995, p. 9) state: ‘It is not clear that the requisite majority of Board members would have voted for the adoption of those statements without the compromises afforded by bypassing the income statement and taking those items directly to equity.’ While referring to the statements on marketable securities, the comment is an equally good description of other projects producing direct entries to equity, where the Board resorted to OCI to smooth an otherwise fluctuating income figure, accepting that volatility could not be managed or otherwise avoided. Hence, OCI was used to incorporate current values into the balance sheet, while retaining traditional concepts of realization and matching in the income statement. Johnson and Swieringa (1996b, p. 120) termed this approach an ‘uncoupling’ of the income statement and the balance sheet, ‘ending the requirement that the financial statements articulate with one another'. Instead of introducing such an uncoupling more fundamentally, OCI was ‘expanded on a piecemeal basis [ … ] without consideration of the potential longer-term consequences’ (Johnson and Reither 1995). Inadvertently, these projects set precedents for an accounting treatment that was expanded over the years. When the project on reporting comprehensive income labeled the entry as OCI without (an attempt at) achieving a definitional consensus, an unusual accounting practice became institutionalized.

The subsequent sense-making of OCI seemed to follow the Wittgensteinian notion that a concept is understood through its use and, vice versa, that use determines meaning. However, the attempts at defining OCI were unable to find a common conceptual denominator of the practice (e.g. Barker 2004, Van Cauwenberge and De Beelde 2007, Rees and Shane 2012). Standard setters realized that conceptual considerations did not lead to OCI and that finding a definition inductively might not bring more than a descriptive approach. Yet, any conceptual fresh start to income reporting, as explored by Barker (2003, 2004), is complicated by practical factors, such as the IASB's focus on IFRS adoption in Japan, where a strong attachment to net income and recycling, and an equally strong aversion to comprehensive income, prevails (e.g. Van Mourik and Katsuo 2015).

The piecemeal approach that expanded OCI also made further conceptual considerations necessary, primarily because of the use of reclassification adjustments. In the currency translation project, they were employed provisionally, pending the decisions on comprehensive income. Initially, the FASB had preferred not to use recycling, as translation adjustments were already included in comprehensive income. Applying this concept consistently would have no need for recycling. Yet, as the FASB issued further OCI standards, the practice was carried forward as part of a compromise and, eventually, preferred over the arguments of SFAS No. 52. Recent discussions show that recycling has become a more contentious issue: As standard setters need to fit in current views, it is asserted that reclassification adjustments sustain income as the primary performance indicator and that eliminating recycling undermines the income concept.

Without conceptual agreement on performance reporting, OCI will remain a peculiar practice without guidance for subsequent decisions, as shown by the IASB's recent Exposure Draft, where OCI is fitted into a convoluted position between the recognition of income items, determination of profit or loss, and presentation or classification issues. Using a dual understanding of relevance as high-level guidance means that the qualitative characteristic is constructed as a conceptual remedy without any appropriate first-order criterion for excluding items from profit or loss. As the paper shows, relevance has never been the reason for OCI such that an anchoring to this characteristic is unlikely to be of more use to the IASB than the currently lacking definition of OCI. In that sense, the current proposals on OCI make it unlikely that financial performance will be ‘robust and tinker-free’, at least not from a standard-setting perspective (Hoogervorst 2014, p. 8). Instead, they might even increase the risk of standard setters employing the practice arbitrarily.

Acknowledgements

I gratefully acknowledge the comments and suggestions received from Richard Barker, Kees Camfferman, Sebastian Hoffmann, Jim Leisenring, Tim Lucas, Christoph Pelger, Peter Walton, Tatsumi Yamada and Stephen Zeff. I also thank the associate editor and an anonymous reviewer for their help and support in further developing this paper.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. E-mail from Tim Lucas, the FASB's former Director of Research and Technical Activities, 2 December 2014.

2. This approach was apparently inspired by the ASBJ (Hoogervorst 2014). In a comment letter, the ASBJ suggested that ‘OCI is “the linkage factor” that is used when measurements that are relevant from the perspective of reporting an entity's financial position differ from the measurements that are relevant from the perspective of reporting an entity's financial performance’ (emphasis in the original).

3. SFAS No. 80 required recognition of the value changes of hedging instruments directly in equity, if they hedged assets recorded at fair value, with value changes recognized in equity under SFAS No. 12. In 1998, the FASB issued SFAS No. 133 with its familiar distinction of fair value and cash flow hedges, where the effective portion of a cash flow hedge is taken to OCI. Conceptual discussions of OCI did not feature strongly in the two standards.

4. A perusal of contemporary reviews of annual reports suggests that Carey's (1948) assessment was too optimistic and that there was ‘no sign of any decline in the practice of direct entries to earned surplus’ (Sanders 1949, p. 318; see also Werntz and King 1946, American Institute of Accountants 1949).

5. To increase transparency, ARB No. 35, issued in October 1948, asked for a separate display in the surplus statement of items excluded from net income.

6. Bailey (1948, pp. 11–12) was not convinced of the supporting role of disclosures: ‘It is seldom possible to give enough information with respect to a special item to permit the reader to judge its propriety as an item to be included or excluded in determining the income.’

7. To reflect this compromise, Regulation S-X was amended in December 1950 by Accounting Series Release No. 70, while the CAP issued ARB No. 41 in July 1951.

8. I thank Jim Leisenring for pointing out this issue.

9. The dissenters criticized the standard for disregarding the underlying economic circumstances and making the recognition of unrealized gains in income depend on the classification of securities. This criticism has reappeared over the years in projects on financial instruments (see note 12) and relates closely to the discussion of business-model and intent-based accounting (Leisenring et al. 2012).

10. All SEC Commissioners agreed with Breeden's testimony and supported the idea of extending use of market valuation (Schuetze 2006). However, current values were not supported unanimously. Among others, SEC General Counsel James Doty and Chairman of the Federal Reserve Board Alan Greenspan raised concerns about current value accounting (Kirk 1991, Beresford 1993, Johnson and Swieringa 1996a).

11. By then, the FASB had issued two disclosure standards relating to marketable securities: SFAS No. 105 in March 1990 and SFAS No. 107 in December 1991.

12. The dissenters criticized that SFAS No. 115 allowed accounting based on intent and did not prevent gains trading. Hence, what arose out of the need to agree on the accounting for financial instruments appears to have made a long-lasting impact in that later financial instruments projects, particularly the one resulting in IFRS 9, have labeled this kind of accounting more explicitly as being based on business models and asserted its higher relevance for users, a view that is rejected, among others, by Leisenring et al. (2012).

13. In the IASB's revision of its financial instruments standard, measurement at fair value through OCI was initially upheld for some equity instruments only (IFRS 9.5.7.5), whereas the final standard expanded this measurement approach to include debt instruments managed in a mixed business model (IFRS 9.4.1.2A).

14. This section draws from Detzen (2014), who discusses the foreign currency translation project in detail. The discussion here summarizes the key points from the standard setting project.

15. SFAS No. 52 carried forward SFAS No. 8's treatment of foreign currency transactions and the recognition of transaction gains and losses in income (para. 15; for exceptions see para. 20).

16. While making a case based on conceptual arguments, the dissenters did not mention that the temporal method had been in conflict with the common notion of prudence, that is, the recognition of losses as they occur. In the case of depreciating currencies, SFAS No. 8 deferred the recognition of translation losses for items translated at the historical exchange rate.

17. The format was also driven by the period of high inflation in the late 1970s and early 1980s (see, for example, Tweedie and Whittington 1984). The FASB's inflation standard, SFAS No. 33, required disclosure of the effect of changing prices either in a statement format or as a reconciliation of income. The latter approach was said to be well in line with comprehensive income (Tweedie and Whittington 1984, p. 178).

18. The similarity between the two standards was also demonstrated by the SEC's 1994 decision to allow foreign registrants to use portions of a revised IAS 21 as one of only three IAS standards accepted in accounts filed with the SEC (Camfferman and Zeff 2007, p. 317).

19. SFAS No. 106 on other post-retirement benefits includes requirements that ‘are similar to the ones established in SFAS 87, with some new differences to accommodate the demands of some constituents’ (Miller et al. 1998, p. 137). Other post-retirement benefits was made a separate project in 1984 so as not to delay the project on pension accounting. The project also sparked considerable controversy (e.g. Van Riper 1994).

20. The same year, the disclosure standard SFAS No. 36 was issued as an interim solution, requiring footnote disclosure of a company's unfunded accumulated benefit obligation.

21. A further measure to reduce volatility was the recognition of an intangible asset for unamortized costs of plan amendments in the case of prior service costs. The intangible would be amortized over active plan participants’ remaining years of service. For an illustration of the proposed rules, see Lucas and Miller (1983).

22. In a Senate hearing on pension plans, FASB Chairman Donald J. Kirk gave testimony on the Board's proposals, having to defend not only the FASB's intent to present pension plans in financial statements, but also the Board as a private initiative (Kirk 1983). Van Riper (1994, pp. 103–4, 120–2, 127–8) further demonstrates the opposition to the Board's pension proposals.

23. The three dissenters were generally supportive of recognizing a pension liability, but criticized how SFAS No. 87 determined pension cost and the pension obligation.

24. SFAS No. 87 required about 20% of public companies, mostly in heavy or unionized industries, to considerably increase their pension liabilities (Berton 1985). Following Wyatt (1990), it can be argued that the standards on pensions and other post-employment benefits established a broader sense of accountability for social phenomena, but did not reflect a conceptually appropriate accounting solution.

25. E-mail from Tim Lucas, project manager on the pensions project, dated 2 December 2014.

26. The project coincided with the one on performance reporting and recognition of pension income and expenses, including actuarial gains and losses, was proposed according to the new ‘matrix’ format (see Section 7). When this format was abandoned, the IASB proceeded with recognition of actuarial gains and losses in OCI, but required entities that adopted this option to present changes in equity in a separate performance statement, termed ‘statement of recognized income and expense’.

27. The AAA's Financial Accounting Standards Committee criticized this decision, arguing that the project would raise fundamental issues beyond income presentation (Linsmeier et al. 1997).

28. The latter proposal was inspired by the UK ASB's approach in Financial Reporting Standard 3 and the two Boards went on to cooperate on the project (Johnson and Reither 1995). Given that the IASC considered a similar approach at the time, Johnson and Swieringa (1996b, p. 121) identified ‘an emerging trend toward reporting financial performance in two statements'.

29. See the responses cited in SFAS No. 130, the staff summary of comment letters to the Financial Accounting Standards Advisory Council, and Financial Executives Institute (FEI 1996). For an analysis of the comment letters, see Yen et al. (2007).

30. Initially, the FASB included an exception for minimum pension liability adjustments due to practicability reasons (SFAS No. 130, paras. 19, 91). Following the expansion of OCI in SFAS No. 158, reclassification adjustments were also required for pension items recorded in OCI.

31. This decision was taken in an October 2001 meeting, with Japanese Board member Tatsumi Yamada being the only dissenter. Foreshadowing the concerns from Japan, Yamada considered the decision premature and requiring too drastic changes to accounting practice, because he saw profit or loss as an important performance indicator (e-mail from Tatsumi Yamada, 24 June 2015).

32. The suggested matrix format also implied an unwelcome end to hedge accounting and the IASB had introduced an exception for cash flow hedges, undermining its principled approach to income presentation.

33. The IASB had retreated from a mandatory single income statement late in its deliberations, primarily to avoid confrontation with constituents. Phase A then resulted only in small changes beyond SFAS No. 130, which is why the FASB refrained from issuing a separate Exposure Draft (Camfferman and Zeff 2015).

34. The Staff Draft has not yet been issued as an accounting standard and the project is currently on hold. The FASB has re-initiated deliberations and added the project to its research agenda in January 2014.

35. While the Boards appeared supportive of this view, the IASB's then senior director of technical activities, Alan Teixeira, expressed his personal view of OCI as a standards-level topic that should not be part of the conceptual framework (Abela et al. 2014).

36. An alternative approach, which conflicted with the Board's views, was to prohibit recycling. According to the Board, this approach would imply that profit or loss was conceptually not different from any other subtotal.

37. This dual approach to measurement is evident throughout the Draft and results from the IASB's resolution not to express a preference for one measurement basis (Hoogervorst 2015). While OCI is thus described consistently with the remaining Exposure Draft, this approach risks making the framework descriptive at the expense of providing standard-setting guidance.