«How well does net income measure ﬁrm performance? A discussion of two studies1 Douglas J. Skinner* University of Michigan Business School, 701 ...»
Journal of Accounting and Economics 26 (1999) 105—111
How well does net income measure
A discussion of two studies1
Douglas J. Skinner*
University of Michigan Business School, 701 Tappan Street, Ann Arbor, MI 48109-1234, USA
Received 1 December 1998
The papers by Dhaliwal, Subramanyam and Trezevant (1998) and Vincent (1998) both
examine whether stock returns are more highly associated with net income or an alternative measure of ﬁrm performance in contexts that are of some current interest to accounting regulators. However, since neither paper does a very good job of motivating their basic economic questions, we are left with results that are not all that interesting or surprising.
Both papers would have beneﬁted greatly from a clearer delineation of the economic rationale for their tests and predictions. 1999 Elsevier Science B.V. All rights reserved.
JEL classiﬁcation: L85; M41 Keywords: Capital markets; Comprehensive income; Real Estate Investment Trusts (REITs); Summary measures of performance
1. Introduction At ﬁrst glance, the papers by Dhaliwal, Subramanyam, and Trezevant (hereafter, DST, 1998) and Vincent (1998) seem disparate. While the former deals with Financial Accounting Standards Board (FASB) Statement of Financial Accounting Standards No. 130, ‘Reporting Comprehensive Income’ (SFAS-130), * Tel.: (734)-764-1239; fax: (734)-936-0282; e-mail: email@example.com.
I am grateful to KPMG for ﬁnancial support and to Patty Dechow for helpful discussions.
0165-4101/99/$ — see front matter 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 5 - 4 1 0 1 ( 9 9 ) 0 0 0 0 5 - 1 106 D.J. Skinner / Journal of Accounting and Economics 26 (1999) 105—111 the latter deals with the reporting of Funds from Operations (FFO) by Real Estate Investment Trusts (REITs). However, as it turns out, the basic research question that underlies both studies is fundamentally the same — how well does net income measure ﬁrm performance compared to possible alternative measures? (DST compare net income to comprehensive income while Vincent compares net income to FFO.) Both papers address topics that are of current concern to policymakers. In the case of comprehensive income, the FASB has just issued SFAS-130 (June 1997), and is likely to be interested in evidence on whether comprehensive income is more highly associated with stock returns than net income. In the case of FFO, the REIT industry group (NAREIT) has argued for several years that FFO should replace net income in these ﬁrms’ ﬁnancial reports, and the SEC has already allowed these ﬁrms to include FFO in their ﬁnancial statements on a supplementary basis, so there seems to be some regulatory sympathy for the use of FFO as a performance measure in this industry.
Despite the currency of these topics, I have concerns about the motivation for each of these papers, which I outline in turn. In my view, neither paper does a convincing job of outlining the economic rationale for their basic research approaches, with the result that not a great deal is learned from either paper.
2. Motivation for the research
One thing that neither paper does particularly well is provide an economic rationale for why we might expect the variable of interest (comprehensive income or FFO) to outperform net income as a summary measure of ﬁrm performance and why the best approach to studying this question is through a stock-returns association study.
To motivate their paper, DST cite arguments from the accounting literature about whether we should use the ‘current operating performance’ approach or the ‘all inclusive’ approach to income measurement. At the risk of oversimpliﬁcation, GAAP has traditionally preferred the all-inclusive approach because this approach provides a more objective and veriﬁable measure of earnings, and so is more desirable for contracting purposes. On the other hand, stock market investors are more likely to prefer a current operating performance approach, under which the ﬁrm reports a number that is closer to ‘core earnings’ (i.e., excludes transitory earnings components). This follows because stock market investors are most interested in predicting future cash ﬂows/earnings and so are principally interested in those components of earnings that persist into future A contemporaneous paper by Gore and Scott (1998) also assesses alternative measures of ﬁrm performance in the REIT industry using a stock-returns association approach and reaches similar conclusions to Vincent (1998).
D.J. Skinner / Journal of Accounting and Economics 26 (1999) 105—111 107 periods. This being the case, it is hard to see why the authors chose a stockreturns based approach since this is a context where comprehensive income would seem to be relatively less useful to users. To give comprehensive income a fair chance, it would seem to be better to look at contracting contexts where there are good reasons to utilize all-inclusive measures of performance (e.g., where performance is being measured ex-post for compensation purposes).
To see more clearly why stock returns are likely to be more highly associated with net income than with comprehensive income, consider what comprehensive income comprises and how it compares to net income. Under previous (preSFAS 130) GAAP, three non-capital items are taken directly to stockholders’
equity, thus violating ‘clean surplus’ accounting:
1. Unrealized gains and losses on ‘available-for-sale’ securities pursuant to SFAS-115 (accounting for investments),
2. The net loss associated with the minimum liability pension adjustment pursuant to SFAS-87 (pension accounting), and
3. Foreign currency translation adjustments pursuant to SFAS-52 (foreign currency translation).
A principal objective of SFAS-130 is to eliminate these clean surplus violations by requiring companies to report a new number (in addition to net income) called ‘Comprehensive Income’, deﬁned as net income adjusted to include the eﬀects of these three items (which together comprise ‘other comprehensive income’).
It is hard to see how any of these three items have implications for the ﬁrm’s future operating performance or cash ﬂows, which is the principal focus of stock market investors. First, past changes in the value of investment securities have no implications for future changes, and so are unlikely to be related to future earnings or cash ﬂows. Thus, while these changes reﬂect changes in the value of the ﬁrm’s assets and so will likely be reﬂected in current equity values, they are completely transitory. Second, the other two components of other comprehensive income are essentially accounting adjustments that are diﬃcult to interpret economically and which sophisticated analysts tend to ignore in estimating future earnings and cash ﬂows; these items arguably add ‘noise’ to reported earnings (e.g., see White et al., 1998 (Chapters 15 and 18)). As a result, it is hard to imagine circumstances under which these items would be correlated with stock returns. Overall, it is not clear why we would expect comprehensive income to be more highly correlated with stock returns than net income, since it includes items that have few, if any, implications for the ﬁrm’s future cash ﬂows and earnings.
These are apparently the largest items. SFAS-130 (para. 39) indicates that there are actually eight of these types of ‘dirty surplus’ items.
108 D.J. Skinner / Journal of Accounting and Economics 26 (1999) 105—111 The problem that DST face in motivating their paper is that it is hard to see any economic justiﬁcation for SFAS-130 itself, so it is hard to motivate empirical tests for an associated academic study. As the statement (see SFAS-130 para.
1) makes clear, no new information is disclosed as a result of the statement: “This statement establishes standards for reporting and display of comprehensive income and its components2It does not address issues of recognition or measurement.” In other words, all the statement really does is to display previously disclosed information in a diﬀerent way. Apparently, there was some concern that ﬁnancial statement users were unable to fully process this information previously, and that the new manner of disclosure would correct this deﬁciency. This concern is diﬃcult to understand. The argument implies that investors process information about recognized ﬁnancial statement items diﬀerently depending on where these items are reported in the ﬁnancial statements. This is an even more extreme version of the long-standing recognition vs. disclosure debate under which some argue that investors process items diﬀerently depending on whether they are disclosed in footnotes or recognized in the ﬁnancial statements. As Bernard and Schipper (1994) observe, there are some economic reasons why we might expect investors to place diﬀerent interpretations on items depending on whether they are recognized vs. disclosed (e.g., recognized items may be more reliable). However, these arguments are more diﬃcult to make in the case of SFAS-130. In the end, one gets the impression that the motivation for SFAS-130 is principally political rather than economic. At the same time that SFAS-130 was being developed, the FASB was also considering the highly controversial issue of accounting for derivative securities, and in particular preferred a mark-to-market approach under which unrealized gains and losses on derivative securities would be reported as part of income. Because some constituents (most notably ﬁnancial institutions) strongly objected to including these items in income, while some on the Board strongly objected to including these items as additional ‘dirty surplus’ items in stockholders’ equity, a compromise position was reached under which certain of these items are included as part of comprehensive income, which is a convenient ‘middle ground’ between these alternatives. Overall, there seem to be two problems that underly the motivation for DST. First, the economic rationale for SFAS-130 itself is weak, so it is hard Two quotes from a Wall Street Journal article on the statement illustrate this argument: “The new ﬁgure will shine a bright, embarrassing light on items that are now buried in shareholders’ equity” and “The new rule will jog loose unexpected information.” See The Wall Street Journal, 30, September 1997, at C1, “FASB Rule Will Oﬀer Walk on Wild Side” by Elizabeth McDonald.
Consistent with this, SFAS-133, “Accounting for Derivative Instruments and Hedging Activities” (issued in June 1998), classiﬁes certain unrealized gains and losses on derivative instruments as part of ‘other comprehensive income’.
D.J. Skinner / Journal of Accounting and Economics 26 (1999) 105—111 109 to understand exactly what it is we are setting out to test. Second, given the deﬁnition of comprehensive income, it is diﬃcult to understand why we would expect comprehensive income to be more highly associated with stock returns than net income. As a result, it is not too surprising that the authors ﬁnd that net income dominates comprehensive income in most of their tests.
A similar argument can be made about the motivation of the Vincent paper, although here the argument is somewhat stronger. In the case of REITs, GAAP requires depreciation of real property, causing reported net income to be relatively low. The industry argues that including this depreciation expense distorts net income as a performance measure, and so advocates the use of FFO instead. FFO essentially comprises net income before depreciation and before the inclusion of gains and losses on property sales and debt restructurings. Thus, FFO arguably is closer to being a measure of ‘core earnings’ for REITs, and so one can more plausibly see why this measure may be more highly correlated with stock prices than net income.
Once again, however, Vincent does not do a good job of spelling out the economic argument for FFO as a superior measure of ﬁrm performance.
Instead, she simply documents the industry argument that FFO is somehow ‘better’ because it excludes depreciation, but fails to tell us precisely what it is about depreciation expense in this industry that causes it to distort performance measurement. Presumably, since others have studied the association between stock returns and the components of accruals, there is something unique about the economics of the REIT industry that make it an interesting industry to study. Unfortunately, the author does little to explain why this might be. The author also argues that “REITs provide an opportunity to examine claims that traditional, accounting-based performance measures are inferior alternatives, particularly for fast-growing ﬁrms whose operations do not conform well to the traditional manufacturing or retailing model.” Yet it is hard for me to see how adjusting net income for depreciation and some other items to arrive at FFO is an example of an ‘alternative reporting model.’ A related problem is that it is not clear why investors should care which of these alternative measures are reported as the primary performance measure in the ﬁnancial statements, since it seems to be easy to adjust back and forth between the two measures based on published information. (Vincent is able to obtain very good estimates of FFO by adjusting net income for depreciation and the other items using information disclosed in her sample ﬁrms’ ﬁnancial statements.) This being the case, it is hard to see why the form of disclosure should matter, as long as the components of performance are separately disclosed. Thus, it is unclear why the author expects one or other measure to provide diﬀerent amounts of information to investors, unless one again assumes that investors cannot make simple adjustments based on numbers that are reported on the face of ﬁnancial statements.
110 D.J. Skinner / Journal of Accounting and Economics 26 (1999) 105—111
3. Research design considerations