Linda Smith Bamber
J.M. Tull Professor of Accounting
University of Georgia
John (Xuefeng) Jiang
Assistant Professor
Michigan State University
Kathy R. Petroni*
Deloitte /Michael Licata Professor of Accounting
Michigan State University
Isabel Yanyan Wang
Assistant Professor
Michigan State University
October, 2007
We are grateful to Joe Anthony, Ben Ayers, Michael Bamber, Stephen Brigham, John Hassell,
Marilyn Johnson, Yen-Jung Lee, Tom Linsmeier, Kin Lo, Tom Omer, Karen Sedatole, Min
Shen, Jenny Tucker, Dave Weber, Matt Weiland, Eric Yeung, workshop participants at the
University of Connecticut, Indiana University at Indianapolis, University of Kentucky, and
Michigan State University, and participants in the 2007 American Accounting Association
Annual Meeting, the Financial Accounting and Reporting Section Mid-Year Meeting, and the
2007 Canadian Academic Accounting Association Annual Conference for insightful comments that helped us improve this paper.
* Corresponding author: N250 Business College Complex, Eli Broad College of
Business, Michigan State University, East Lansing MI, 48824. Phone: 517-432-2924. Fax: 517-
432-1101. Email: petroni@bus.msu.edu
.

Comprehensive Income: Who’s Afraid of Performance Statement Reporting?
Abstract
Our study provides new insight into why most firms do not follow policymakers‟ preference to report comprehensive income in a performance statement, and instead relegate it to the statement of changes in equity. We argue that managers believe reporting comprehensive income in the more salient performance statement will lead financial statement users to perceive the firm‟s performance as more volatile and therefore have a negative impact on stock prices and evaluations of managerial performance. Our empirical evidence on a broad cross-section of firms shows that managers with stronger equity-based incentives and less secure positions are less likely to adopt the more transparent performance reporting. Our evidence demonstrates that even though the reporting location choice is benign in a traditional rational markets view, managers act as if they believe that comprehensive income reporting location matters.
Keywords: Accounting choice, Comprehensive income, Executive compensation, Managerial job security

1.
Introduction
Financial reporting standards in the United States currently allow firms to report comprehensive income in either an income-statement-like format as part of a statement of performance,
1
or in a statement of changes in equity (Financial Accounting Standards Board
[FASB] FAS 130). In traditional models of financial markets, rational investors fully process information regardless of its location, so it does not matter where firms report comprehensive income. Recent empirical evidence suggests that this may be the case. For example, in supplementary analysis, Chambers et al. (2007) do not find a statistically significant difference in the pricing of other comprehensive income that is reported across the two locations.
2
The FASB, however, believes the location of comprehensive income does matter because it has stated a preference for the performance statement presentation – referred to as performance reporting – viewing this as the more transparent presentation (FASB 1997, paragraph 67). In addition, the International Accounting Standards Board (IASB 2007) recently ruled that firms must report comprehensive income in a performance statement by 2009 . Apparently, managers also believe reporting location matters because contrary to policymakers‟ preferences, most firms relegate comprehensive income to the statement of changes in equity (Campbell et al. 1999;
Bhamornsiri and Wiggins 2001; Pandit et al. 2006). If managers believed the location decision had little consequence, firms would follow the FASB‟s recommendations because managers place a great deal of importance on acquiring a reputation for transparent reporting (Graham et al. 2005, 54). Our study seeks to shed light on why managers avoid performance reporting.
The only empirical evidence on determinants of firms‟ comprehensive income reporting
1
Under FAS 130, Reporting Comprehensive Income, the income-statement-like format can take one of two specific forms: (1) a statement that includes the information of an income statement as well as comprehensive income, or (2) a separate statement that begins with net income and ends with comprehensive income.
2
For reasons discussed in footnote 8, we believe that further research is necessary to better understand investors‟ response to the comprehensive income reporting location choice.
1

location choice is the Lee et al. (2006) investigation in the property-liability insurance industry.
They provide evidence that insurers who relegate comprehensive income to the statement of changes in equity are more likely to smooth their earnings by cherry-picking realized gains and losses on available-for-sale securities. Apparently, these insurers believe that performance reporting would make their attempts to smooth earnings more transparent. The insurance industry is an excellent setting to investigate the link between firms‟ comprehensive income reporting choices and the selective sales of available-for-sale securities. Opportunities to cherrypick are ripe in this industry because the average available-for-sale portfolio is about 40% of total assets (Godwin et al. 1998).
Outside the insurance industry, however, available-for-sale securities constitute a smaller proportion of firm assets, which likely reduces the opportunity for cherry-picking. Yet firms outside the insurance industry are even more likely (than insurers) to relegate comprehensive income to the statement of changes in equity (Lee et al. 2006). Thus, we examine the role of other potential impediments to performance reporting in a broad cross-section of firms.
Yen et al. (2007) analyze the comment letters on the FASB‟s controversial Exposure
Draft Reporting Comprehensive Income (FASB 1996), which spurred more than twice the average number of comment letters as the FASB‟s first 100 accounting standards (Yen et al., 59 citing Tandy and Wilburn 1992). Yen et al. (2007, 71) conclude that “[l]etter writers were overwhelmingly opposed to the requirement to disclose [comprehensive income] in a statement of financial performance.” Over 80% of the letters expressed concern about external financial statement users‟ reactions to more salient reporting of comprehensive income. More than 40% specifically noted that comprehensive income would be more volatile than net income or expressed concern that the volatility of comprehensive income misrepresents the underlying
2

economics of the firm‟s operations. Letters expressed concern that performance reporting would cause investors to perceive that the firm is riskier. A number of letters further asserted that companies would change their operations to reduce the volatility of comprehensive income “in order to „head off‟ a negative user reaction” (Yen et al. 2007, 69).
3
These comment letters suggest that when managers were making their initial comprehensive income reporting choice (i.e., before managers actually experienced investors‟ responses to their comprehensive income disclosures), many feared that reporting comprehensive income in a salient performance statement would lead investors and other stakeholders to increase their assessments of the volatility of the firm‟s performance.
4
We hypothesize that managers more likely to be hurt by increased perceived volatility of the firm‟s performance would not follow the FASB‟s preference, but would instead relegate comprehensive income to the less salient statement of changes in equity. As described more fully in the next section, managers in the Graham et al. (2005) survey say they believe increases in the perceived volatility of firm performance adversely affect both the firm‟s stock price and assessments of the manager‟s own performance. We therefore expect managers with more powerful equity-based incentives (who suffer more from lower stock prices) and managers with less job security (who face greater risks from unfavorable performance evaluations) would be more likely to avoid reporting comprehensive income in the more salient (performance statement) location.
We test our expectations on the initial comprehensive income reporting choice made by
Standard & Poor‟s (S&P hereafter) 500 firms during the 1998 to 2001 period. Consistent with
3
We independently analyzed the 97 comment letters that managers of our sample firms sent in response to the initial exposure draft. All 97 letters opposed performance reporting, often expressing concern that performance reporting would increase perceived volatility of the firm‟s performance. Managers‟ actions were consistent with positions expressed in the comment letters, as each of these 97 firms actually chose to relegate comprehensive income to the
4 statement of changes in equity.
Barth et al. (1995) and Hodder et al. (2006) provide empirical evidence that comprehensive income is more volatile than net income (in the banking industry). In the next section we discuss research that concludes that the salience of a disclosure can affect market participants‟ perceptions of the firm‟s performance.
3

our hypotheses, firms that disregard policymakers‟ preferences for performance reporting are headed by CEOs with more powerful equity-based incentives and less job security than firms that follow policymakers‟ recommendations.
5
These results are robust to controlling for other comprehensive income items and other variables often associated with accounting choice, such as leverage, industry, and firm size. Additional analysis of the small set of firms that changed their comprehensive income reporting location further supports our inferences.
We also find that firms with larger (absolute) unrealized gains and losses on availablefor-sale securities are less likely to report comprehensive income in a (salient) performance statement. Thus, firms with more opportunity to manage earnings by selectively selling available-for-sale marketable securities tend to avoid performance reporting that would increase the likelihood of detection (Hunton et al. 2006).
Our study contributes to the literature in several ways. First, and most important, our results suggest managers‟ reluctance to follow policymakers‟ preference for reporting comprehensive income in a performance statement reflects their concerns that more salient reporting of comprehensive income may negatively impact stock prices and investors‟ perceptions of managerial competence. This evidence is relevant to the FASB who is currently deliberating whether to require all firms to report comprehensive income in a performance statement (FASB 2006a), consistent with the IASB‟s recent decision . Our evidence demonstrates that even though the reporting location choice is benign in a traditional rational markets view, managers act as if they believe that comprehensive income location matters.
5
As detailed in Section 3, our job security measure is based on two dimensions of CEO power derived from the management literature. Following prior research on determinants of accounting choice (e.g., Cheng and Warfield
2005; Bergstresser and Philippon 2006), we focus on the CEO’s incentives, because the CEO‟s preferences likely affect the firm‟s accounting choices, even if the CFO or controller is literally responsible for the choice. For example, if the CEO prefers smoother reported performance, the firm‟s accounting choices are likely to further this goal.
4

A second contribution of our study is new evidence that managers‟ broader job security concerns play a significant incremental role (beyond their specific equity-based compensation incentives) in determining financial reporting choices. Graham et al. (2005) provide surveybased evidence suggesting that managers‟ career concerns are important drivers of financial reporting, and note that this is an “under-explored” issue (Graham et al. 2005, 24). Our results help address this void by providing large-scale archival evidence that managers‟ job security concerns do appear to affect their accounting choices.
Third, we extend prior research on the effects of equity-based incentives on earnings management (e.g., Cheng and Warfield 2005; Coles et al. 2006; Bergstresser and Philippon
2006). Specifically while this prior research shows that equity-based incentives affect accounting estimates, we extend this literature by showing that equity incentives also affect a transparencyrelated accounting choice – the decision to disclose accounting information in a more or less salient location.
Finally, because firms self-select where to report comprehensive income, factors affecting this choice could also affect the valuation of comprehensive income. Future research on the valuation effects of comprehensive income reporting location could use our model of the determinants of the location to control for self-selection biases.
The paper proceeds as follows. The next section develops testable hypotheses. The third section describes the research method. The fourth section presents the empirical results, and the fifth section concludes.
2. Hypothesis development
This section develops our hypotheses that managers who report comprehensive income in the less salient statement of changes in equity have stronger equity-based incentives and less job
5

security. First, we explain why managers are likely to believe that performance reporting leads financial statement users to increase their perception of the volatility of the firm‟s performance.
Second, we discuss why managers would expect an increase in the perceived volatility of firm performance to hurt their stock prices and performance evaluations. Third, we use these discussions to develop testable hypotheses.
Because the comprehensive income reporting choice is purely one of location (i.e., the content of the disclosure is the same) we can focus on a parsimonious set of explanations that abstracts from many traditional determinants of accounting choice that are unlikely to drive a location choice. For example, existing contracts relying on accounting numbers (e.g., debt contracts) should not affect the location where managers report comprehensive income.
Why managers believe that performance reporting leads to higher perceived volatility of firm performance
There are several reasons why managers making their initial comprehensive income reporting choice (i.e., before actually experiencing investors‟ responses to their comprehensive income disclosures) would have expected performance reporting to lead to higher perceived volatility of firm performance. First, other comprehensive income items are generally more volatile than net income (Barth et al. 1995; Hodder et al. 2006; and Figure 1 of this paper) and laboratory experiments confirm that reporting comprehensive income in a performance statement increases the salience of this volatile measure. Hirst and Hopkins (1998) find that professional financial analysts more fully assimilate the implications of an electronic company‟s comprehensive income when that company uses performance reporting. Indeed, when the company reports comprehensive income in the statement of changes in equity, half the analystsubjects do not even recall seeing the term comprehensive income (consistent with Brown‟s
(1997) evidence that financial analysts regard the statement of changes in equity as one of the
6

least useful components of the annual report). Maines and McDaniel (2000) demonstrate that when nonprofessional investors evaluate the performance of insurance firms, they weight comprehensive income more heavily when it appears in a performance statement than when it appears in a statement of changes in equity. Furthermore, Maines and McDaniel‟s (2000, 179) experimental evidence demonstrates that “investors‟ judgments of corporate and management performance reflect the volatility of comprehensive income only when it is presented in a statement of comprehensive income.” 6
Comment letters responding to the FASB‟s Exposure
Draft confirm that many managers feared that more salient reporting of comprehensive income would lead to higher perceived volatility of firm performance (Yen et al. 2007).
Second, users often fail to fully assimilate hard-to-process information, such as a disclosure about complex financial transactions, which in turn impairs users‟ ability to correctly assess the disclosure‟s implications for future earnings (i.e., to assess differential persistence).
For example, in Hirshleifer and Teoh‟s (2003, HT, hereafter) model, when firms saliently disclose “lumpy” earnings components (employee stock option expense in their specific example), investors overestimate the persistence of the current period realization of this lumpy expense. This insight is relevant to our setting because the components of other comprehensive income are unrealized gains and losses on (1) available-for-sale (hereafter AFS) securities; (2) foreign currency translations; (3) minimum pension obligations; and (4) certain hedging and derivative activities. These unrealized gains and losses are complex, stem from volatile market forces (e.g., stock market trends, currency exchange rates, interest rates), and are transitory in nature. Applying HT‟s inattention to differential persistence notion suggests that when firms
6
Similarly, in Hirshleifer and Teoh‟s (2003) model, investors with limited attention and limited information processing ability more fully assimilate information that is more salient, and fail to fully assimilate identical information that – even if it is relevant – is less prominent. This finding implies that reporting comprehensive income in the more salient performance statement would lead users to focus more heavily on (the more volatile) comprehensive income, which in turn may lead them to perceive the firm‟s performance as more volatile.
7

report comprehensive income more saliently, financial statement users will overestimate the persistence of the transitory other comprehensive income items. In turn this will lead users to perceive the firm‟s performance as more volatile.
The third reason for managers to expect salient reporting of comprehensive income to increase users‟ perceptions of the volatility of the firm‟s performance is that FAS 130 requires firms to report only a subset of their unrealized gains and losses (i.e., only the four categories listed above). In contrast, unrealized gains and losses on other assets and liabilities of the firm, which may be natural or planned hedges for the four items subject to FAS 130 reporting, are not recognized. If a firm reports comprehensive income in a performance statement, users likely will assimilate the implications of the unrealized gains and losses included in comprehensive income
(Hirst and Hopkins 1998; Maines and McDaniel 2000). However, HT‟s limited attention notion suggests that users likely will fail to balance these (saliently-reported) other comprehensive income items against offsetting unrealized gains and losses on other assets and liabilities that are not recognized in the current accounting model. Thus, when a firm starts disclosing other comprehensive income in a salient performance statement, to the extent these items are in fact hedged by other unrealized gains and losses that are not subject to FAS 130 disclosure requirements, users are likely to perceive the firm‟s performance as more volatile.
7
Relatedly, Bloomfield et al. (2006) develop a model showing that if a firm‟s other comprehensive income items are correlated with other information about the firm that is publicly available (for example, when the unrealized gains and losses on the firm‟s AFS securities are correlated with the firm‟s returns), investors fail to recognize fully the redundancy of the information in other comprehensive income items. As a result, investors place more weight on
7
For example, HT (2003, 380) note that their model “suggests that firms that hedge may be viewed by investors as more risky than those that do not if hedge profits are marked-to-market whereas the long-term business risk the firm is hedging is not marked to market.”
8

the other comprehensive income items, thus affecting stock price. The authors test their model in a series of experimental financial markets and show that when the correlated items are large, the firm‟s stock price is indeed more volatile when comprehensive income appears in the more salient performance statement than when it appears in a statement of changes in equity.
In sum, evidence from laboratory experiments (Hirst and Hopkins 1998; Maines and
McDaniel 2000; Bloomfield et al. 2006) and theory (Hirshleifer and Teoh 2003), as well as comment letters on the initial Exposure Draft (Yen et al. 2007), support our maintained assumption that when making their initial comprehensive income reporting choice, managers expected performance reporting to lead to higher perceived volatility of firm performance. That is, our maintained assumption centers on how managers expected investors to respond to their comprehensive income reporting location choice before those managers had experience with investors‟ responses to their actual comprehensive income disclosures. For our maintained assumption to hold, all that is relevant is what managers believed at the time they made their initial comprehensive income reporting choice . Although managers‟ expectations are not directly observable today, all the evidence available at the time (from research and comment letters) supports our assumption that managers had good reason to be concerned that performance reporting might lead financial statement users to place more weight on comprehensive income.
8
8
One recent archival study addresses a related but different question -- whether investors experienced in real-world comprehensive income reporting actually do place more weight on other comprehensive income items reported in a performance statement, over the period 1998-2003. (Note that this is an entirely different question than how managers expected investors to respond.) Chambers et al. (2007) report that other comprehensive income items are associated with share prices when comprehensive income appears in a statement of changes in equity but not when it appears in a performance statement. This result is, however, inconclusive. First, the (apparent) difference in pricing responses across the two reporting locations is not statistically significant (i.e., the pricing multiple on other comprehensive income items is not significantly different across the two reporting locations). Second, tests of the pricing implications of reporting location choice should control for self-selection of reporting location. Finally, tests of the pricing implications should also carefully consider the component of other comprehensive income that represents the reclassification adjustments for realized gains and losses. Extant research has either not addressed the reclassification adjustment or has not treated it consistently. For example, if the reclassification adjustment for realized gains and losses on marketable securities was separately disclosed on the face of the financial statements,
Chambers et al. (2007) include it in the “Other” category, but if not, they included it as part of the unrealized gains
9

Furthermore, as reported in Section 4, our analysis of the small set of firms that changed their comprehensive income reporting location suggests that managers continue to believe that reporting location matters, even after they have experience with investors‟ reactions to actual comprehensive income disclosures. (Specifically, when CEOs‟ circumstances change such that they have more to lose from higher perceived volatility, they are more likely to switch to relegate comprehensive income to the statement of changes in equity.) Finally, to the extent that our maintained assumption is incorrect (i.e., if managers did not fear performance reporting would render comprehensive income more salient), this would work against finding the hypothesized effects.
Why managers would expect an increase in the perceived volatility of firm performance to hurt stock prices and their performance evaluations
Above we argued that managers fear that reporting comprehensive income more saliently would lead to higher perceived volatility of firm performance. We now consider why these managers would have expected higher perceived volatility to hurt stock prices (and performance evaluations).
9
Many managers state they believe that – even keeping cash flows constant – stakeholders perceive more volatile earnings paths as more risky (see Graham et al. 2005).
10
Evidence from laboratory experiments confirms that both professional and nonprofessional investors associate variability in earnings with higher firm risk (Farrelly et al. 1985; Lipe 1998). Koonce et al.
9 and losses on marketable securities.
HT demonstrate theoretically that the salience of a disclosure can affect stock price. Daniel et al. (2002) review the burgeoning literature in finance suggesting that rational arbitrage will not necessarily eliminate mispricing arising from investors‟ cognitive biases.
10
Prior research on how earnings variability affects market participants‟ perceptions of firm value and risk focuses on net income. Yet policymakers consider comprehensive income to be a notion similar to earnings (e.g., FASB
1997, paragraph 26) and comprehensive income is often referred to as an all-inclusive measure of income (FASB
1997, paragraph 2). Thus, we believe it is reasonable for managers to expect investors would react to volatility in comprehensive income similarly to volatility in net income, especially when comprehensive income appears in a performance statement.
10

(2005) also show that financial statement users perceive uncontrollable items as increasing risk, so the relatively uncontrollable nature of other comprehensive income items (that largely stem from unpredictable market forces) exacerbates users‟ perceptions of the firm‟s risk.
Survey evidence (Graham et al. 2005) further demonstrates that managers believe higher perceived risk directly lowers stock price by increasing the firm‟s cost of capital, consistent with theory (Trueman and Titman 1988; Goel and Thakor 2003) and archival evidence (e.g., Gebhardt et al. 2001; West 1970). Higher perceived risk also lowers stock price indirectly if the firm‟s profits decline because customers and suppliers offer less favorable terms of trade to firms considered riskier (consistent with Trueman and Titman‟s theory (1988) and Sommer‟s (1996) archival evidence). Behavioral research supports managers‟ concerns that reporting comprehensive income in the more salient performance statement increases perceived risk
(Maines and McDaniel 2000; Hirst et al. 2004) and hurts the firm‟s stock price (Maines and
McDaniel 2000; Hirst et al. 2004; Hunton et al. 2006).
It is also reasonable for managers to expect increases in the perceived riskiness of the firm to hurt their performance evaluations that are often a function of profits and stock prices
(which both suffer from an increase in perceived risk). For example, Maines and McDaniel
(2000) show that investors evaluating managers‟ performance penalize managers for volatility in other comprehensive income items when comprehensive income appears in the salient performance statement (and not when it appears in the statement of changes in equity).
Finally, managers also likely feared that reporting comprehensive income in a performance statement could ultimately elevate comprehensive income to become the new focal performance measure. Evidence of this concern is managers‟ fierce opposition to the FASB‟s initial Exposure Draft proposal (FASB 1996) that would have required firms to report
11

comprehensive income on a per-share-basis (Yen et al. 2007). This would have paved the way for comprehensive income to become the premier performance measure. Because comprehensive income is generally more volatile than net income, if comprehensive income becomes the new focal performance measure, it would be more difficult to meet or beat benchmarks.
11
Missing benchmarks hurts stock price (Barth et al. 1999; Skinner and Sloan 2002), and more than threequarters of the respondents in the Graham et al. (2005) survey believe that meeting benchmarks helps to improve their professional reputations.
Hypotheses
Above we argued that managers believe reporting comprehensive income in a more salient performance statement will increase the perceived volatility of the firm‟s performance, which they in turn expect to hurt both stock price and evaluations of their own performance. We expect that managers with the most to lose from lower stock price and poorer performance evaluations will be most reluctant to use performance reporting.
Managers whose wealth is more sensitive to changes in the firm‟s stock price (i.e., managers who have more powerful equity-based incentives) have more to lose from a lower stock price, and thus are more likely to prefer reporting methods that reduce the perceived volatility of firm performance (Goel and Thakor 2003).
12
This leads to our first hypothesis
(stated in the alternative):
H1: The likelihood that a firm avoids performance reporting, and instead relegates comprehensive income to the statement of changes in equity, increases in the power of the CEO’s equity-based incentives.
11
Supplementary analysis not tabulated here supports our claim that firms with more volatile earnings are less likely to meet or beat benchmarks based on prior year earnings and analyst forecasts.
12
As footnote 23 explains in more detail, owners grant equity compensation to give managers incentives to assume risk in operational and investment decisions . Equity compensation does not give managers incentives to make financial reporting choices that inflate the volatility of reported earnings, for a given level of volatility in economic earnings. Indeed, the Graham et al. (2005) survey shows that most managers prefer smoother earnings streams.
12

Poor performance evaluations pose a more serious threat for managers with less job security. CEOs are rightfully concerned about job security because forced turnover has been increasing. Specifically, forced dismissals increased significantly from 1971 to 1994 (Huson et al. 2001) and further increased by 170% from 1995 to 2003 (Lucier et al. 2004). This leads to our second hypothesis (stated in the alternative):
H2: The likelihood that a firm avoids performance reporting, and instead relegates comprehensive income to the statement of changes in equity, increases as the CEO’s job becomes less secure.
3 . Research method
Sample selection and descriptive statistics
The comprehensive income data must be hand-collected, so we start with the S&P 500 firms as of December 1998 from Compustat‟s Price, Dividend, and Earnings file. We chose 1998 because that is the year of the required adoption of FAS 130. We drop 46 firms for which we cannot reliably identify comprehensive income or executive compensation data from 1998 to
2001.
13
To more cleanly isolate the determinants of firms‟ long-term comprehensive income reporting policy choices, we follow Lee et al. (2006) and drop 14 firms that changed their comprehensive income reporting choice between 1998 and 2001.
14
Thus, our empirical tests investigate the comprehensive income reporting choices of the remaining 440 firms. This broad cross-sectional sample is representative of large firms in the U.S. economy. Furthermore, the accounting choices of these firms are also of interest in their own right, as they comprise 66% of the total market value of NYSE, AMEX, and NASDAQ. About 80% of our sample firms
13
We drop 43 firms that did not report comprehensive income and three additional firms identified as outliers in our subsequent analysis. Examination of the three outliers‟ 10-K reports revealed that all three were involved in confounding spin-off type organizational changes in 1998. Moreover, two of the outlier firms spun-off from the same parent, but we cannot separately identify their CEO compensation because Execucomp sums the cash pay and options across the two firms (i.e., treating the cash and options as if they came from the same firm).
14
The low rate of switches confirms that the comprehensive income reporting choice is a long-term disclosure commitment. We later examine the small set of firms that changed their comprehensive income reporting location.
13

reported other comprehensive income items for the first time in 1998, while the remaining 20% first reported other comprehensive income in 1999, 2000, or 2001.
Of the 440 sample firms, only 19% (85 firms) report comprehensive income in a performance statement.
15
In contrast, 81% (355 firms) disregard policymakers‟ stated preference and report comprehensive income in a statement of changes in equity. These results on our broad cross-sectional sample demonstrate that the high incidence of performance reporting that Lee et al. (2006) observe in their sample of property-liability insurers (approximately half of their sample insurers use performance reporting) appears rather unique.
16
Therefore, as Lee et al.
(2006, 687) note, one should exercise caution in generalizing inferences from the propertyliability industry to the broader population.
Table 1 displays descriptive statistics for the first year the firms reported comprehensive income. Other comprehensive income is material for the mean (median) firm: the absolute value of other comprehensive income is 12.7% (7.2%) of the absolute value of net income.
Our maintained assumption is that at the time managers make the initial comprehensive income reporting decision ; they expect that the future stream of comprehensive income will be more variable and less predictable than the future stream of net income. This maintained assumption is consistent with evidence that other comprehensive income items are (1) more transitory than net income
17
and (2) less subject to managers‟ control in that these items are driven largely by unpredictable market forces such as interest and exchange rates.
15
Of the 85 firms that use performance reporting, 69 use a separate statement of comprehensive income, while the remaining 16 use a combined statement of net income and comprehensive income. Because so few firms use a combined statement, we classify the two performance reporting formats in one category, following Lee et al. (2006).
16
Additional analyses confirm that the insurance firms in our sample are significantly more likely than the noninsurance firms to report comprehensive income in a performance statement (p < 0.03).
17
In our sample, the average signed total of other comprehensive income items is insignificantly different from zero for 93% of our sample firms, and the serial correlation in other comprehensive income is mildly negative (-0.10).
14

Our assumption is also supported by evidence that the ex post volatility of comprehensive income on average exceeds that of net income. We compute the standard deviation of both income series from two years before the firm starts reporting comprehensive income until
2004.
18
We then divide the standard deviation of comprehensive income by the standard deviation of net income. As expected, Table 1 shows that comprehensive income is 29% (9%) more volatile than net income for the mean (median) firm (the difference in distributions is significant at p < 0.001). Figure 1 displays the distribution of the volatility of comprehensive income relative to net income. The shaded bars to the right of the dotted line show that for 78% of the sample firms, comprehensive income actually turns out to be more volatile than net income, ex post .
19
This evidence supports our maintained assumption that at the time of the initial comprehensive income reporting decision, the average manager would expect comprehensive income to be more volatile than net income.
20
The last four rows of Table 1 provide evidence on the relative magnitudes of each of the four components of other comprehensive income (in the year the firm first reported comprehensive income). Specifically, we divide the absolute value of each component by the sum of the absolute values of all four components. We also record the number of firms reporting a non-zero value for each component. The most common other comprehensive income item is unrealized gains and losses on foreign currency translation (FORCURR), with 81% of our sample firms reporting this item. This component represents more than half of all other comprehensive income for the mean and median firm. The next most common component is
18
Throughout the study we use as-reported comprehensive income data. We are able to obtain these data for two years before the year the firm initially reports comprehensive income, because FAS 130 requires that in the initial reporting year firms must also disclose comparative comprehensive income data for the previous two years.
19
Even a small (e.g., 1%) difference in uncontrollable volatility of the performance measures can be cause for concern when missing a performance benchmark by a penny causes serious stock market consequences.
20
If we confine our analysis to firms whose comprehensive income is more volatile than net income, ex post , we continue to find that firms whose CEOs have greater equity incentives and lower job security are more likely to relegate comprehensive income to the statement of changes in equity.
15

unrealized gains and losses on AFS securities (MKTSEC), reported by 60% of our sample firms.
For the mean (median) firm this component represents more than 30% (8%) of other comprehensive income. Less than 40% of our sample firms report a pension component
(PENSION),
21
and less than 4% of our sample firms report a derivatives-related component
(DERIVATIVES) because most firms started reporting comprehensive income before FAS 133 became effective in 2001. Since the derivative component is so rare in our sample, we do not separately consider it in our empirical analysis.
Empirical proxies for CEO reporting incentives
Power of the CEO’s equity-based incentives
We hypothesize that managers with more powerful equity-based incentives have greater incentives to minimize the perceived volatility of firm performance and so prefer reporting comprehensive income in the statement of changes in equity. Bergstresser and Philippon (2006) measure the power of equity-based incentives as the sensitivity of the CEO‟s stock and stock option holdings to changes in stock prices.
22
Specifically, Bergstresser and Philippon (2006) first measure the dollar change in the value of a CEO‟s stock and stock option holdings that would arise from a one percentage point increase in the firm‟s stock price:
21
Of course, these data precede FAS 158 (FASB 2006b), which will increase the magnitude of the pension component of other comprehensive income because FAS 158 requires firms to recognize in comprehensive income changes in the over- or under-funded status of defined-benefit postretirement plans.
22
Owners grant stock option compensation to give risk-averse managers incentives to assume more risk in their operational and investment decisions. This is quite distinct from, and does not mean that, option compensation gives managers incentives to make financial reporting choices that inflate the volatility of reported earnings. The
Graham et al. (2005) survey shows that most managers prefer less volatile earnings streams. Similarly, Cheng and
Warfield (2005) argue that CEOs with large equity incentives prefer a smoother earnings series. This is likely because, as explained earlier, managers believe that more volatile earnings streams reduce the value of the firm‟s shares, which in turn reduces the value of the shares the manager already owns. Furthermore, Lambert et al. (1991) show that if the probability that a manager‟s options will finish in-the-money is sufficiently high (and the vast majority of our sample‟s options are in-the-money), risk-averse managers have incentives to decrease volatility because they do not want to risk a fall in stock price that will render the options out-of-the money. Results from the
Sawers et al. (2006) laboratory experiment are consistent with the Lambert et al. (1991) prediction that managers with in-the-money options make less risky choices.
16

ONEPCT = the effect of a one percentage point increase in the firm‟s stock price on the value of the firm‟s shares held by the CEO (i.e., 0.01 × share price × number of shares the CEO owns) plus the effect of a one percentage point increase in the firm‟s stock price on the value of the CEO‟s options, calculated following Core and Guay (1999, 2002).
23
Bergstresser and Philippon (2006) use ONEPCT to construct EQUITY_INCENTIVE, which is the portion of the CEO‟s total annual compensation from the firm stemming from a one percentage point increase in the firm‟s stock price:
EQUITY _ INCENTIVE

ONEPCT

ONEPCT
SALARY where:

BONUS
SALARY = the CEO‟s salary;
BONUS = the CEO‟s cash bonus.
(1)
We obtained the stock and stock option ownership information and salary and bonus amounts from ExecuComp, measured at the fiscal year-end of the first year the sample firm reports other comprehensive income.
24
Job Security
Our second hypothesis is that CEOs whose jobs are less secure are more likely to prefer relegating comprehensive income to the statement of changes in equity . Because the board of directors determines whether to retain or dismiss the CEO, CEOs who wield less power with the
23
Specifically, we divide the CEO‟s options into three groups: (1) those awarded in the current year, (2) those awarded in previous years but not yet exercisable, and (3) those that are currently exercisable. For each group, we use ExecuComp data to extract or construct measures of the exercise price and other variables in the Black-Scholes option formula. Core and Guay (2002) show that this procedure (first used in Core and Guay 1999) leads to unbiased estimates of the sensitivity of option value to changes in stock prices in broad cross-sectional samples of firms whose options are in-the-money, and these estimates capture more than 99% of the variation in option portfolio value. Only three of our sample firms have out-of-the-money unexercisable options, and omitting these three firms does not affect our inferences. Thus, we believe we have an unbiased estimate.
24
The results reported in detail here are based on Bergstresser and Philippon‟s (2006) measure that directly captures the sensitivity of a CEO‟s total equity holdings to changes in stock price. Our results are robust to alternative specifications of this equity incentive ratio, such as treating all options as exercisable and deep in-the-money (see equation (9) in Bergstresser and Philippon 2006) or deflating ONEPCT by a firm‟s total assets or market value.
Our inferences are also robust using Cheng and Warfield‟s (2005) measure of CEO equity incentives (i.e., a CEO‟s total stock and option holdings deflated by the firm‟s total outstanding shares).
17

board suffer lower job security (i.e., higher forced turnover rates if perceived or real performance declines). A review of prior research in management and accounting identifies two primary determinants of the balance of power between CEOs and the board: (1) whether the CEO also chairs the board of directors (hereafter termed CEO-chair duality ) and (2) the independence of the board members.
CEO-chair duality increases the CEO‟s power over the board. As Lucier et al. (2004, 13) note, “a chief [executive] who is also chairman has far more influence” on the board of directors.
Furthermore, both Lucier et al. (2004) and Desai et al. (2006) provide empirical evidence that
CEOs who also serve as chairman of the board of directors enjoy greater job security.
CEOs also wield more power over the board when a greater percentage of directors are firm insiders. As Weisbach (1988) notes, it can be costly for inside directors to challenge the
CEO to whom their careers are tied. Similarly, the management and accounting literatures argue that boards that are more independent are more likely to fire CEOs (e.g., Fredrickson et al. 1988;
Friedman and Singh 1989; Laux 2006) for poor performance. Consistent with this view, Huson et al. (2001) demonstrate that more independent boards are associated with higher forced CEO turnover. We therefore expect CEOs of firms with less independent boards of directors to enjoy greater job security.
Because both CEO-chair duality and board independence affect CEOs‟ power in relation to their boards – and thus their job security – our proxy for job security sums these two measures
(measured in the first year the firm reports other comprehensive income):
JOB_SECURITY = CHAIRMAN + DIRECTORS (2) where:
CHAIRMAN = indicator variable that equals one if the CEO is also the chairman of the board of directors; 0 otherwise; and
18

DIRECTORS = an indicator variable that equals one if the percentage of independent directors on the firm‟s board is smaller than the sample median; 0 otherwise.
Higher values of JOB_SECURITY suggest the CEO has more power and thus more job security.
25
Model
We test the determinants of firms‟ comprehensive income location choices using the following logit model:
AVOID_PERF = β
0
+ β
1
EQUITY_INCENTIVE + β
2
JOB_SECURITY + β
3
MKTSEC_DUM
+ β
4
PENSION_DUM + β
5
FORCURR_DUM + β
6
DISC_QUAL
+ β
7
LEVERAGE + β
8
LSIZE + ε (3) where:
AVOID_PERF = indicator variable that equals one if the firm avoids reporting comprehensive income in a performance report and instead reports in a statement of changes in equity, otherwise zero;
EQUITY_INCENTIVE = the sensitivity of the CEO‟s stock and stock option holdings to a one percent change in the firm‟s stock price, following
Bergstresser and Philippon (2006), as explained above;
26
JOB_SECURITY = a measure of the CEO‟s job security, as explained above;
MKTSEC_DUM = indicator variable that equals one if the absolute value of the unrealized gains and losses from the firm‟s AFS securities (handcollected from the firm‟s 10-K report and then deflated by total assets), averaged over the two years prior to and the initial comprehensive income reporting year, exceeds the sample median; otherwise zero;
PENSION_DUM = indicator variable that equals one if the absolute value of the unrealized gains and losses resulting from changes in the minimum pension obligation (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the two years prior to
25
Our job security variable is a parsimonious measure of the balance of power between the CEO and the board of directors who determines whether to retain or dismiss the CEO. Our inferences are, however, robust to a number of alternative measures of CEO job security, as explained in Section 4.
26
Using the natural log or the rank of EQUITY_INCENTIVE does not affect our inferences.
19

and the initial comprehensive income reporting year, exceeds the sample median; otherwise zero;
FORCURR_DUM = indicator variable that equals one if the absolute value of unrealized gains and losses from foreign currency translation
(hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the two years prior to and the initial comprehensive income reporting year, exceeds the sample median; otherwise zero;
DISC_QUAL = a disclosure quality factor extracted from analyst following, institutional holdings, and bid-ask spreads, following Lee et al.
(2006), as of the end of the initial comprehensive income reporting year;
27
LEVERAGE = long-term debt deflated by total assets, as of the end of the initial comprehensive income reporting year; and
LSIZE = log of the market value of the firm‟s common shares outstanding, as of the end of the initial comprehensive income reporting year.
H1 predicts that the EQUITY_INCENTIVE coefficient will be positive if CEOs whose stock and stock option holdings are more sensitive to changes in the firm‟s stock price avoid performance reporting (and instead report comprehensive income in a statement of changes in equity). H2 predicts that the coefficient on JOB_SECURITY will be negative, because we expect CEOs whose jobs are less secure to avoid the more salient performance reporting.
In addition to the primary variables used in our hypothesis tests, we control for a number of other factors. First, we control for the magnitudes of each of the other comprehensive income items by including dummy variables that are coded one if the absolute value of the other comprehensive income item (hand-collected from the firm‟s 10-K report) deflated by total assets, averaged over the three years starting two years before the initial comprehensive income
27
To maximize the power of the factor analysis and minimize estimation error in the extracted disclosure quality factor, we start with all firms listed in the 1998 Compustat file that have data for the bid-ask spread (from CRSP), analyst following (from I/B/E/S), and institutional holdings data (from CDA Spectrum). A principal factor analysis yields one common disclosure quality factor that, as expected, is a negative function of the bid-ask spread, and a positive function of analyst following and institutional holdings. The factor explains about 56% of the total variation in the original variables. We then apply these factor coefficients to the bid-ask spread, analyst following, and institutional holdings to compute the disclosure quality factor for our sample firms.
20

reporting year, exceeds the sample median. These variables are denoted as MKTSEC_DUM,
PENSION_DUM, and FORCURR_DUM. These controls are necessary because our analysis of the comment letters opposing FAS 130 revealed that most of the resistance to performance reporting (that mentioned specific components of comprehensive income) focused on the unrealized gains and losses on AFS securities. Almost twice as many critics expressed concern about the effect of the volatility in unrealized gains and losses on AFS securities as mentioned pension or foreign currency translation components. A second reason for including these controls is that the magnitudes of the other comprehensive income items affect the volatility of comprehensive income and, thus might affect the firm‟s comprehensive income reporting location choice. Our controls are based on an average of the annual magnitudes of the other comprehensive income items because, given the transitory nature of these items, we expect an average to better reflect the expected future magnitude of each component (than would the amount observed in any one year). We use an above- versus below-the-median bifurcation to specify these variables because “large” values of these other comprehensive income items might affect managers‟ comprehensive income reporting decisions, but managers‟ reporting choices are unlikely to be a linear function of the highly skewed raw values of the other comprehensive income items. However, our inferences are robust using the ranks of the other comprehensive income items.
28
Because Lee et al. (2006) find that firms that committed to higher disclosure quality in the past are more likely to follow policymakers‟ preference to report comprehensive income in
28
We do not separately control for the component of other comprehensive income related to hedging and derivatives because this component is zero for 96% of our sample firms. (Adding a control for this component does not affect our inferences.) Further sensitivity tests reveal that controlling for the signed values of the other comprehensive income items does not affect our inferences. This result is not surprising given the transitory nature of other comprehensive income items and the fact that the average signed value of other comprehensive income is insignificantly different from zero for 93% of our sample firms.
21

the performance statement, we control for their measure of disclosure quality. Their empirical results suggest that the coefficient on DISC_QUAL will be negative (i.e., firms that committed to higher disclosure quality in the past are more likely to choose performance reporting). We also control for leverage (LEVERAGE) because Graham et al. (2005) provide some evidence suggesting that managers of more levered firms are more concerned with smoothing earnings to minimize the perceived risk of the firm. Therefore, more highly levered firms may be more likely to report comprehensive income in a statement of changes in equity. Similar to Lee et al.
(2006), we also control for the log of firm size (LSIZE).
4. Results
Descriptive statistics
Panel A of Table 2 reports descriptive statistics for our sample firms. The mean of
EQUITY_INCENTIVE shows that, on average, a 1% change in the firm‟s stock price will lead to a 29% change in the CEO‟s total annual compensation from the firm.
29
The mean of
JOB_SECURITY is 1.29, and long-term debt averages 21.6% of total assets. Not surprisingly, our sample S&P 500 firms are large, with a median (mean) market value of equity of $7.9
($20.2) billion.
30
Panel B of Table 2 reports correlations among the independent variables. CEO equity incentives (EQUITY_INCENTIVE) are greater for firms whose CEOs have more job security, consistent with powerful CEOs using their influence to get more equity compensation (Morse et al. 2007). Equity incentives are associated also with several of the control variables. For
29
The mean value of EQUITY_INCENTIVE in our study (0.29) is similar to Bergstresser and Philippon‟s (2006,
517) mean of 0.24. Because equity incentives increased during the 1990s, it is not surprising that the mean value of
EQUITY_INCENTIVE is somewhat higher in our sample period that starts in 1998 than in Bergstresser and
Philippon‟s (2006) period that starts in 1994.
30
The indicator variables for the other comprehensive income items are coded as zero when reported values fall in the lower half of the distribution and are coded as one when values fall in the upper half. For MKTSEC_DUM and
FORCURR_DUM, exactly half of the observations are coded as zero and the other half coded as one, so the median
(and mean) values of these variables are 0.5 by construction. For PENSION_DUM, only 38.6% of the sample firms report a pension component, so more than half of the observations are coded as zero.
22

example, equity incentives are greater in larger firms, and equity incentives are also associated with the MKTSEC_DUM and PENSION_DUM components of other comprehensive income.
Consistent with prior research, DISC_QUAL is higher for larger firms (Lang and Lundholm
1993; Kasznik and Lev 1995).
Logit results
Table 3 displays the results of estimating the logit model in Equation 3 with and without controlling for industry. We control for industry by adding to Equation 3 a dummy variable for each major economic sector identified by the Global Industry Classification Standards (GICS) codes.
31
The benchmark industry (whose effect is impounded in the intercept) is Energy.
All significance levels are based on robust standard errors that correct for heteroscedasticity and dependence across firms within the same industry (Froot 1989; Wooldridge 2002). The results support our hypothesis that managers with more powerful equity-based incentives will be less willing to follow policymakers‟ recommendation to report comprehensive income in the more salient performance statement. Specifically, firms whose CEOs have higher
EQUITY_INCENTIVE are more likely to avoid reporting comprehensive income in a performance statement (p

0.001) whether or not we control for industry. Table 3 also shows that firms with CEOs that enjoy greater JOB_SECURITY are less likely to reject policymakers‟ recommendation for performance reporting, consistent with our second hypothesis (p ≤ 0.04 in both cases).
Firms with large absolute unrealized gains and losses on AFS securities (i.e.,
MKTSEC_DUM = 1) are more likely to relegate comprehensive income to the statement of
31
We use the GICS industry classification because Bhojraj et al. (2003) show that it generates more homogenous industry groups (for example, in terms of contemporaneously correlated stock returns, valuation multiples, operating characteristics, and forecasted growth in earnings, sales, and R&D) than SIC codes or the Fama-French (1997) algorithm. Moreover, Bhojraj et al. (2003) show that the GICS advantage is more pronounced for S&P 500 firms than for mid- and small-cap firms.
23

changes in equity (p < 0.01 in both cases). This is consistent with the notion that firms enjoying more opportunity to manage earnings through selective sale of AFS securities are less apt to choose the reporting location that would render any such earnings management more transparent.
To get a sense of the economic impact of our findings, we calculate the percentage change in odds for these key explanatory variables (untabulated). This measures the effect of a change in the independent variable on the odds of reporting comprehensive income in a statement of changes in equity rather than in a performance statement assuming that all other independent variables are at their means.
32
A one-standard-deviation increase in the value of
EQUITY_INCENTIVE increases the odds the manager will avoid performance reporting by
66.1%. A one-unit increase in the job security measure (which ranges from zero to two) reduces the odds the manager avoids performance reporting by 28.7%. Moving from the bottom to the top half of the distribution on MKTSEC increases the odds that the firm relegates comprehensive income to the statement of changes in equity by 63.1%. These effects are economically significant.
With respect to the remaining control variables, DISC_QUAL is marginally negative (p =
0.08) in the model without industry controls, but becomes insignificant when we control for industry effects.
33
This result provides evidence from a broad cross-sectional sample of firms that is weakly consistent with the Lee et al. (2006) finding that insurers that have a history of higher quality reporting are more likely to report comprehensive income in the more salient performance statement. The pension and foreign currency components of other comprehensive
32
For continuous variables, the percentage change in odds is 100 [exp(β coefficient for variable i and δ i i
x δ i
) – 1], where β i is the estimated
is the sample standard deviation of variable i. For discrete variables, the percentage change in odds is 100 [exp(β i
) – 1].
33
In additional analyses, we find that the disclosure quality index varies across industries and therefore conclude that controlling for industry abstracts from these industry-average effects of disclosure quality on the comprehensive income reporting choice.
24

income, leverage, and firm size are not significant explanators of firms‟ comprehensive income reporting choice. However, many of the industry dummy variables are significant, indicating that there are systematic differences across industries in the likelihood that a firm within an industry using performance reporting.
Analysis of firms changing their comprehensive income reporting location
To help ensure that our inferences are not simply an artifact of some unidentified correlated omitted variable, we now examine the (necessarily small) set of firms that changed their comprehensive income reporting. We start with the 14 firms in the S&P 500 (as of
December 1998) omitted from our primary sample because they changed their comprehensive income reporting choice between their initial reporting year and 2001. We then examined the
133 firms included in the October 2002 S&P 500 that were not in our original sample (which was based on the December 1998 S&P 500) and identified six firms that changed their comprehensive income reporting location between 1998 and 2001. This approach yields a sample of 20 firms that changed their comprehensive reporting location.
34
Table 4 Panel A provides descriptive statistics on these “change” firms. Specifically, we report the changes in our explanatory variables between the year the firm first reported comprehensive income and the year the firm changed its comprehensive income reporting location. We denote these “change” variables with the prefix CHG.
For the 13 firms that changed away from salient performance reporting to instead relegate comprehensive income to the statement of changes in equity, EQUITY_INCENTIVE increased , while JOB_SECURITY decreased , on average. In stark contrast, for the seven firms that
34
Restricting our analysis to the 14 change firms deleted from our original sample yields similar inferences. We report results based on the larger sample to help ensure that our inferences are robust. We investigated additions to the S&P 500 that changed their comprehensive income reporting only between 1998 and 2001, because recent research suggests that the Sarbanes-Oxley Act that took effect in 2002 significantly affected firms‟ financial reporting (e.g., Lobo and Zhou 2006; Bartov and Cohen 2007).
25

switched from the statement of changes in equity to the more salient performance reporting,
EQUITY_INCENTIVE decreased and JOB_SECURITY increased . The difference across reporting locations for the change in EQUITY_INCENTIVE is significant (p = 0.07), and the difference in the changes in JOB_SECURITY is also significant (p = 0.04).
35
Given the small sample size, we consider these results consistent with increasingly powerful equity incentives and reduced job security reducing the likelihood of reporting comprehensive income in the more salient performance statement. None of the other explanatory variables changed significantly between the initial comprehensive income reporting year and the year the firm changed the comprehensive income reporting location.
Panel B reports the results of a logit analysis on the change firms, where the dependent variable equals one for firms that switch from performance reporting to reporting comprehensive income in a statement of changes in equity; zero otherwise. We include only the change version of our two variables of primary interest, CHGEQUITY_INCENTIVE and
CHGJOB_SECURITY, because we have so few degrees of freedom, and Panel A of Table 4 shows that none of the other variables differ significantly across changes in reporting location.
Consistent with the univariate change analysis, we find that the coefficient is positive on
CHGEQUITY_INCENTIVE (p = 0.018) and negative on CHGJOB_SECURITY (p = 0.015).
That is, CEOs with increasingly powerful equity incentives and decreasing job security are more likely to switch away from performance reporting. These change results are consistent with our levels analysis and indicate that managers continue to act as if reporting location matters, further supporting our hypotheses that CEOs with higher equity incentives and less job security are more likely to report comprehensive income less saliently.
35
Nonparametric Wilcoxon rank sum tests lead to similar inferences (p = 0.03 for both EQUITY_INCENTIVE and
JOB_SECURITY), so the results are not attributable to a few extreme values.
26

To assess whether these changes in reporting location are associated with changes in
CEOs we searched for CEO changes in all of our 20 location change firms. We found that seven of them changed CEOs between the time the firm initially reported comprehensive income and the time the firm changed its reporting location. Four of the firms changed away from performance reporting and three of them changed to performance reporting. This suggests that there is no systematic direction of change associated with a change in CEO.
Additional analyses
Our primary analysis uses Equation 3 for parsimony. However, our results are robust to controlling for a number of other variables. First, we control for the extent to which comprehensive income actually turns out to be more volatile than net income, ex post, by adding the standard deviation of comprehensive income divided by the standard deviation of net income to Equation 3. Our inferences remain robust. The coefficient on this control variable is insignificant, similar to the results in Lee et al. (2006), but this is not necessarily surprising.
First, the MKTSEC_DUM, FORCURR_DUM, and PENSION_DUM independent variables already capture some of the volatility of other comprehensive income items. Second, ex post realized volatility is a very noisy measure of the (unobservable) ex ante volatility managers expected when they made the initial comprehensive income reporting choice. Over a long horizon, the average manager should have expected comprehensive income to be more volatile than net income, but it would have been almost impossible to predict accurately the market forces (e.g., fluctuations in stock market trends, foreign currency exchanges rates, and interest rates) and thus the actual magnitudes of the gains and losses that comprise the other comprehensive income items. Indeed, we find that the correlation between pre-FAS 130 relative volatility (volatility of comprehensive income relative to net income) and post-FAS 130 relative
27

volatility is insignificantly different from zero (p = 0.26), which confirms that it would have been extremely difficult for managers making the initial comprehensive income reporting choice to predict future volatility. Third, if managers expected comprehensive income to be more volatile but were unable to quantify the magnitude of this higher volatility stemming from uncontrollable market forces, it would be the managers with the most to lose from higher volatility who would be most likely to avoid performance reporting. We argue that because most managers should expect comprehensive income to be more volatile than net income, it is managers‟ sensitivity to the perceived volatility of the firm‟s performance (not the ex post measured level of volatility) that better explains their comprehensive income location choice.
We also repeat the analysis after including an indicator variable equaling one if the firm uses an industry-specialist auditor, because Lee et al. (2006) find that insurance firms using an industry specialist are more likely to report comprehensive income in the performance statement.
In our sample, the industry-specialist auditor is not significant, and including it does not affect our inferences. Similarly, including other variables sometimes associated with accounting choice, such as a proxy for the firm‟s financing needs (issuance of debt and equity, scaled by total assets) and the natural log of the market-to-book ratio (to control for growth opportunities) does not affect our inferences, and the coefficients on these control variables are not significant.
We also reestimate our logit model after eliminating firms in the financial industry (which generally have more extensive AFS securities) or firms in high-tech industries (which tend to have larger equity incentives). Our primary inferences are unaffected by these two sensitivity tests.
Our inferences are also robust to a number of alternative specifications of job security.
First, Collins et al. (2006) find some evidence that the percentage of gray directors (in addition to
28

CEO-chair duality and a smaller proportion of independent directors) is associated with stock option backdating, suggesting that the percentage of gray directors may be associated with CEO power. Supplementing our job security variable by adding an indicator variable identifying firms with above-median percentage of gray directors does not affect our inferences. Second, it is possible that the CEO‟s job security could be affected by the extent to which the firm‟s corporate governance tilts the balance of power toward managers and away from shareholders. Adding an indicator variable that identifies firms whose managers wield greater-than-median power relative to shareholders (per the Gompers et al. (2003) GSCORE) does not affect our inferences (beyond marginally increasing the explanatory power of JOB_SECURITY). Third, although poor firm performance is associated with increased CEO turnover (e.g., Warner et al. 1988; Murphy and
Zimmerman 1993), we do not expect the short-term, historical value of a time-varying characteristic (such as stock returns at the initial comprehensive income reporting date) to drive management‟s comprehensive income reporting location choice. This choice is a long-term disclosure commitment that should reflect managers‟ assessment of the likelihood that salient reporting of comprehensive income will in the future adversely affect investors‟ perceptions, thereby putting the CEO‟s job at risk. Therefore, it is not surprising that including annual returns in our logit model does not affect our inferences, nor is this variable significant. Finally, to the extent that our measure of job security is a noisy reflection of the unobservable degree of the
CEO‟s job security, this should simply reduce the power of our tests.
Our inferences are also robust using ordered logit with three categories, ordered as follows: comprehensive income reported in statement of changes in equity, performance
29

reporting in a separate statement of comprehensive income, and performance reporting in a single combined statement of net income and comprehensive income.
36
5. Conclusions
Policymakers prefer that firms report comprehensive income in a performance statement rather than in the statement of changes in equity. As this preference is simply a reporting location choice, one would expect managers to acquiesce because they place a great deal of importance on developing a reputation for transparent reporting (Graham et al. 2005, 54). However, over
80% of our sample S&P 500 firms do not follow policymakers‟ preference, and instead relegate comprehensive income to the statement of changes in equity. Our study provides new insight into why so many firms are reluctant to follow policymakers‟ preference for reporting comprehensive income in the more salient performance statement.
We draw on prior research to explain why we expect that at the time they made the initial comprehensive income reporting choice, managers were concerned that reporting comprehensive income – which is typically more volatile than net income – in a performance statement would increase the perceived volatility of the firm‟s performance. We hypothesize that managers most likely to be hurt by higher perceived volatility are those with more powerful equity-based incentives (that will be devalued by a lower stock price) and those with less job security (who have more to fear from a poor performance evaluation). Consistent with our hypotheses, our evidence suggests that when CEOs have more powerful equity-based incentives or less secure positions, the firm is less likely to report comprehensive income in the more salient performance
36
Ideally, we would also like to include an explicit control for cherry-picking similar to Lee et al. (2006).
Unfortunately, we are not able to estimate their measure for our broad-based sample of firms because their measure requires a time-series of realized gains and losses on AFS securities and only about one-quarter of the firms in our sample disclose this information. Lee et al. (2006) were able to estimate this variable for their sample of propertyliability insurers because insurers are uniquely required by FAS 60 (paragraph 50) Accounting and Reporting by
Insurance Enterprises to disclose separately their realized gains and losses on all investments. Although we cannot directly capture ex post cherry-picking, our MKTSEC_DUM variable is similar in spirit because it captures the ex ante opportunity to cherry-pick (which is a necessary, though not sufficient condition for actual cherry-picking).
30

statement and is more likely to relegate it to the statement of changes in equity. The magnitudes of these effects are economically significant, and the results hold even after controlling for industry and other variables often related to accounting choice.
Our evidence suggests that at the time managers made the initial comprehensive income reporting location decision, they believed that performance reporting would be more salient – consistent with the views of the FASB and IASB and consistent with evidence from laboratory studies (e.g., Hirst and Hopkins 1998; Maines and McDaniel 2000). Analysis of the small set of firms that change their comprehensive income reporting location further supports our inferences.
Specifically, CEOs with increasingly powerful equity incentives and decreasing job security are more likely to switch away from performance reporting, suggesting that managers continue to act as if they believe that reporting location matters, and that performance reporting is more salient.
Our evidence that managers with less job security on average make reporting choices that reduce transparency is of interest in its own right, and also helps fill the void that Graham et al.
(2005) identify when they point out the dearth of evidence on how other attributes of managers‟ welfare (beyond equity-based incentives) affect their financial reporting choices. Our finding on the effect of equity-based compensation extends prior research showing that equity-based compensation increases incentives for earnings management (e.g., Cheng and Warfield 2005:
Coles et al. 2006; Bergstresser and Philippon 2006) by providing evidence that equity incentives affect another accounting choice: the decision to disclose comprehensive income in a more or less salient location.
Finally, in comment letters on the initial FAS 130 proposal, managers expressed concern that performance reporting would lead stakeholders to view firms‟ performance as more volatile than warranted by the actual economics. Because we find that managers‟ comprehensive income
31

reporting choices are consistent with their stated concerns, our evidence suggests that managers act as if the concerns expressed in the comment letters are real (as distinct from excuses). For example, our results suggest that managers are concerned that investors may overact to transitory other comprehensive income items that are saliently reported.
Future research can further probe investors‟ actual reactions to comprehensive income disclosures, in order to better assess the ex post validity of managers‟ ex ante concerns. For example, the results of laboratory experiments suggest that managers have a reason to be concerned that presenting comprehensive income in a performance statement has the potential to increase users‟ assessments of the firm‟s risk and hurt the firm‟s stock price and the manager‟s performance evaluation. It is of interest to learn whether these inferences can be triangulated with evidence from real capital markets. It would also be of interest to learn whether there is any empirical support for managers‟ concern that investors overreact to salient disclosure of the transitory comprehensive income items. Such research could help policymakers decide whether reporting comprehensive income in a performance statement or in the statement of changes in equity results in stock prices and evaluations of CEO performance that better reflect the firm‟s true economics.
32

References
Barth, M., J. Elliott, and M. Finn, 1999. Market rewards associated with patterns of increasing earnings. Journal of Accounting Research 37(2): 387-413.
Barth, M., W. Landsman, and J. Wahlen. 1995. Fair value accounting: Effects on banks‟ earnings volatility, regulatory capital, and value of contractual cash flows. Journal of
Banking & Finance 19 (3-4): 577-605.
Bartov, E., and D.A. Cohen. 2007. Mechanisms to meet/beat analyst earnings expectations in the pre-and post-Sarbanes-Oxley eras. Working Paper, New York University.
Bergstresser, D., and T. Philippon. 2006. CEO incentives and earnings management. Journal of
Financial Economics 80 (3): 511-529.
Bhamornsiri, S. and C. Wiggins. 2001. Comprehensive income disclosures. The CPA Journal
(October): 54-56.
Bhojraj, S., C. Lee, and D. Oler. 2003. What‟s my line? A comparison of industry classification schemes for capital market research. Journal of Accounting Research 41 (5): 745-774.
Bloomfield, R., M. Nelson, S. Smith. 2006. Feedback loops, fair value accounting and correlated investments. Review of Accounting Studies, 11 (2/3): 377-416.
Brown, P. 1997. Financial data and decision making by sell-side analysts. Journal of Financial
Statement Analysis 2(3): 43-48.
Canadian Accounting Standards Board. 2005. Decision Summary , November 2, 2005, available at: http://www.acsbcanada.org/index.cfm/ci_id/28794/la_id/1
Canadian Institute of Chartered Accountants (CICA). 2005. CICA Handbook – Accounting:
Section 1530, Comprehensive Income . Toronto: CICA.
Campbell, L., D. Crawford, and D. Franz. 1999. How companies are complying with the comprehensive income disclosure requirements. Ohio CPA Journal (Jan-March): 13-19.
Chambers, D., T. J. Linsmeier, C. Shakespeare, and T. Sougiannis. 2007. An evaluation of SFAS
No. 130 Comprehensive Income Disclosures. Review of Accounting Studies 12: 557-593.
Cheng, Q. and T. Warfield. 2005. Equity incentives and earnings management. The Accounting
Review 80 (2): 441-476.
Coles, J., M. Hertzel, and S. Kalpathy. 2006. Earnings management around employee stock option reissues. Journal of Accounting & Economics 41 (1-2): 173-200.
33

Collins, D., G. Gong, and H. Li. 2006. Corporate governance and backdating of executive stock options. Working paper.
Core, J. and W. Guay. 1999. The use of equity grants to manage optimal equity incentive levels.
Journal of Accounting and Economics 28 (2): 151-184.
Core, J. and W. Guay. 2002. Estimating the value of employee stock option portfolios and their sensitivities to price and volatility. Journal of Accounting Research 40 (3): 613-630.
Daniel, K., D. Hirshleifer, and S. Teoh. 2002. Investor psychology in capital markets:
Evidence and policy implications. Journal of Monetary Economics 49(1):139-209.
Desai, H., C. Hogan, and M. Wilkins. 2006. The reputational penalty for aggressive accounting:
Earnings restatements and management turnover. The Accounting Review 81 (1): 83-112.
Dhaliwal, D., K. Subramanyam, and R. Trezevant. 1999. Is comprehensive income superior to net income as a measure of firm performance? Journal of Accounting & Economics 26:
43-67.
Fama, E., and K. French. 1997. Industry costs of equity. Journal of Financial Economics 43
(2): 153-193.
Farrelly, G., K. Ferris, and W. Reichenstein. 1985. Perceived risk, market risk, and accounting determined risk measures. The Accounting Review 60 (2): 278-288.
Financial Accounting Standards Board (FASB). 1982. Accounting and Reporting by Insurance
Enterprises No. 60 . Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 1996. Exposure draft: Reporting comprehensive income . Norwalk, CT: FASB.
Financial Accounting Standards Board (FASB). 1997. Statement of Financial Accounting
Standards 130: Reporting comprehensive income No. 130 . Norwalk, CT: FASB
Financial Accounting Standards Board (FASB). 2006a. Project Update: Financial Statement
Presentation, as updated on April 26, 2006, Available from www.fasb.org.
Financial Accounting Standards Board (FASB). 2006b. Statement of Financial Accounting
Standards 158:
Employers’ accounting for defined benefit pension and other postretirement plans No. 158. Norwalk, CT: FASB.
Fredrickson, J., D. Hambrick, and S. aumrin. 1988. A model of CEO dismissal. Academy of
Management Review 13 (2): 255-270.
34

Friedman, S. and H. Singh. 1989. CEO succession and stockholder reaction: The influence of organizational context and event content. Academy of Management Journal 32 (4): 718-
744.
Froot, K. A. 1989. Consistent covariance matrix estimation with cross-sectional dependence and heteroskedasticity in financial data. Journal of Financial and Quantitative Analysis
24(3): 333–355.
Gebhardt, W., C. M. C. Lee, and B. Swaminathan. 2001. Toward an implied cost of capital.
Journal of Accounting Research 39 (1): 135-176.
Godwin, N., K. Petroni, and J. Wahlen. 1998. Fair value accounting for property-liability insurers and classification decisions under FAS 115. Journal of Accounting, Auditing and
Finance 13 (3): 207-239.
Goel, A. and A. Thakor. 2003. Why do firms smooth earnings? Journal of Business 76 (1): 151-
192.
Gompers, P., J. Ishii, and A. Metrick. 2003. Corporate governance and equity prices. The
Quarterly Journal of Economics 118 (February): 107-155.
Graham, J., C. Harvey, and S. Rajgopal. 2005. The economic implications of corporate financial reporting. Journal of Accounting & Economics 40 (1-3): 3-73.
Hirshleifer, D., and S. H. Teoh. 2003. Limited attention, information disclosure, and financial reporting. Journal of Accounting & Economics 36 (1-3): 337-386.
Hirst, D., and P. Hopkins. 1998. Comprehensive income reporting and analysts' valuation judgments. Journal of Accounting Research 36 (Supplement): 47-75.
Hirst, E., P. Hopkins, and J. Wahlen. 2004. Fair values, income measurement, and bank analysts‟ risk and valuation judgments. The Accounting Review 79 (2): 453-472.
Hodder, L., P. Hopkins, and J. Wahlen. 2006. Risk-relevance of fair-value income measures for commercial banks. The Accounting Review 81 (2): 337-375.
Hunton, J., R. Libby, and C. Mazza. 2006. Financial reporting transparency and earnings management. The Accounting Review 81 (1): 135-157.
Huson, M., R. Parrino, and L. Starks. 2001. Internal monitoring mechanisms and CEO turnover:
A long-term perspective. The Journal of Finance 56 (6): 2265-2297.
International Accounting Standards Board (IASB). 2007. IAS 1 Presentation of Financial
Statements: A Revised Presentation . London, United Kingdom: IASB.
35

Kasznik, R., and B. Lev. 1995. To warn or not to warn: management disclosures in the face of an earnings surprise. The Accounting Review 70 (1): 113-134.
Koonce, L., M. McAnally, and M. Mercer. 2005. How do investors judge the risk of financial items? The Accounting Review 80 (1): 221-241.
Lambert, R. D. Larcker, and R. Verrecchia. 1991. Portfolio considerations in valuing executive compensation Journal of Accounting Research 29 (1): 129-149.
Lang, M., and R. Lundholm. 1993. Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research 31 (2): 246-271.
Laux, V. 2006. Board independence and CEO turnover. Working paper, Goethe University
Frankfurt.
Lee, Y., M. Shen, and K. Petroni. 2006. Cherry picking, disclosure quality, and comprehensive income reporting choices: The case of property-liability insurers. Contemporary
Accounting Research 23 (3): 655-692.
Lipe, M. 1998. Individual investors' risk judgments and investment decisions: The impact of accounting and market data. Accountin,g Organizations and Society 23 (7): 625-640.
Lobo, G., and J. Zhou. 2006. Did conservatism in financial reporting increase after the
Sarbanes-Oxley Act? Initial evidence. Accounting Horizons 20 (1): 57-73.
Lucier, C., R. Schuyt, and J. Handa. 2004.
CEO Succession 2003: The Perils of ‘Good’
Governance . McLean, VA: Booz Allen Hamilton.
Maines, L., and L. McDaniel. 2000. Effects of comprehensive income characteristics on nonprofessional investors‟ judgments: the role of financial-statement presentation format.
The Accounting Review 75 (2): 179-207.
McKelvey, R. and W. Zavoina. 1975. A statistical model for the analysis of ordinal level, dependent variables. Journal of Mathematical Sociology 4: 103-120.
Morse, A., V. Nanda, and A. Seru. 2005. Are incentive contracts rigged by powerful CEOs?
Working paper.
Murphy, K. and J. Zimmerman. 1993. Financial performance surrounding CEO turnover.
Journal of Accounting and Economics 16 (1-3): 273-315.
Pandit, G., A. Rubenfield, and J. Phillips. 2006. Current NASDAQ corporation methods of reporting comprehensive income. Mid-American Journal of Business 21 (1): 13-19.
36

Sawers, K., A. Wright, and V. Zamora. 2006. Loss aversion, stock-based compensation and managerial risk-seeking behavior. Working paper.
Skinner, D. and R. Sloan. 2002. Earnings surprises, growth expectations, and stock returns or don‟t let an earnings torpedo sink your portfolio.
Review of Accounting Studies 7 (2-3):
289-312.
Sommer, D. 1996. The impact of firm risk on property-liability insurance prices. The Journal of Risk and Insurance 63(3): 501-514.
Tandy, P. and N. Wilburn. 1992. Constituent participation in standard-setting: The FASB‟s first 100 statements, Accounting Horizons 6 (2): 47-58.
Trueman, B. and S. Titman. 1988. An explanation for accounting income smoothing. Journal of Accounting Research 26 (Supplement):127-139.
Warner, J., J. Watts, and K. Wruck. 1988. Stock prices and top management changes. Journal of Financial Economics 20, 461-492.
Weisbach, M. 1988. Outside directors and CEO turnover. Journal of Financial Economics 20:
431-460.
West, R. 1970. An alternative approach to predicting corporate bond ratings. Journal of
Accounting Research 8 (1): 118-125.
Wooldridge, J. 2002. Econometric Analysis of Cross Section and Panel Data . Cambridge,
Massachusetts: The MIT Press.
Yen, A., D. E. Hirst, and P. Hopkins. 2007. A content analysis of the comprehensive income exposure draft comment letters. Research in Accounting Regulation 19: 53-79.
37

Figure 1
Volatility in Comprehensive Income Relative to Net Income
45
40
35
30
25
20
15
10
Std.Dev.CI = Std.Dev.NI
NI
5
0
[0
.3
7,
0.
58
)
[0
.5
8,
0.
79
)
[0
.7
9,
1.
00
)
(1
.0
0,
1
.2
1)
[1
.2
1,
1
.4
2)
[1
.4
2,
1.
63
)
[1
.6
3,
1.
84
)
[1
.8
4,
2.
05
)
[2
.0
5,
5.
62
]
Relative Volatility (Std.Dev. CI/Std. Dev. NI)
38

Table 1
Descriptive statistics for comprehensive income and its components in the year S&P 500 firms initially report comprehensive income during 1998-2001
N Mean Std Dev
Lower
Quartile
Median
Upper
Quartile
Absolute value of other comprehensive income
Absolute value of net income
Standard deviation of comprehensive income
Standard deviation of net income
(n>1)/440 = 78%
Absolute value of MKTSEC
Sum of the absolute values of other comprehensive income items
(n>0)/440 = 60%
Absolute value of PENSION
Sum of the absolute values of other comprehensive income items
(n>0)/440 = 39%
Absolute value of FORCURR
Sum of the absolute values of other comprehensive income items
(n>0)/440 = 81%
Absolute value of DERIVATIVES
Sum of the absolute values of other comprehensive income items
(n>0)/440 = 4%
440 0.127 0.221
440 1.288 0.576
440 0.309 0.385
440 0.113 0.242
440 0.554 0.407
440 0.022 0.135
0.027
1.004
0.000
0.000
0.075
0.000
0.072
1.093
0.083
0.000
0.700
0.000
0.142
1.400
0.691
0.088
0.954
0.000
MKTSEC is the unrealized gains and losses from the firm‟s AFS securities. PENSION is the unrealized gains and losses resulting from changes in the minimum pension obligation. FORCURR is the unrealized gains and losses from foreign currency translation.
DERIVATIVES is the unrealized gains and losses related to hedging and derivative activity.
39

Table 2
Panel A: Descriptive statistics for independent variables
Variable
EQUITY_INCENTIVE
JOB_SECURITY
MKTSEC_DUM
PENSION_DUM
FORCURR_DUM
DISC_QUAL
LEVERAGE
Mean
0.292
1.291
0.5
0.386
0.5
1.7
0.216
Std. Dev.
0.243
0.613
0.501
0.487
0.501
0.583
0.143
Lower
Quartile
0.117
1
0
0
0
1.26
0.11
Median
0.209
1
0.5
0
0.5
1.66
0.199
Upper
Quartile
0.386
2
1
1
1
2.085
0.317
LSIZE 9.047 1.277 8.233 8.979 9.818
Market value of equity (in $ millions) 20,243 34,990 3,764 7,931 18,366
EQUITY_INCENTIVE is the sensitivity of the CEO‟s stock and stock option holdings to a 1% change in the firm‟s stock price, following Bergstresser and Philippon (2006). JOB_SECURITY is the sum of two indicator variables (CHAIRMAN + DIRECTORS). CHAIRMAN is an indicator variable that equals one if the CEO is also the chairman of the board of directors; 0 otherwise. DIRECTORS is an indicator variable that equals one if the percentage of independent directors on the firm‟s board is smaller than the sample median; 0 otherwise. MKTSEC_DUM is an indicator variable that equals one if the average absolute value of the unrealized gains and losses from the firm‟s AFS securities (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the three years starting two years before the initial comprehensive income reporting year, exceeds the sample median; 0 otherwise. PENSION_DUM is an indicator variable that equals one if the average absolute value of the unrealized gains and losses resulting from changes in the minimum pension obligation (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the three years starting two years before the initial comprehensive income reporting year, exceeds the sample median; 0 otherwise. FORCURR_DUM is an indicator variable that equals one if the average absolute value of unrealized gains and losses from foreign currency translation (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the three years starting two years before the initial comprehensive income reporting year, exceeds the sample median; 0 otherwise. DISC_QUAL is the disclosure quality factor extracted following Lee et al. (2006), where higher values indicate higher quality disclosure. LEVERAGE is long-term debt deflated by total assets. LSIZE is the log of the market value of the firm‟s common shares outstanding.
40

Table 2 (continued)
Panel B: Correlations among independent variables
EQUITY_INCENTIVE JOB_SECURITY MKTSEC_DUM PENSION_DUM FORCURR_DUM DISC_QUAL LEVERAGE LSIZE
EQUITY_INCENTIVE
JOB_SECURITY
MKTSEC_DUM
0.167
(<0.001)
0.156
(0.001)
0.193
(<0.001)
-0.043
(0.372)
0.143
(0.003)
-0.052
(0.277)
-0.195
(<0.001)
0.004
(0.931)
-0.140
(0.003)
0.048
(0.319)
-0.052
(0.277)
-0.145
(0.002)
0.126
(0.008)
-0.057
(0.229)
0.109
(0.022)
-0.201
(<0.001)
0.001
(0.980)
-0.252
(<0.001)
0.351
(<0.001)
0.041
(0.390)
0.136
(0.004)
PENSION_DUM -0.192
(<0.001)
-0.002
(0.974)
-0.140
(0.003)
0.112
(0.019)
-0.020
(0.676)
0.108
(0.023)
-0.094
(0.049)
FORCURR_DUM
DISC_QUAL
0.066
(0.165)
-0.058
(0.224)
-0.145
(0.002)
0.112
(0.019)
-0.014
(0.765)
-0.020
(0.670)
-0.026
(0.587)
0.190
(<0.001)
-0.045
(0.346)
0.111
(0.020)
-0.032
(0.505)
-0.023
(0.625)
0.018
(0.704)
0.582
(<0.001)
LEVERAGE -0.239
(<0.001)
0.446
(<0.001)
-0.009
(0.852)
0.040
(0.400)
-0.265
(<0.001)
0.137
(0.004)
0.134
(0.005)
-0.107
(0.025)
-0.006
(0.900)
-0.035
(0.469)
0.019
(0.697)
0.568
(<0.001)
-0.186
(<0.001)
-0.180
(0.000)
LSIZE
EQUITY_INCENTIVE is the sensitivity of the CEO‟s stock and stock option holdings to a 1% change in the firm‟s stock price, following Bergstresser and Philippon
(2006). JOB_SECURITY is the sum of two indicator variables (CHAIRMAN + DIRECTORS). CHAIRMAN is an indicator variable that equals one if the CEO is also the chairman of the board of directors; 0 otherwise. DIRECTORS is an indicator variable that equals one if the percentage of independent directors on the firm‟s board is smaller than the sample median; 0 otherwise. MKTSEC_DUM is an indicator variable that equals one if the absolute value of the unrealized gains and losses from the firm‟s AFS securities (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the three years starting two years before the initial comprehensive income reporting year, exceeds the sample median; 0 otherwise. PENSION_DUM is an indicator variable that equals one if the absolute value of the unrealized gains and losses resulting from changes in the minimum pension obligation (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the three years starting two years before the initial comprehensive income reporting year, exceeds the sample median; 0 otherwise.
FORCURR_DUM is an indicator variable that equals one if the absolute value of unrealized gains and losses from foreign currency translation (hand-collected from the firm‟s 10-K report and then deflated by total assets), averaged over the three years starting two years before the initial comprehensive income reporting year, exceeds the sample median; 0 otherwise. DISC_QUAL is the disclosure quality factor extracted following Lee et al. (2006). LEVERAGE is long-term debt deflated by total assets. LSIZE is the log of the market value of the firm‟s common shares outstanding. Pearson correlations appear above the diagonal, and Spearman correlations appear below the diagonal. Two-tailed p-values are in parentheses. Correlations significant at the 5% level or lower appear in bold.
41

Table 3
Logit analysis of comprehensive income reporting choice
(Dependent variable = 1 if the firm reports comprehensive income in a performance statement; otherwise 0)
(1) (2)
Primary Variables:
Predicted sign
Without industry controls
With industry controls
EQUITY_INCENTIVE
JOB_SECURITY
-
+
2.091
(0.001)
-0.338
(0.029)
1.575
(0.001)
-0.349
(0.036)
Control Variables :
MKTSEC_DUM
PENSION_DUM
FORCURR_DUM
DISC_QUAL
LEVERAGE
LSIZE
Industry Dummies:
Consumer Discretionary
Consumer Staples
Financials
Health Care
Industrials
Information Technology
Materials
Telecommunication Service
Utilities
Intercept
Observations
McKelvey and Zavoina's R
2
?
?
?
?
?
?
?
?
?
?
?
+
?
-
?
?
0.489
(0.002)
-0.083
(0.641)
0.389
(0.151)
-0.434
(0.081)
0.257
(0.805)
0.029
(0.850)
1.381
(0.361)
440
0.117
0.777
(0.018)
0.928
(0.000)
0.565
(0.043)
1.783
(0.000)
-0.148
(0.398)
1.521
(0.000)
2.067
(0.000)
0.541
(0.029)
0.482
(0.019)
0.599
(0.003)
-0.029
(0.857)
0.196
(0.442)
-0.249
(0.241)
0.457
(0.655)
-0.016
(0.931)
0.832
(0.656)
440
0.186
42

Table 3 reports the results of the logit analysis of comprehensive income reporting choices. Independent variables are defined in table 2, except industry dummies are indicator variables that equal 1 if a firm is in the given industry, otherwise zero. In column (2), the benchmark industry is Energy. Reported p-values (in parentheses) are based on
White-adjusted standard errors, corrected for correlation across firms within a given industry (Froot 1989;
Wooldridge 2002). Reported p-values are based on one-tailed significance levels for variables with predictions, and two-tailed significance levels for variables without predictions.
43

Table 4
Analysis of the 20 firms that changed comprehensive income reporting location
Panel A: Univariate tests of differences between firms changing reporting location
Variable
(1)
Change to
Statement of
Equity
(n = 13)
(2)
Change to
Performance
Reporting
(n = 7)
Expected difference:
(1)-(2)
Actual difference:
(1)-(2)
P-value of difference
CHGEQUITY_INCENTIVE 0.059 -0.141 + 0.200 0.073
CHGJOB_SECURITY -0.539 0.429 - -0.967 0.039
0.615 0.286 ? 0.330 0.179 CHGMKTSEC_DUM
CHGPENSION_DUM
CHGFORCURR_DUM
0.231
0.308
0.143
0.429
?
?
0.088
-0.121
0.647
0.627
CHGDISC_QUAL
CHGLEVERAGE
0.068
0.320
-0.019
0.259
-
?
0.087
0.061
0.707
0.473
CHGLSIZE 8.957 8.723 ? 0.233 0.672
Panel B: Logit analysis of determinants of firms’ decision to change comprehensive income reporting location
Dependent Variable=1 if change is to a statement of changes in equity and 0 if change is to a performance statement
CHGEQUITY_INCENTIVE
Predicted sign
+
Coefficient
(p-value)
13.462
(0.018)
CHGJOB_SECURITY -
-2.873
(0.015)
McKelvey and Zavoina's (1975) R
2
= 0.821
CHGEQUITY_INCENTIVE is the difference in EQUITY_INCENTIVE between the year the firm changed its comprehensive income reporting location and the initial year it reported comprehensive income. Similarly, CHGJOB_SECURITY,
CHGMKTSEC_DUM, CHGPENSION_DUM, CHGFORCURR_DUM, CHGDISC_QUAL, CHGLEVERAGE, and
CHGLSIZE are the changes in the original variables (defined in Table 2) between the year the firm changed its comprehensive income reporting location and the initial year the firm reported comprehensive income. Reported p-values are based on twosample t tests with one-tailed significance levels for variables with predictions and two-tailed significance levels for variables without predictions. Reported p -values in Panel B are based on robust standard errors.
44