Voluntary disclosure of operating income
We would like to thank Dan Dhaliwal, Debra Jeter, Irene Tutticci, an anonymous reviewer, Ian Zimmer (Deputy Editor), and participants at the annual AFAANZ Conference for their valuable comments.
Abstract
This study investigates whether New Zealand firms’ voluntary disclosure of operating income, which is also known as earnings before interest and tax, in the income statement is related to the investment opportunity set. New Zealand provides an ideal setting to examine this because New Zealand generally accepted accounting principles do not require the disclosure of operating income as an intermediate income number in arriving at net income (earnings) in the income statement. We hypothesize and find evidence that firms with high assets-in-place and high leverage are more likely to voluntarily disclose operating income/earnings before interest and tax. However, the assets-in-place finding is sensitive to alternative measures of the investment opportunity set.
1. Introduction
Discretionary disclosure research has focussed on earnings disclosures (Skinner, 1994), earnings forecasts (Frankel et al., 1995), disclosure levels (Lang and Lundholm, 1996; Botosan, 1997), disclosure medium (Tasker, 1998; Frankel et al., 1999), the inclusion of balance sheet information with quarterly earnings announcements (Chen et al., 2002) and managerial emphasis in earnings press releases (Bowen et al., 2005). These studies use the discretionary disclosure framework in Dye (1985) and Verrecchia (1983), where firms make voluntary accounting disclosures that are useful to investors in assessing firm value.
Skinner (1994) finds that managers voluntarily make good news disclosures to differentiate themselves from firms that are not performing as well and that they make pre-emptive bad news disclosures to prevent litigation that may arise from withholding or delaying bad news. Frankel et al. (1999) find that high growth firms use conference calls to reduce information asymmetries because their ‘accounting quality’ is low (Tasker, 1998). Chen et al. (2002, p. 231) find that US firms voluntarily disclose balance sheet information when “current earnings are less informative, or when future earnings are more uncertain” (italics in original), and Bowen et al. (2005) provide evidence that managers use emphasis in earnings press releases to draw attention to pro forma earnings that are more ‘value relevant’.
None of these studies links the voluntary disclosure to the firm’s investment opportunity set (IOS). Smith and Watts (1992) indicate that a firm’s IOS drives its policy selection, and it is used to explain cross-sectional variation in financial, dividend and compensation policies (Smith and Watts, 1992; Gaver and Gaver, 1993) and accounting policies (Watts and Zimmerman, 1986; Zimmer, 1986; Watts and Zimmerman, 1990; Skinner, 1993; Dhaliwal et al., 1999; Wong and Wong, 2001). The purpose of this study was to extend the IOS’s relation to voluntary disclosures. Leftwich et al. (1981) examine the relation between firms’ IOS and the voluntary disclosure of interim financial information, but their evidence is inconclusive. Hossain et al. (2005) find the voluntary disclosure of future oriented information by New Zealand firms is related to the level of growth options in their IOS.
This study investigates whether New Zealand firms’ voluntary disclosure of operating income, which is also known as earnings before interest and tax (EBIT), in the income statement is related to the IOS. We argue that investors in firms with assets-in-place find the operating income number/EBIT an appropriate measure of performance for facilitating efficient contracting between the parties to the firm. For example, debt contracts between the firm and debtholders include an interest cover covenant that is defined using operating income/EBIT (Emanuel, 1976; Smith and Warner, 1979; Whittred and Zimmer, 1986). As assets-in-place are efficiently financed by debt, operating income presents a measure of performance of the assets-in-place to all the capital providers (debtholders and equityholders). Consistent with our prediction, we find that assets-in-place are positively related to the voluntary disclosure of operating income/EBIT in the income statement.
Assets-in-place are associated with leverage, so there is an endogenous relation between leverage and the voluntary disclosure of operating income/EBIT in the income statement. That is, assets-in-place determine the voluntary disclosure of operating income/EBIT in the income statement and leverage such that we would observe a relation between the latter two variables. These findings are consistent with managers’ voluntary disclosure of a performance measure being driven by the firms’ IOS.
Our study contributes to the literature on policy choices and voluntary management disclosures. Watts and Zimmerman (1990, p. 151) suggest a relation between the IOS and ‘internal accounting performance measurement systems’, which we extend to accounting performance measurement systems in general. To date and to our knowledge, no one has documented a relation between the IOS and performance measures, so this paper provides the first such evidence. We therefore extend the literature on the correlation between the firm’s IOS with its financial, dividend, compensation and accounting policies to an operating income performance measure.
New Zealand provides an ideal institutional environment to examine the relation between the IOS and the discretionary disclosure of operating income in the income statement. US generally accepted accounting principles (GAAP) require the disclosure of operating income in the income statement, followed by other non-operating income and expenses (including interest expense), then income taxes in arriving at net income (earnings). New Zealand GAAP, both pre- and post-IFRS,1 does not require the disclosure of operating income as an intermediate income number in arriving at net income (earnings) in the income statement. However, we observe that some New Zealand companies voluntarily disclosed operating income or EBIT on the path to arriving at net income (earnings) in the income statement. We examine whether this discretionary disclosure is driven by the firm’s IOS, thereby shedding light on the IOS–disclosure policy relation.
The next section motivates our hypothesis. Section 3 presents the research design, data and empirical results, and we present our conclusion in Section 4.
2. Hypothesis development
2.1. Performance reporting
New Zealand GAAP does not decompose the net income (earnings) number into the operating and non-operating components. In fact, the notion of an operating income that measures the performance arising from the day-to-day core operating activities (Koller et al., 2005, p. 164) is not contemplated. Performance reporting under New Zealand GAAP is prescribed in FRS-2 Presentation of Financial Reports and FRS-9 Information to be Disclosed in Financial Statements,2 where one of the income numbers is referred to as ‘operating surplus (deficit)’, which is defined in FRS-2 paragraph 4.9 and FRS-9 paragraph 4.25 as:
-
‘Operating surplus (deficit)’ is a measure of financial performance resulting from transactions and other events during a period, excluding
- (a)
changes in equity resulting from the effect of extraordinary items;
- (b)
changes in reserves recognized in the statement of movements in equity;
- (c)
currency translation differences permitted to be recognized in the statement of movements in equity; and
- (d)
distributions to and contributions by owners.
New Zealand GAAP requires extraordinary items (item (a) above) to be disclosed separately in the income statement following the ‘operating surplus (deficit)’ to arrive at the ‘net surplus (deficit)’ for the period (FRS-2 paragraphs 4.6 and 6.10 and appendix 1). Items (b) and (c) are the dirty surplus items that by pass the income statement. In the USA, SFAS 130 Reporting Comprehensive Income includes these items to arrive at comprehensive income, which reflects all changes to equity other than those from owner activities, which are item (d) above.
FRS-9 requires the operating revenue and operating surplus (deficit) from continuing activities to be disclosed separately (paragraph 6.2) from the operating revenue and operating surplus (deficit) from discontinued activities (paragraph 6.4). Together, these two combine to produce the ‘operating surplus (deficit)’ for the period (paragraph 6.5 of FRS-9).
Paragraph 4.10 of FRS-2 notes that the ‘operating surplus (deficit) for the period includes the deduction of tax expense’. On the surface, this may suggest that the ‘operating surplus (deficit)’ is a net operating profit after tax (NOPAT). However, this is not so because the operating expenses that are deducted from operating revenues to arrive at the ‘operating surplus (deficit)’ include ‘interest expense and similar charges’ (paragraph 6.13(h) of FRS-9). Hence, New Zealand GAAP’s ‘operating surplus (deficit)’ is not NOPAT, but is net income (before extraordinary items) for equityholders.
The anomalous feature of New Zealand GAAP is the use of the description ‘operating surplus (deficit)’ to describe the net income (before extraordinary items) to equityholders. The ‘operating’ descriptor in reporting performance is typically used to denote a company’s income from core ongoing activities, where ‘financing revenues and expenses (mainly interest expense) are excluded from operating income’ (Wild et al., 2004, p. 317).
In summary, New Zealand GAAP, in spite of using the ‘operating surplus (deficit)’ terminology, does not require the disclosure of operating income which measures a company’s performance from operations before considering the effects of financing. There was no requirement in New Zealand to disclose operating income as an intermediate income number on the path to the net income (earnings) result in the income statement. Notwithstanding this, operating income is not unfathomable. Relatively simple adjustments can be made to the ‘operating surplus (deficit)’ to arrive at operating income, given that interest expense and income tax expense are required to be disclosed in the financial statements (paragraph 6.13(h) of FRS-9 and paragraph 6.12 of FRS-2, respectively). Nevertheless, we do observe a number of New Zealand companies voluntarily disclosing operating income or EBIT in their income statements and this leads to the research question of why some firms voluntarily made this disclosure. The next section develops the argument for the IOS link to the voluntary disclosure of operating income/EBIT in the income statement.
2.2. Investment opportunity set
Myers (1977) argues that the mix of debt and equity in financing the firm’s assets is a function of the firm’s asset structure, which comprises assets-in-place (assets already owned) and growth opportunities (future assets). Myers (1977) indicates that agency costs are inversely (directly) related to the firm’s assets-in-place (growth opportunities) because it is easier to monitor existing assets (for example, property, plant and equipment) than discretionary expenditures (for example, research and development) that are used to create future assets. Hence, debtholders are prepared to finance assets-in-place but not growth opportunities, which are financed by the residual claimants (equityholders).
Smith and Warner (1979) examine the ways in which debt contracts control the conflict of interest between debtholders and equityholders, and they report that debt covenants restrict the firm from engaging in certain activities after the debt has been issued. One covenant prescribes a minimum interest cover ratio, and this constraint is also found in Australian and New Zealand debt contracts (Emanuel, 1976; Whittred and Zimmer, 1986). Its prevalence and survival indicates that the interest cover constraint is an efficient contractual solution for the firm. Interest coverage is calculated by dividing operating income/EBIT by interest and it indicates a firm’s ability to cover its interest charges based on its ongoing operating profits.
In the presence of assets-in-place and debt, the operating income/EBIT performance measure is an efficient value driver that is used to value the firm using a multiples analysis (Soffer and Soffer, 2002; Penman, 2004; Wild et al., 2004; Koller et al., 2005). Valuation multiples are usually derived from a set of firms that are comparable with the target firm in terms of their operating characteristics and capital structure. While selecting firms from the same industry controls for the former, it is usually difficult to find firms that match on capital structure. A practical way to deal with differences in capital structure is to use an unlevered, pre-tax multiple, such as an EBIT multiple, and apply it to the operating income/EBIT value driver to arrive at an estimate of the firm value or the equity value as if the firm had no leverage in its capital structure.
In summary, the IOS determines the mix of debt and equity in the firm. In the presence of assets-in-place and debt finance, operating income/EBIT is an efficient performance measure that is used for debt contracting and in valuing the firm. Hence, our first hypothesis is:
H1: Assets-in-place are positively related to the voluntary disclosure of operating income/EBIT in the income statement.
Assets-in-place also determine the amount of debt in the firm’s capital structure; so, there is an endogenous relation between leverage and the voluntary disclosure of operating income/EBIT in the income statement. That is, assets-in-place simultaneously drive the voluntary disclosure of operating income/EBIT in the income statement and leverage. Hence, our second hypothesis is:
H2: Leverage is positively related to the voluntary disclosure of operating income/EBIT in the income statement.
2.3. Proprietary information
Verrecchia (1983) describes proprietary cost as the cost associated with disclosing proprietary information. Disclosing proprietary information could be harmful to the firm. The proprietary information could give competitors some market advantage, provide labour unions and lenders with bargaining power to renegotiate contracts, and draw the attention of regulators to scrutinize the firm. Verrecchia analyses the role of proprietary information on management’s decision to disclose information.
Where a proprietary cost exists, investors are unsure whether information is withheld because it is bad news, or whether it is good news that is not disclosed because the value of the good news to investors does not justify the proprietary cost. This gives rise to a threshold level of disclosure where certain information is withheld from investors. Verrecchia indicates that the threshold level increases with the proprietary cost, thereby giving managers more discretion as the proprietary cost increases.
Managers’ decision not to disclose operating income/EBIT in the income statement is unlikely to be influenced by proprietary cost. New Zealand GAAP requires disclosure of interest expense and typically this appeared in the notes to the financial statements. This enables investors, if they so desire, to construct the operating income/EBIT number from the pre-tax net income number in the income statement and the interest expense in the notes to the financial statements. Therefore, it is unlikely that proprietary cost is a factor that influences management’s decision not to disclose operating income/EBIT in the income statement.
2.4. Non-proprietary information
Dye (1985) suggests three reasons why management might withhold non-proprietary information. First, management suppresses non-proprietary information because investors are uncertain about the kind of information that management possess. Second, non-proprietary information may be withheld if the value of the firm is dependent on a vast array of proprietary and non-proprietary information and non-disclosure of the latter is unlikely to adversely affect that value.
Finally, managers may withhold non-proprietary information if disclosure exacerbates principal-agent problems between shareholders and management. For example, more frequent reporting of financial information may be useful for forecasting net income but not for monitoring purposes. Verrecchia notes that the inability of Leftwich et al. (1981) to find a relation between the voluntary disclosure of interim reports and proxies for the existence of moral hazard between the firm’s debtholders and shareholders/manager is because interim reports are inapt for contracting purposes. Conflicts between debtholders and shareholders/manager are better solved with debt covenants than through the frequency of financial reporting.
Non-proprietary information is unlikely to be a consideration in management’s decision not to disclose operating income/EBIT in the income statement. The first two reasons indicated above do not apply because interest expense is not withheld. Rather, it is disclosed in the notes to the financial statements if it is not shown on the face of the income statement, thereby enabling the computation of operating income/EBIT. In relation to the third reason above, it is our hypothesis that operating income/EBIT provides an appropriate performance measure for the firm’s capital providers (debtholders and shareholders) to assess the manager’s actions in situations where the firm comprises principally assets-in-place. Debt contracts have an interest cover covenant (Emanuel, 1976; Bowen et al., 1981; Leftwich, 1983), where interest cover is EBIT divided by interest expense. This constraint demonstrates the pertinence of EBIT as a monitoring mechanism. As a result, the disclosure of operating income/EBIT in the income statement resolves, rather than exacerbates, the principal-agent problems by facilitating better contracting between the firm’s capital providers.
2.5. Control variables
Lang and Lundholm (1993) and Frankel et al. (1999) find that firm size is related to firms’ voluntary disclosures. First, firm size proxies for information production costs and these costs decrease with firm size. Second, Skinner’s (1994) legal cost hypothesis posits that damages from securities litigation are related to firm size. However, neither of these explanations applies to the voluntary disclosure of operating income/EBIT in the income statement. Information production costs in this case are minimal; so, it is not likely to be an influencing factor, and it is difficult to conceive of potential litigation costs associated with the non-disclosure of operating income/EBIT in the income statement given the ease with which that number can be computed from the information in the financial statements. We nevertheless include firm size as a control variable to capture the effects of any omitted variables.
Lang and Lundholm (1993), Botosan (1997) and Frankel et al. (1999) find that better performing firms are more likely to make voluntary disclosures. For example, Lang and Lundholm (1993) report a mean (median) ROA for firms holding conference calls of 5 per cent (4 per cent), compared with −9 per cent (2 per cent) for non-conference call firms. We include ROA in our analysis to control for the profitability effect on the voluntary disclosure of operating income/EBIT in the income statement. We also include a control variable for loss companies because these firms may have incentives to disclose information voluntarily (Chen et al., 2002).
Tasker (1998) explores the role of analyst following on voluntary disclosures. She argues that voluntary disclosures using conference calls are used to communicate information to analysts who follow the firm. Botosan (1997) finds a negative relation between the level of disclosure and the cost of equity capital where the analyst following is low but no such association where the analyst following is high. We therefore include analyst following in our empirical tests to control for this effect on disclosure.
3. Research design, data and results
3.1. Research design
()Myers (1977) description of the firm as a combination of assets-in-place and growth options is well understood, but measuring the components is more challenging. Our AIP measure above has been used in previous research (e.g. Leftwich et al., 1981; Smith and Watts, 1992; Gaver and Gaver, 1993; Skinner, 1993; Wong and Wong, 2001). Smith and Watts (1992) indicate that, as assets are measured at depreciated historical cost, the AIP variable is likely to involve a significant measurement error for firms with long-lived assets. However, this problem is less severe for New Zealand companies as they frequently revalue their non-current assets. Similarly, Bradbury et al. (2003), in their study on New Zealand companies’ choice of goodwill amortization period, indicate that their AIP measure is less likely to have significant measurement error because of the frequent revaluation of non-current assets.
Other researchers (e.g. Collins and Kothari, 1989; Chung and Charoenwong, 1991; Gaver and Gaver, 1993; Skinner, 1993; Hossain et al., 2000, 2005) use the market-to-book value of assets (MKTBVA) and the earnings/price ratio to capture the amount of growth options in a firm’s IOS. We also use the MKTBVA measure, but not the earnings/price ratio because both these parameters are sensitive to the capital structure of the firm. To overcome this problem, we use NOPAT divided by the market value of the assets (MKTA), hereafter NOPAT/MKTA. This ratio relates post-tax operating earnings that are available to all capital providers to the market value of the firm, and is better suited for this study which examines the impact of assets-in-place on the voluntary disclosure of operating income. In summary, we attempt to capture the IOS variable using three measures: AIP (our primary variable of interest), MKTBVA and NOPAT/MKTA.
The specification in equation (1) takes the firm’s IOS as pre-determined and assumes that variation in the IOS leads to whether firms report operating income. An endogeneity concern is simultaneity (Larcker and Rusticus, 2008), where the IOS variable in equation (1) could be partially determined by the disclosure variable. If this is so, simultaneous equation models are needed to address the simultaneity. For example, in examining New Zealand companies’ voluntary disclosure of prospective information, Hossain et al. (2005) indicate that firm attributes can affect a firm’s disclosure to stakeholders, but the relation could also go the other way in that the disclosure could reduce the firm’s cost of capital which enables the firm to raise more capital to invest in growth projects. Using a simultaneous system of equations, Hossain et al. (2005) find that New Zealand companies’ voluntary disclosure of prospective information and their IOS are simultaneously determined.
The simultaneity issue addressed in Hossain et al. (2005) is appropriate because there is evidence that disclosures that reduce information asymmetry is associated with a lower cost of capital (Botosan, 1997; Sengupta, 1998; Healy et al., 1999; Lang and Lundholm, 2000; Botosan and Plumlee, 2002) and this, in turn, influences firms’ investment decisions. For our study that examines the relation between the IOS and the reporting of an operating performance measure, we are not aware of any evidence, nor is there a plausible explanation, for why reporting an operating performance measure could affect the IOS. As a result, we follow Smith and Watts (1992) and assume the IOS is pre-determined.
3.2. Data
We examine companies listed on 30 November 2005 Datex Company Annual Reports database. We review each company’s most recent income statement (statement of financial performance) prior to 30 November 2005 to see if they disclose EBIT. Of the 172 companies on the database, a number are excluded for the following reasons: 36 are finance or trust-type companies and their capital structure is unique to their industry, seven are overseas companies, five do not have financial statements on the database and six have negative equity. This leaves 118 companies of which 40 disclose EBIT in the income statement and 78 do not. Panel A of Table 1 summarizes our sample selection process.
| Panel A: Sample selection | |
| Companies listed on the Datex Annual Reports database on 30 November 2005 | 172 |
| Less: Companies in the finance sector or that are trusts | 36 |
| Companies based overseas | 7 |
| Companies without the required financial statement data | 5 |
| Companies that have negative equity | 6 |
| Companies in the final sample | 118 |
| Companies that disclose operating income/EBIT | 40 |
| Companies that do not disclose operating income/EBIT | 78 |
| Companies in the final sample as above | 118 |
| Panel B: Industry distribution | |
| Agriculture and fishing | 9 |
| Building materials and construction | 3 |
| Consumer | 18 |
| Energy | 7 |
| Food and beverages | 5 |
| Forestry and forest products | 4 |
| Intermediate and durables | 13 |
| Investment | 16 |
| Leisure and tourism | 6 |
| Media and telecommunications | 10 |
| Mining | 2 |
| Other services | 9 |
| Ports | 5 |
| Property | 4 |
| Textiles and apparrel | 3 |
| Transport | 4 |
| Companies in the final sample as above in panel A | 118 |
To measure the independent variables in our regression model, we use data from: (i) the companies’ financial statements, as found on the Datex Company Annual Reports database and (ii) from the Datex Company Information database to obtain the necessary information to calculate the market capitalization of equity and to determine if financial analysts provide earnings forecasts for the firm.
Panel B of Table 1 presents the distribution of the 118 companies in the sample by industry classification. About 48 per cent of the companies in the sample are in the consumer, intermediate and durables, investment and media and telecommunications industries, while the remaining percentage of companies is spread among a number of other industry sectors.
3.3. Results
Panel A of Table 2 reports the descriptive statistics for the independent variables in the logit regression analyses (see Table 4) and the results of one-tailed t-tests and Mann–Whitney U-tests of the difference in the mean values between firms that disclose EBIT in the income statement and firms that do not make this disclosure (except for the ANCOV and LOSS variables). For our three measures of the IOS (AIP, MKTBVA and NOPAT/MKTA), panel A shows the following:
| Panel A: Descriptive statistics and univariate tests of variables in logit regression analysis (except for ANCOV and LOSS) | ||||||||
|---|---|---|---|---|---|---|---|---|
| DISC = 0 (n = 78) | DISC = 1 (n = 40) | Test of differences, P-value (one-tailed) | ||||||
| Mean | Median | Standard deviation | Mean | Median | Standard deviation | t-test | Mann–Whitney U-test | |
| AIP | 0.205 | 0.079 | 0.256 | 0.302 | 0.265 | 0.028 | 0.023 | 0.002 |
| MKTBVA | 2.694 | 1.516 | 3.666 | 1.901 | 1.477 | 1.322 | 0.045 | 0.379 |
| NOPAT/MKTA | 0.008 | 0.041 | 0.143 | 0.024 | 0.042 | 0.081 | 0.250 | 0.358 |
| LEV | 0.346 | 0.357 | 0.209 | 0.484 | 0.514 | 0.192 | 0.000 | 0.001 |
| SIZE | 10.742 | 10.736 | 1.903 | 12.105 | 12.284 | 2.093 | 0.000 | 0.001 |
| ROA | −0.028 | 0.059 | 0.323 | 0.045 | 0.066 | 0.146 | 0.045 | 0.163 |
| Panel B: Contingency table analysis to examine the relation between operating income/EBIT disclosure and analyst coverage (ANCOV) | ||||
|---|---|---|---|---|
| DISC | Total | |||
| 0 | 1 | |||
| ANCOV | 0 | 48 | 12 | 60 |
| 1 | 30 | 28 | 58 | |
| Total | 78 | 40 | 118 | |
| Chi-square statistic | 10.523 | |||
| p-value (two-tailed) | 0.001 | |||
| Panel C: Contingency table analysis to examine the relation between operating income/EBIT disclosure and loss companies (LOSS) | ||||
|---|---|---|---|---|
| DISC | Total | |||
| 0 | 1 | |||
| LOSS | 0 | 57 | 35 | 92 |
| 1 | 21 | 5 | 26 | |
| Total | 78 | 40 | 118 | |
| Chi-squared statistic | 3.202 | |||
| p-value (two-tailed) | 0.074 | |||
- DISC denotes a binary variable equal to 1 if operating income/EBIT is disclosed in the income statement, 0 otherwise. AIP denotes book value of property, plant and equipment divided by the sum of market capitalization of equity, minority interest and liabilities. MKTBVA denotes market value of total assets (found by taking the sum of market capitalization of equity, minority interest and liabilities) divided by the book value of total assets. NOPAT/MKTA denotes net operating profit after tax divided by the market value of total assets. LEV denotes total liabilities divided by total assets. SIZE denotes natural log of total assets. ROA denotes net operating profit after tax divided by total assets. ANCOV denotes a binary variable equal to 1 if financial analysts provide earnings forecast for the firm, 0 otherwise. LOSS denotes a binary variable equal to 1 if firm makes a loss, 0 otherwise.
| Variable | Sign | Regression 1 | Regression 2 | Regression 3 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Coefficient | t-statistic | P-value (two-tailed) | Coefficient | t-statistic | P-value (two-tailed) | Coefficient | t-statistic | P-value (two-tailed) | ||
| INTERCEPT | ? | −4.553 | −2.961 | 0.003 | −4.460 | −2.524 | 0.012 | −4.422 | −2.817 | 0.005 |
| AIP | + | 1.223 | 1.375 | 0.169 | ||||||
| MKTBVA | − | −0.008 | −0.095 | 0.925 | ||||||
| NOPAT/MKTA | + | −1.422 | −0.776 | 0.438 | ||||||
| LEV | + | 2.818 | 2.574 | 0.010 | 2.544 | 2.426 | 0.015 | 2.537 | 2.447 | 0.014 |
| SIZE | ? | 0.163 | 1.155 | 0.248 | 0.198 | 1.362 | 0.173 | 0.196 | 1.426 | 0.154 |
| ROA | ? | −0.035 | −0.035 | 0.972 | −0.138 | −0.150 | 0.881 | 0.333 | 0.292 | 0.771 |
| ANCOV | ? | 0.825 | 1.718 | 0.086 | 0.807 | 1.673 | 0.094 | 0.803 | 1.675 | 0.094 |
| LOSS | ? | 0.376 | 0.432 | 0.666 | 0.228 | 0.273 | 0.785 | 0.121 | 0.137 | 0.891 |
| McFadden R-squared | 0.147 | 0.135 | 0.137 | |||||||
| Likelihood ratio statistic | 22.280 | 20.444 | 20.701 | |||||||
| p-value | 0.001 | 0.002 | 0.002 | |||||||
| n | 118 | 118 | 118 | |||||||
- DISC denotes a binary variable equal to 1 if operating income/EBIT is disclosed in the income statement, 0 otherwise. AIP denotes book value of property, plant and equipment divided by the sum of market capitalization of equity, minority interest and liabilities. MKTBVA denotes market value of total assets (found by taking the sum of market capitalization of equity, minority interest and liabilities) divided by the book value of total assets. NOPAT/MKTA denotes net operating profit after tax divided by the market value of total assets. LEV denotes total liabilities divided by total assets. SIZE denotes natural log of total assets. ROA denotes net operating profit after tax divided by total assets. ANCOV denotes a binary variable equal to 1 if financial analysts provide earnings forecast for the firm, 0 otherwise. LOSS denotes a binary variable equal to 1 if firm makes a loss, 0 otherwise.
- •
The mean AIP for firms that disclose EBIT in the income statement is 0.302, while the mean AIP for firms not making this disclosure is 0.205. The t-test indicates that the difference is significant at the 5 per cent level, while the Mann–Whitney U-test indicates significance at the 1 per cent level;
- •
The mean MKTBVA for firms that disclose EBIT in the income statement is 1.901, while it is 2.694 for firms not making this disclosure. The t-test indicates that the difference is significant at the 5 per cent level, but it is insignificant when the Mann–Whitney U-test is used; and
- •
The mean NOPAT/MKTA for firms that disclose EBIT in the income statement is 0.024, while the mean NOPAT/MKTA for firms not making this disclosure is 0.008. Both a t-test and a Mann–Whitney U-test indicate that the difference is not significant.
The mean LEV for firms that disclose EBIT in the income statement is 0.484, while the mean LEV for firms not making this disclosure is 0.346. Both the t-test and the Mann–Whitney U-test indicate the difference is significant at the 1 per cent level.
The mean SIZE (as measured by the natural log of total assets) for firms that disclose EBIT in the income statement is 12.105, while the mean SIZE for firms not making this disclosure is 10.742. Both the t-test and the Mann–Whitney U-test indicate the difference is significant at the 1 per cent level. The mean ROA for firms that disclose EBIT in the income statement is 0.045, while the mean ROA for firms not making this disclosure is −0.028. The t-test indicates that the difference is significant at the 5 per cent level. However, the Mann–Whitney U-test shows that the difference is not significant.
Panels B and C of Table 2 provide descriptive statistics for the analyst coverage (ANCOV) and loss companies (LOSS), respectively. They also provide an analysis of the relation between firms disclosing EBIT in the income statement and ANCOV and loss companies. For ANCOV, the results indicate that analysts provide an earnings forecast for 70 per cent (i.e. 28 of 40 companies) of the firms that disclose EBIT, while it is about 39 per cent (i.e. 30 of 78 companies) for firms that do not disclose EBIT. The results also show that there exists a relation between firms disclosing EBIT in the income statement and ANCOV. The chi-squared statistic of this relation is 10.523, which is significant at the 1 per cent level (two-tailed test). For LOSS, the results show that about 13 per cent of the companies that disclose EBIT are loss making firms (i.e. five of 40 companies), while about 27 per cent of the companies that do not disclose EBIT make a loss (i.e. 21 of 78 companies). The chi-squared statistic of the relation between firms disclosing EBIT in the income statement and loss companies is 3.202, which is significant at the 10 per cent level (two-tailed test).
In general, the univariate results reported in Table 2 indicate that a significant difference across firms that disclose EBIT in the income statement and those that do not: (i) exists for AIP and not for our two other measures of the IOS (i.e. MKTVBVA and NOPAT/MKTA); (ii) exists for LEV; (iii) exists for SIZE, ANCOV and LOSS; and (iv) does not exist for ROA.
Table 3 reports the correlations for the independent variables used in the regressions. In particular, we calculate Pearson correlations (shown above the diagonal) and the Spearman correlations (shown below the diagonal) for our independent variables. The only exception to this is the correlation between ANCOV and LOSS, where we calculate and report the phi-coefficient, which is appropriate when the two variables in the calculation are both categorical in nature and can only take on one of two values.
| AIP | MKTBVA | NOPAT/MKTA | LEV | SIZE | ROA | ANCOV | LOSS | |
|---|---|---|---|---|---|---|---|---|
| AIP | −0.315 (0.001) | 0.103 (0.269) | 0.081 (0.385) | 0.328 (0.000) | 0.155 (0.094) | 0.170 (0.059) | −0.270 (0.003) | |
| MKTBVA | −0.349 (0.000) | −0.092 (0.322) | −0.268 (0.003) | −0.346 (0.000) | −0.297 (0.001) | −0.160 (0.083) | 0.200 (0.028) | |
| NOPAT/MKTA | 0.223 (0.015) | −0.172 (0.063) | 0.173 (0.061) | 0.323 (0.000) | 0.668 (0.000) | 0.210 (0.023) | −0.590 (0.000) | |
| LEV | 0.264 (0.004) | 0.028 (0.760) | 0.101 (0.276) | 0.388 (0.000) | 0.263 (0.004) | 0.280 (0.002) | −0.320 (0.000) | |
| SIZE | 0.406 (0.000) | −0.264 (0.004) | 0.426 (0.000) | 0.380 (0.000) | 0.444 (0.000) | 0.540 (0.000) | −0.560 (0.000) | |
| ROA | 0.165 (0.075) | 0.174 (0.060) | 0.826 (0.000) | 0.220 (0.017) | 0.410 (0.000) | 0.280 (0.003) | −0.640 (0.000) | |
| ANCOV | 0.274 (0.003) | 0.025 (0.789) | 0.353 (0.000) | 0.274 (0.003) | 0.561 (0.000) | 0.398 (0.000) | −0.359 (0.000) | |
| LOSS | −0.366 (0.000) | 0.182 (0.049) | −0.717 (0.000) | −0.301 (0.001) | −0.567 (0.000) | −0.717 (0.000) | −0.359 (0.000) |
- AIP denotes book value of property, plant and equipment divided by the sum of market capitalization of equity, minority interest and liabilities. MKTBVA denotes market value of total assets (found by taking the sum of market capitalization of equity, minority interest and liabilities) divided by the book value of total assets. NOPAT/MKTA denotes net operating profit after tax divided by the market value of total assets. LEV denotes total liabilities divided by total assets. SIZE denotes natural log of total assets. ROA denotes net operating profit after tax divided by total assets. ANCOV denotes a binary variable equal to 1 if financial analysts provide earnings forecast for the firm, 0 otherwise. LOSS denotes a binary variable equal to 1 if firm makes a loss, 0 otherwise. Pearson correlations appear on the upper diagonal, while Spearman correlations appear on the lower diagonal. The only exception to this is for the correlation between ANCOV and LOSS, which is calculated by the phi-coefficient. Two-tailed p-values appear in parentheses.
The correlations shown in Table 3 are generally not high, although many of the results are significant. Most surprising are the low correlations between each of our three IOS measures (i.e. AIP, MKTBVA and NOPAT/MKTA). Specifically, the Pearson (Spearman) correlation between AIP and MKTBVA is −0.315 (−0.349), between AIP and NOPAT/MKTA is 0.103 (0.223) and between MKTBVA and NOPAT/MKTA is −0.092 (−0.172). The correlations between each of the IOS measures, respectively, with LEV are also not high. For AIP and LEV, the Pearson (Spearman) correlation is 0.081 (0.264), for MKTBVA and LEV it is −0.268 (0.028) and for NOPAT/MKTA it is 0.173 (0.101). The positive Spearman correlation between MKTBVA and LEV is unusual, although it is not significant.
In summary, the low association between the variables shown in Table 3 suggests that multicollinearity is not expected to be a major concern in the estimation of the logit regressions that follow.
Table 4 presents the results of the logit regression model that we estimate, as specified in equation (1) (i.e. using AIP as the measure of the IOS). These results are represented in Table 4 under the heading Regression 1. Regression 1 shows that the coefficient of the AIP variable is positive, as expected, and it is significant at the 10 per cent level (one-tailed test). While this result is not strong, it does provide some support for hypothesis 1 that assets-in-place are positively related to the voluntary disclosure of operating income/EBIT in the income statement. The coefficient of the LEV variable is also positive, as expected, and it is significant at the 1 per cent level using a one-tailed test. This finding provides support for hypothesis 2, which says that leverage is positively related to the voluntary disclosure of operating income/EBIT in the income statement.
Of the control variables included in Regression 1 (i.e. SIZE, ROA, ANCOV and LOSS), only ANCOV is significant (at the 10 per cent level using a two-tailed test). The sign of the coefficient on the ANCOV variable is positive, indicating that if financial analysts provide earnings forecasts for a firm, there is a higher probability that the firm discloses operating income/EBIT in the income statement. Regression 1 appears to be well specified, with the McFadden R-squared measure being 14.7 per cent and the likelihood statistic being significant at the 1 per cent level (two-tailed test).
To investigate the sensitivity of the results to alternative measures of the IOS, we also re-estimate Regression 1 by replacing the AIP variable with MKTBVA and NOPAT/MKTA, respectively. These regressions are reported in Table 4 as Regression 2 and Regression 3.
Regression 2 estimates the regression with MKTBVA. The results show that the coefficient estimate of MKTBVA is negative, as expected, but it is not significant. However, the coefficient of the LEV variable continues to be positive and significant at the 1 per cent level (using a one-tailed test). This finding is similar to the result in Regression 1 and again provides support for hypothesis 2. The only other variable that is significant in Regression 2 is the ANCOV variable (at the 10 per cent level for a two-tailed test), which again shows a positive relation with the voluntary disclosure of operating income/EBIT. The McFadden’s R-squared measure for Regression 2 is 13.5 per cent with the likelihood ratio statistic being 20.444 and significant at the 1 per cent level.
Regression 3 in Table 4 replaces the AIP variable in Regression 1 with NOPAT/MKTA. The results show that this variable has a negative sign on its coefficient estimate, which is unexpected, but it is also not significant. However, the coefficient of the LEV variable continues to have the expected positive sign and to be significant (again at the 1 per cent level using a one-tailed test). As in Regressions 1 and 2, Regression 3 also shows that ANCOV is significantly positive at the 10 per cent level using a two-tailed test, indicating once more that firms where a financial analyst provides an earnings forecast are more likely to disclose voluntarily operating income/EBIT. The explanatory power of Regression 3 is similar to that of Regressions 1 and 2 – the McFadden R-squared measure is 13.7 per cent and the likelihood ratio statistic of 20.701 is significant at the 1 per cent level.
In general, the logit regression results in Table 4 provide: (i) support for hypothesis 1 when the AIP variable is used to measure the IOS and (ii) support for hypothesis 2 that LEV is associated with the voluntary disclosure of operating income/EBIT. Overall, this study provides some zevidence for a phenomenon that has not been documented previously in the accounting literature.3
4. Conclusion
This paper extends existing research on managers’ discretionary disclosure. Previous IOS research has not related the IOS to discretionary disclosure, and previous disclosure research has not related disclosure to the IOS. To our knowledge, apart from Leftwich et al. (1981), who do not find results consistent with an IOS–disclosure relation, and Hossain et al. (2005), who find that firms’ voluntary disclosure of prospective information and their IOS are determined simultaneously, this study provides further empirical evidence that the IOS drives discretionary disclosure of operating income/EBIT in the income statement. Watts and Zimmerman (1990) suggest that there may be a link between the IOS and accounting performance measurement systems, and we have been able to provide evidence of this.
Our overall prediction is that managers are more likely to voluntarily disclose operating income/EBIT in the income statement where operating income/EBIT presents an appropriate measure of performance to the firm’s capital providers, thereby facilitating efficient contracting between the parties to the firm. Consistent with this prediction, we find that firms with high assets-in-place and, hence, high leverage are more likely to voluntarily disclose operating income/EBIT, the pre-tax measure of NOPAT that is used by capital providers to assess manager’s performance. However, the finding for assets-in-place is sensitive to alternative measures of the IOS and that when alternative measures are used to proxy for this construct, the relation is no longer significant. Despite this, the relation between leverage and the voluntary disclosure of operating income/EBIT is significant and robust in all the analyses that we conduct. Taken together, the findings in our study contribute to the literature on how the IOS drives policy selection and discretionary disclosure.
