4 Ownership, competition, and financial disclosure.

Abstract:

A firm's incentive to disclose has been linked empirically to a range of variables, including information asymmetry, agency costs, political costs, and proprietary costs. While the intuition underlying each of the variables seems plausible, Verrecchia (2001) argues that disclosure models can be characterized as an eclectic mingling of highly idiosyncratic economic-based models, and challenges researchers to take the first steps to unification. First, we investigate the role of ownership and competition variables in explaining voluntary segment disclosures in Australian firms and find support for both these variables. Second, drawing on theory supported by the corporate governance, strategic management and industrial organization literatures, we introduce a new economic variable that unifies both ownership and competition variables. We find that the unifying variable performs better than our model focusing on ownership and competition variables alone. We conduct a series of robustness tests on the model and find that its significance is not affected by the inclusion of disclosure control variables identified in prior literature, the change in standard, and acquisitions and disposals of physical assets.

Keywords:

VOLUNTARY DISCLOSURES; OWNERSHIP; COMPETITION.

1. Introduction

Verrecchia (2001) categorizes disclosure research into three broad groups--association-based, discretionary-based, and efficiency-based. Association-based research investigates the relation between exogenous disclosure and change in investors' individual actions. Discretionary-based research investigates how firms use their discretion regarding information that does not require mandatory disclosure. Efficiency-based research examines unconditional disclosure choices characterized by endogenous consumers.

The focus of this paper is on the discretionary-based research, which models the firm's incentives or disincentives to disclose as a function of a range of variables including information asymmetry, agency costs, political costs, and proprietary costs. (1) Verrecchia (2001) argues that this disclosure literature can best be characterized as an eclectic mingling of highly idiosyncratic economic-based models, and challenges researchers to take the first steps at unification. First, we investigate the role of ownership and competition variables in explaining voluntary segment disclosures in Australian firms. Second, we introduce a new economic variable OC, that unifies both the ownership and competition variables.

Previous disclosure studies have found significant results for variables originating from agency, political cost, information asymmetry and proprietary cost theories. Agency theory suggests that as a means of mitigating divergent interests between principals and agents, firms may use different methods, which include bonus share plans, performance-based contracts and voluntary disclosures. Further, the corporate governance literature suggests that large shareholders (e.g. institutional investors) play an active role in the monitoring and control of firms and have an implicit obligation to other shareholders in ensuring that firms are run in the best interests of all shareholders. The large shareholders have a greater willingness to discipline poorly performing management and more incentive to intervene and exercise 'voice' (Mayer 1997). Thus, large shareholders have the ability to mitigate the agency problems inherent in a firm by influencing the voluntary disclosures made by the firm. We acknowledge these findings and construct a disclosure model testing our ownership variable O.

Theories that explain a firm's decision not to disclose, such as proprietary cost theory; consider costs that arise when the disclosure of private information may harm the firm's competitive position. Firms operating in a low-competition environment have less incentive to disclose private information, as it has more potential to harm their competitive position. Conversely, firms operating in a highly competitive environment may have greater incentive to disclose, as there is potentially less risk to their competitive position and, in fact, the release of additional information could benefit the firm by reducing information asymmetries between management and the shareholders (Hayes & Lundholm 1996; Harris 1998; Botosan & Stanford 2005). This release of additional information provides shareholders and other users with information to better validate the results of the firm. We acknowledge these findings and construct a disclosure model testing our competition variable C.

After examining the roles of ownership and competition variables on disclosure, we then examine the role of a new variable which unifies both ownership and competition. Our OC variable draws upon the corporate governance, strategic management and industrial organization literatures. Corporate governance suggests an interaction between ownership and competition, as the impact of corporate governance mechanisms may be influenced by the degree of market competition (Mayer 1997; Januszewski, Koke & Winter 2001; Aoki 1994; Aghion & Howitt 1996). Further, shareholders are more able to monitor the actions of management when firms operate in a competitive environment. (Nickell 1996; Nickell, Nicolitsas & Dryden 1997).

The strategic management and industrial organization literatures also suggest an interaction between ownership and competition. These literatures refer to the success of a firm depending on the internal environment such as ownership structure and incentive schemes and the external environment in which the firm operates, such as competition. Porter (1981) argues that a successful firm must match its internal competencies and values to its external environment, and Saloner (1991) adds that internal issues and external issues are important inputs to a firm's decision-making approach.

Our unifying variable OC is measured as the product of ownership and competition. We use the percentage of shares held by the top 20 shareholders to proxy for ownership, and we measure the level of industry competition as 1 minus the Herfindahl index of industry concentration. We test whether this new variable enhances our ability to explain voluntary disclosure decisions. To test our hypothesis, we focus on voluntary segment disclosures, because they have been found to be value-relevant in forecasting sales and profits. (2) Our empirical analyses use firms' segment disclosures for 2001 under the original Australian Accounting Standards Board (AASB) standard and for 2002/2003 under the revised standard. This unique regulatory background featuring a change in segment reporting standard provides us with an additional test of robustness for the model featuring our OC variable. Our results demonstrate that both the O and C variables individually are statistically significant in explaining voluntary disclosures. Further, we find strong results that the OC variable does in fact enhance the ability of the model to explain voluntary disclosure. We conduct a series of robustness tests on our OC variable and find that the OC variable is robust to the inclusion of variables measuring the change in standard and acquisitions and disposals of other firms.

A greater understanding of the incentives to disclose financial information is timely, as jurisdictions worldwide are currently undergoing international harmonization projects with accounting standards which will result in changes to disclosure practices. On January 1, 2005, Australian equivalents were adopted of the International Financial Reporting Standards (IFRS), and the Financial Accounting Standards Board (FASB) is currently committed to a project with the IASB to harmonize it's standards.

The remainder of the paper is organized as follows. Section 2 outlines the institutional setting of AASB 1005 Segment Reporting. Section 3 summarizes the main hypotheses regarding discretionary disclosures. Section4 outlines our methodology and empirical measures, and section 5 provides the descriptive statistics. Section 6 reports the results of our empirical tests. Section 7 contains a discussion of the robustness tests, and section 8 describes the main conclusions and suggests avenues for future research.

2. Institutional Setting

Segment reporting refers to the disclosure of results from operating in markets with different rates of profit, different degrees of risk, and different opportunities for growth. Segment reporting disclosures are useful for investment decision-making. Disaggregated industry and geographic segment data can provide analysts and investors with important, incremental information about the different markets within which the company operates. (3) Local and international standards require the disclosure of information regarding business and geographic segments. The Australian Accounting Standards Board (AASB) has released two standards on segment reporting: AASB 1005 Financial Reporting by Segments and the revised AASB 1005 Segment Reporting (4). The current U.S. standard is SFAS 131 Reporting Disaggregated Information about a Business Enterprise, and the current international standard is IAS 14 Segment Reporting.

2.1 Segment Reporting Standards in Australia Australia's original segment standard AASB 1005 Financial Reporting by Segments was released in 1986 and required firms to disclose segment revenue, segment result, and the carrying amount of segment assets for both industry and geographical segments. In August 2000, the AASB issued the revised standard AASB 1005 Segment Reporting whereby firms are required to identify their segments in line with their internal organizational structure and internal reporting system. This approach, known as the 'management approach,' (5) differs markedly from the original 'industry approach'. The revised standard also allows firms to choose whether line of business (LOB) or geographic area (GEO) will be a primary or a secondary segment disclosure. (6) The required disclosures for primary segments are much more extensive than for secondary segments. The standard stipulates that firms' must disclose segment revenue, segment profit, segment assets, segment liabilities, acquisition of segment assets, depreciation and amortization of segment assets, other non-cash segment expenses, segment share of the net profit/result of associates or other investees and segment carrying amount of investments in the associates. The revised standard also encourages voluntary disclosure of additional information such as segment cash flows and segment inventory write-downs. For secondary segments, the standard requires disclosure of segment revenues, the carrying amount of assets, and the cost of property, plant and equipment, and intangible assets acquired during the period. A segment's profit is not a required disclosure for secondary segments, under the revised AASB 1005.

2.2 International Harmonization of Segment Standards

In recent years, there has been a trend towards increasing global comparability in financial reports. In 2005, Australian companies preparing financial reports under the Corporations Act 2001 must comply with the Australian equivalents to the IFRS of the IASB for financial years beginning on or after January 1, 2005. While the U.S. is not adopting IFRS, it is an important partner to the IASB and is working on projects to reduce the differences in FASB and IASB standards. The implementation of IFRS and the harmonization of standards internationally will have repercussions for companies in the preparation and presentation of financial reports.

3. Theoretical Framework

Positive accounting theory deals with management's motives in making accounting choices. Within this framework, disclosure research focuses on the role of capital market incentives in the firm's disclosure decisions. Verrecchia (2001) categorizes disclosure research into three broad groups. The first, association-based research, investigates the relation between exogenous disclosure and the change in investors' individual actions. The second, discretionary-based research, investigates how firms use their discretion in revealing information when reporting is not mandated. The third, efficiency-based research, examines unconditional disclosure choices characterized by endogenous consumers.

Discretionary-based disclosure research, the focus of this study, considers the incentives and disincentives for disclosing additional financial information in a capital market setting. Incentives to disclose include: (a) mitigating the affects of information asymmetry; (b) decreasing potential political costs; and, (c) monitoring agents to reduce agency costs. The primary disincentive for disclosing additional financial information is potential proprietary costs. Table 1 summarizes past research in terms of hypotheses tested and test results. While these studies have found significant results for a number of variables, to our knowledge no study has yet tested the significance of a variable unifying different theories of discretionary disclosure.

3.1 Information Asymmetry Hypothesis

Informational asymmetry impedes the efficient allocation of resources. It arises when markets do not perfectly aggregate private information, and can lead to higher transaction costs, lower liquidity, and, ultimately, mis-pricing of the firm's shares. The effects of information asymmetry can be mitigated in a number of ways, including contracting, regulation through information intermediaries. Accounting disclosures are also a means of disseminating information to less informed parties. Several studies have examined the role of information asymmetry proxies and the presence of voluntary disclosure. Botosan (1997), for example, uses analyst following to proxy for information asymmetry and finds that firms with lower analyst following have a propensity for higher disclosure, and, consequently, experience a reduction in their costs of capital. For firms with higher analyst following, she finds no significant relation. Past studies also find that size is related to the level of information asymmetry. Atiase (1985), Bamber (1987), and Diamond and Verrecchia (1991) find information is incorporated in stock price more quickly for large firms than small firms. In the same spirit, King, Pownall and Waymire (1990) argue that the incentives for disclosure are greater for larger firms. Studies also investigate the relation between ownership and information asymmetry. Healy, Hutton and Palepu (1999) find a link between increased disclosure and higher percentages of institutional ownership. Jiambalvo, Rajgopal and Venkatachalam (2002) find that current earnings are more likely to reflect future earnings when firms' have high percentages of institutional ownership. These findings are further corroborated by evidence regarding the relation between disclosure and diffused ownership. Mitchell, Chia and Loh (1995) and Aitken, Hooper and Pickering (1997), for example, conclude that voluntary disclosures vary directly with the percentage of significant owners.

3.2 Political Costs

Political costs may also explain voluntary disclosure decisions. Belkaoui and Karpik (1989) find that firms employ a number of devices (including voluntary disclosures) to avoid the attention of external parties such as government regulators, suppliers, and unions. (7) Deegan and Gordon (1996) find that firms with high political visibility in the marketplace increase disclosures as a means of mitigating potential political costs. Distinguishing between the political cost hypothesis and other disclosure theories is often difficult, however. Political costs are usually measured using firm size, which, as noted earlier, is often used as a proxy for information asymmetry.

3.3 Agency Costs

Agency costs arise when principals and agents have conflicting incentives. (8) As a means of mitigating divergent interests, principals may use different incentives to monitor their agents. The possibilities include performance-based contracts, bonus share plans, debt covenants, audit committees, as well as increased disclosure. Bradbury (1991) and Chow and Wong-Boren (1987) document that firm characteristics such as firm size, leverage and fixed assets in place affect voluntary disclosures by influencing the degree of agency and contracting costs experienced by the firm. Holthausen and Leftwich (1983) find support for firm-size and leverage variables.

Several segment disclosure choice studies provide evidence consistent with the Holthausen and Leftwich findings. Foster (1986), for example, notes that firm size is the most commonly-used control variable in disclosure studies, see for example McKinnon and Dalimunthe (1993), Bradbury (1992), and Berger and Hann (2002). Studies also report a positive relation between disclosure and being audited by an international auditing firm. Street and Gray (2001) argue that a 'big four or five' auditor encourages firms to be forthcoming in their disclosures as part of the monitoring process associated with reducing agency costs. Past research also suggests that the proportion of fixed assets in place and voluntary disclosure are related. Bradbury (1992) and Chow and Wong-Boren (1987) argue that firms with large proportions of fixed assets in place are expected to experience lower agency costs, and, consequently, will have less incentive for voluntary disclosures.

Studies also examine the relation between ownership structure and disclosure using explanatory variables such as directors' shareholdings and CEO remuneration. Jensen and Meckling (1976) argue that agency costs increase with the proportion of outside capital. Nagar, Nanda and Wysocki (2003) find a positive relation between disclosure and the level of CEO share ownership. In firm performance studies, ownership structure plays an important, albeit empirically ambiguous, role. Berle and Means (1932) find an inverse correlation between ownership and firm performance. Demsetz (1983), on the other hand, argues there should be no relation between variation in ownership and firm performance because the ownership structure of a firm is a multidimensional variable and should be seen as an endogenous outcome of decisions that reflect the influence of shareholders. In the burgeoning corporate governance literature, studies have shown that the large investors (e.g. the top 20 shareholders) play an active role in the monitoring and control of firms. These shareholders have greater willingness to discipline poorly performing management and more incentive to intervene and exercise voice (Mayer 1997).

3.4 Proprietary Costs

The discretionary disclosure literature also considers theories that explain a firm's decision not to disclose. Dye (2001) posits that, if disclosure is discretionary, firms will choose to disclose favourable information and not to disclose unfavourable information. (9) Related to this are proprietary costs. Proprietary costs arise when private information, if released, may harm the firm's competitive position. Verrecchia (2001) examines the role of proprietary costs in explaining a firm's decision to withhold the release of additional information.

Segment information is important to users of financial reports. Firm operations can vary significantly across line of business and geographic segments, and firm segments can vary according to the rates of profit, levels of risk, and opportunities for growth. Segment disclosures contain value relevant information that may help investors and analysts predict future profits and revenues. At the same time, segment disclosure information may be useful to external, and potentially adversarial, parties such as suppliers, employees, unions and competitors. Consequently, management must exercise discretion, taking into consideration the impact of the market release of potentially harmful information.

In determining an appropriate level of disclosure, firms must, therefore, consider factors such as the competitiveness of the industry in which they operate. Hayes and Lundholm (1996) predict that managers alleviate competitive costs through non-disclosure. The empirical evidence regarding the relation between competition and disclosure is mixed. Verrecchia (1983) and Wagenhofer (1990), on one hand, find that firms in more competitive industries provide less informative disclosures. On the other, Harris (1998) and Botosan and Stanford (2005) find that operations in less competitive industries are less likely to be reported as industry segments. This suggests that managers attempt to conceal information that may allow rival firms to capture these profits. Hayes and Lundholm (1996) find that a firm disaggregates consolidated information into segment information in a highly competitive environment in order to reduce information asymmetries.

Firm performance is another determinant in the decision to disclose. The empirical studies on the relation between firm performance and disclosure are mixed. Lev and Penman (1990) suggest that firms tend to be more forthcoming when the firm is experiencing favorable earnings results. Lang and Lundholm (1993) show otherwise. Prencipe (2004) shows that the competitive costs associated with segment disclosures, however, tend to increase as the profitability of the reporting entity increases. Berger and Hann (2002) also find that firms aggregate segment information when there are large variances in segment profits in order to protect abnormal profits.

On balance, the literature appears to support the position that firms with low competition have higher proprietary costs (and greater potential to make abnormal profits), and, consequently, have less incentive to disclose proprietary information to rivals. Firms in more competitive industries, however, have greater incentive to disclose information in order to reduce information asymmetries.

4. Effect of Ownership and Industry Competition

This paper first investigates the role of ownership and competition variables in explaining voluntary segment disclosures in Australian firms. Second, in the spirit of Verrecchia (2001), we attempt to improve existing economic models of voluntary disclosure by introducing our OC variable that unifies both ownership and competition variables.

Support for our ownership variable is found in the agency and corporate governance literature. Agency theory suggests that as a means of mitigating divergent interests, principals may use different incentives to monitor their agents which include bonus share plans, performance-based contracts and increased disclosures. Studies have examined the relation between ownership structure and agency costs (Jensen & Meckling 1976), and ownership structure and voluntary disclosures (Nagar, Nanda & Wysocki 2003) and found a positive relationship between disclosure and the level of CEO share ownership. Further, corporate governance literature suggests that large shareholders (e.g. institutional investors) play an active role in the monitoring and control of firms and have an implicit obligation to other shareholders in ensuring that firms are run in the best interests of all shareholders. The large shareholders have a greater willingness to discipline poorly performing management and more incentive to intervene and exercise 'voice' (Mayer 1997). Thus, large shareholders have the ability to mitigate the agency problems inherent in a firm by influencing the voluntary disclosures made by the firm. We acknowledge these findings and construct a disclosure model testing our ownership variable O.

Support for our competition variable is found in the proprietary cost literature. Studies have found that firms with low competition have higher proprietary costs (and greater potential to make abnormal profits), and, consequently, have less incentive to disclose proprietary information to rivals. Additionally, the literature suggests that firms in more competitive industries have greater incentive to disclose to reduce information asymmetries between management and the shareholders (Harris 1998; Botosan & Stanford 2005). This release of additional information provides shareholders and other users with information to better validate the results of the firm. We acknowledge these findings and construct a disclosure model testing our competition variable C.

Our study also introduces a new economic variable OC which unifies both ownership and competition variables. This unified variable is implied through the corporate governance literature which suggests a possible interaction between ownership and competition. Mayer (1997) argues that the effectiveness of difference types of governance systems (e.g. ownership structure of the firm) may be influenced by the degree of product market competition. This interaction is also echoed in the industrial organisation literature where Nickell (1996) and Nickell, Nicolitsas & Dryden (1997) suggest that shareholders are more able to monitor the actions of management when firms operate in a competitive environment. Further support for our OC variable is suggested in the strategic management literature where Porter (1981) argues that a successful firm must match both its internal and external environments. This interaction is also suggested in the industrial organization literature where Tirole (1990) and Schmalensee and Willig (1989) describe how decisions concerning ownership structure can impact on the firm's competitive position in the marketplace. (10) Further, Saloner (1991) adds that external factors such as competition, combined with internal factors such as incentive schemes, are important inputs to a firm's decision-making approach.

Thus, we propose that a management's decision to voluntarily disclose segment information depends on the ownership structure of the firm and the competition environment in which the firm operates. Further we propose that the decision can depend jointly on these two factors. We measure the firm's large shareholders using the percentage of shares owned by the top 20 shareholders O. We theorize that, with high levels of O, the large shareholders will influence a firm's voluntary disclosure choices. To proxy for the competition of the firm, we use the degree of industry competition faced by the firm, as measured by 1 minus the Herfindahl index, (1 - HI). (11) The HI for each industry is calculated for 36 industries using the top 500 firms on the Australian Stock Exchange. The Herfindahl index for industry j is

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

where [R.sub.ij] is the revenue of firm i in industry j (as defined by the 4-digit SIC code), [n.sub.j] is the number of firms in industry j, and

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

is the total of revenue for all firms in industry j. Firms operating in more competitive industries (1 - HI) are expected to have more incentive to disclose.

Combining elements, we hypothesize that the product of large shareholders and high competition, OC, is positively related to the amount of segment disclosure. This model can be expressed as;

VD = f(OC) (2)

Where VD is voluntary disclosure and OC represents the product of the ownership variable O and the competition variable C. In model (2), the change in disclosure is modelled as the interaction of both ownership and competition.

The appropriateness of this specification can be determined by comparison with the following models;

VD = f(O) (3)

VD = f(C) (4)

VD = f(O, C) (5)

These models present the alternative views that voluntary disclosure is determined solely by either ownership O (as in the case of model 2) or competition C (as in the case of model 3), or by a combination of the independent influences of ownership and competition (model 5).

Finally, we include both the ownership O and competition C variables and our interaction variable OC to the model. Thus,

VD = f(OC, O, C) (6)

If our hypothesis is correct, that it is the interaction of ownership and competition which determines voluntary disclosure, then model (2) should dominate models (3), (4), (5) and (6).

Next, we explicitly consider the voluntary disclosure work of past investigators. Section 3 and table 1 indicate that firm size, leverage, fixed assets in place and cross listing are the variables that most frequently are found to be associated with voluntary disclosures. We also include auditor and profitability which have also been found to be significant. Therefore we propose the following model based on the current state of the literature:

VD = f(RETURN, FAIP, AUDITOR, LEV, LOGTA, CROSS) (7)

where VD is voluntary disclosure, RETURN is stock return, FAIP is fixed assets in place, AUDITOR is an indicator variable for the prestige of the firm's auditor, LEV is the firm's debt ratio, LOGTA is the natural logarithm of total assets and CROSS indicates whether a firm is cross-listed.

We hypothesise that our O variable has explanatory power in explaining voluntary segment disclosures. We also hypothesise that our C variables has significant explanatory power. We then turn to our unifying, interaction variable OC and hypothesis that OC will provide significant incremental explanatory power. Our measures of the explanatory variables are as follows: (a) stock return, RETURN, is the annualised logarithmic stock return including dividends and price appreciation; (b) fixed assets in place, FAIP, is measured as the book value of fixed assets relative to total assets; (c) AUDITOR is a dummy variable set equal to 1 if the firm's auditor was a 'big five' firm in 2001 or a 'big four' firm in 2002 or 2003, and 0 otherwise; (d) leverage, LEV, is the book value of debt divided by market value of equity and the book value of debt; (e) LOGTA is the natural logarithm of total assets; and, (f) CROSS is an indicator variable which takes a value of 1 if a firm is cross-listed and zero otherwise.

Therefore, to test our hypothesis we consider the following model,

VD = f(OC, RETURN, FAIP, AUDITOR, LEV, LOGTA, CROSS) (8)

We also test both the ownership O and competition C variables individually in this model. Thus,

VD = f (O, C, RETURN, FAIP, AUDITOR, LEV, LOGTA, CROSS) (9)

Finally we test our OC interaction variable and the O and C variables individually alongside the previously tested variables. Thus,

VD = f(OC, O, C, RETURN, FAIP, AUDITOR, LEV, LOGTA, CROSS) (10)

5. Empirical Measures, Sample Selection and Descriptive Statistics

5.1 Data

To examine firms' segment reporting practices, we use the Connect 4 database to access financial reports for the Top 500 Australian companies for the years 2001 through 2003. Segment reporting information, if disclosed, is located in the notes of the financial reports. Of the Top 500 reports examined for the year 2001, 263 disclosed segment information. Under the revised standard in 2002 and 2003, 276 and 286 firms disclosed segment information, respectively, bringing the total to 825 for the entire sample.

As noted earlier, we measure competition using the Herfindahl index. The index is based on the industry groups of the entire sample of Top 500 Australian firms. We categorize the Top 500 Australian firms according to the 36 Global Industry Classification Scheme (GICS) four-digit industry group from the Centre for Research in Finance (CRIF). Table 2 provides a breakdown of the sample and their relevant industries for 2001 through 2003. In no year does a single industry dominate in our assignment of the dependent variable.

5.2 Voluntary Segment Disclosure Variables

The dependent variable used in our model, VD, is a dichotomous variable. Its value is set equal to 1 in 2001 if the firm reported other disclosures aside from the required revenue, result and segment assets. All other firms are coded 0. For the years, 2002 and 2003, VD reflects disclosures that are in addition to the required nine primary items and three secondary items as outlined in section 2.1. Firms making such disclosures are coded 1. All others are coded 0.

5.3 Descriptive Statistics

Table 2 also provides statistics for the voluntary disclosure variable, VD. In 2001, 107 of 263 or 41% of firms provided voluntary segment disclosures. In 2002 and 2003, the percentages were 51 and 67, (12) respectively. Voluntary segment disclosures include items such as: additional segment revenue, amortization of goodwill, write-down of inventory, income tax, significant items, segment bad debts, and segment cash flow from operating activities. Table 3 presents the descriptive statistics of the independent variables for the pooled sample, 2001 through 2003. The mean percentage of shares held by the top 20 shareholders variable is 62.59% across the observations in the sample. The competition variable as measured by 1 minus the Herfindahl index for each industry is 66.53%, and the mean of the product of ownership and competition OC, is 41.66%. More than 82% of the firms had a big 4/big 5 auditor. The average debt ratio is 44.85%, and the average fixed assets in place variable is 59.86%. Approximately 13.21% of firms in the sample were cross-listed and the mean annualised stock return is about -4%.

5.4 Correlation Matrix 

Table 4 contains the pair-wise correlation coefficients between the model variables using the full sample period, 2001 through 2003. Naturally, the interactive variable OC and its components, O and C, are highly positively correlated with correlations of 0.793 and 0.587 respectively. The LOGTA variable is highly correlated with LEV (0.529). The correlation with FAIP is 0.302, and with the product variable, OC, is -0.115. Apart from LOGTA, the OC variable is relatively weakly correlated with the other variables. The lack of correlation between OC and the other independent variables in model (3) mitigates possible concerns about the effects of multi-collinearity in model estimation.

6. Empirical Tests and Results

The focus now turns to examining the role of ownership and competition variables and the incremental contribution of our measure of the interaction variable of the firm, OC, in explaining voluntary segment disclosures, VD. We begin by examining the relation between VD and O and C. We then turn to identifying the contribution of the OC variable.

6.1 Examining the Role of Ownership, Competition and the Interaction OC Variable

Our first test involves regressing voluntary segment disclosures VD on the ownership variable O using a pooled time-series, cross-sectional probit model for the three years of data. The results of the probit regression are reported in Table 5 and are significant in a statistical sense (t-ratio = 1.97). This evidence suggests that firms having high levels of shares owned by top 20 shareholders are more likely to disclose voluntary segment items. Our second test involves regressing voluntary segment disclosures VD on the competition variable C for the three years of data. The results of the probit regression are also reported in table 5 and show that VD and the C are positively related and significant in a statistical sense (t-ratio = 2.86). Our third test involves including both the O and C variables and the results show that these variables are both significant (t-ratio =1.97) and (t-ratio =2.86) respectively. Fourth we identify the contribution of our interaction variable OC alone and as predicted, VD and the interaction variable OC are positively related, and highly significant in a statistical sense (t-ratio =3.30). We also conduct a fifth test which involves regressing VD on each of the variables, that is, O, C and OC and we find that O and C fail to add anything to OC by itself. This can be seen in the results of the likelihood ratio test statistic (LR) which is calculated as:

LR = -2(RLLF-ULLF)

and is distributed as a [[chi square].sub.k] under the null, where k is the number of coefficients being restricted, and RLLF (ULLF) is the restricted (unrestricted) log likelihood function. Therefore:

LR =-2(-564.25+563.59) = 1.32,

which follows a chi-squared distribution with 2 degrees of freedom (i.e. the difference number of parameters in models 5 and 6). The result is reported at the bottom of table 5 (p-value = 0.52). Finally, the Bayesian Information Criterion (BIC) confirms model 5 to be the best model (1141.93).

6.2 Testing the Significance of Previously Tested Variables and the OC and 0 and C Variable

A probit regression is used to test the significance of the variables of models (7) through to (10). We use a pooled time-series, cross-sectional probit model with all three years of data, 2001 through 2003. Table 6 contains the results. Column 2 of table 6 shows the results of model (4) where voluntary disclosure is regressed on previously-tested disclosure variables. The results reported in the table suggest that voluntary disclosure is significantly related to stock return RETURN (t-ratio =-2.59) and cross-listing (t-ratio =2.03). These results are consistent with past studies.

The third column of table 6 contains the probit analysis results When our interaction variable, OC, is included as a regressor alongside commonly tested disclosure variables (see model (8)). The results are as hypothesized. We find that the OC variable is significant and has the predicted sign (t-ratio = 3.96). Firms with high levels of shares owned by top 20 shareholders and high competition appear to be more likely to voluntarily disclose segment data. The results for the other six explanatory variables are similar to those reported for model (7) in the second column. Specifically, we find a significant relation between voluntary segment disclosures and each of RETURN (t-ratio = -2.56), LOGTA (t-ratio = 2.23), and CROSS (t-ratio = 2.17), Again, these results are consistent with the past empirical literature. In summary, our results not only support the voluntary disclosure variables, RETURN and LOGTA, but also justify, both economically and statistically, the inclusion of the OC variable representing the joint effects of the internal and external environments within which the firm operates.

The fourth column of table 6 contains the probit analysis results when both the O and C variable are included alongside commonly tested disclosure variables. The results in column 4 show that both O (t-ratio = 2.46), and C (t-ratio =3.23), are significantly positive when added to the other six explanatory variables. This indicates that these variables are robust to the inclusion of variables previously identified in the literature. The fifth column and last column of table 6 contains the probit analysis results when all three variables, that is, O, C and OC are included alongside commonly tested disclosure variables. The results indicate that adding O and C fails to add anything to the model when OC and the control variables are included. This is also supported by the LR test statistic of 1.34 (p-value = 0.51). Consistent with the results from table 5, the best model is the model with the interaction variable OC and the control variables. The Bayesian Information Criterion (BIC) of 1146.5 also confirms model 8 to be the preferred model.

7. Robustness Tests

In the last section, we demonstrated that our measure of the internal/external environment of the firm, OC, has significant explanatory power in determining voluntary segment disclosures. In this section, we examine the robustness of this result with respect to: (a) the change in reporting standard; and, (b) acquisitions and disposals of other firms.

7.1 Change in Standard

The revised segment reporting standard AASB 1005 Segment Reporting changed segment reporting practices of Australian firms. (Recall that details of the change in the reporting standard were provided in section 2.) To test whether the change affected the structural relation between voluntary disclosure and its determinants, we use a dummy variable approach. Specifically, we create a dummy variable D whose value is 1 in the period after the change in standard and 0 before. The dummy variable is then inserted into model (8) in a manner that tests whether the intercept and slope coefficients changed significantly as a result of the new standard, that is,

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (11)

In order to identify whether there is a structural break associated with the change in standard, model (11) can be compared with model (8). The LR test statistic is 36.12 (p-value = 0.00) indicating that the null hypothesis that the changes in the intercept/slope coefficients are jointly equal to zero (i.e. the same before and after the change in standard) is rejected. The rejection, however, does not distinguish between a change in the relevance of the explanatory variables of the model or an increase in VD (i.e. the intercept) generally.

To identify the cause of the structural break, we conduct an additional regression and likelihood ratio test on the restriction that the change in the slope coefficients are jointly zero. To accomplish this, we estimate the following model;

VD = [[beta].sub.0] + [[beta].sub.1] + RETURN + [[beta].sub.3] + [[beta].sub.4] AUDITOR + [[beta].sub.5] LEV + [[beta].sub.6] LOGTA + [[beta].sub.7] CROSS + [[beta].sub.8] D (12)

As with the previous model, D takes a value of 1 in the period after the change in standard and 0 before. The likelihood ratio test statistic is 8.82, which follows a chi-square distribution with 7 degrees of freedom (i.e. the difference between the number of parameters in models (11) and (12)). The p-value is 0.27, which means that the null hypothesis that the slope coefficients are unchanged as a result of the standard cannot be rejected. Further, a comparison of model (12) with model (8) permits a direct test of the null hypothesis that the change in the intercept term ([[beta].sub.8]) is zero. The likelihood ratio for this test is 43.18, with a p-value of 0.00. This indicates that the structural break is due to a change in the intercept. The results are reported in the second column of table 7. Of note, the coefficient on the intercept dummy variable is positive and significant (t-ratio = 5.20). Apparently, the change in standard resulted in an overall increase in underlying/natural disclosure. The insignificance of the coefficients on the slope dummy variables in model (11) suggests that the effect of the models explanatory variables remains constant in the pre- and post-period.

The significant increase in voluntary disclosures in the post standard period is not surprising as the revised segment standard explicitly refers to voluntary disclosures and provides examples of these disclosures such as segment cash flows and any other relevant items. (13) Thus, a firm merely following the standard's suggestions would be disclosing more than the mandated requirements and would be making voluntary disclosures.

In summary, the economic model featuring OC, RETURN, FAIP, AUDITOR, LEV, LOGTA and CROSS is robust to the change in standard, and continues to apply with the same strength on each variable. However, the level of voluntary disclosure was found to increase following the standard change. In light of these findings, Model (12) featuring the intercept dummy will be used as a benchmark model against which all subsequent robustness tests will be performed.

7.2 Acquisition and Disposal Dummies and Voluntary Disclosures

We further test the robustness of the model by taking into consideration acquisition and disposal activities of the firms. Such activities can affect the number of segments reported and this could also flow on to voluntary segment disclosures. (14) We construct acquisition and disposal dummies for the pooled sample. The acquisition dummy variable is coded 1 if the firm makes a physical asset acquisition during the year, and 0 otherwise, and the disposal dummy is coded 1 if the firm disposes of physical assets during the year, and 0 otherwise. We then include the dummy variable from Model (12) and test whether the acquisition or disposal variables influence voluntary segment disclosure. The results are reported in the third column of table 7. As the table shows, the variables OC (t-ratio = 3.95), RETURN (t-ratio = -2.94), LOGTA (t-ratio = 1.96) and CROSS (t-ratio = 2.00) remain significant. The firm size variable, LOGTA, is less significant than in the previous regressions. One possible explanation for this result is multicollinearity--large firms are more likely to be involved in acquisition/disposal activity. To test the joint impact of the ACQ and DISP variables, we use a likelihood ratio test. The result, reported at the bottom of the third column, is that the restriction that the variables ACQ and DISP are jointly zero is not rejected (p-value = 0.57).

8. Conclusions

A firm's incentive to disclose has been linked empirically to a range of variables, including information asymmetry, agency costs, political costs and proprietary costs. First, this paper investigates the role of ownership and competition variables in explaining voluntary segment disclosures in Australian firms. Agency theory suggests that large shareholders have the ability to mitigate the agency problems inherent in a firm by influencing the voluntary disclosures made by the firm. Proprietary cost theory suggests that firms operating in a highly competitive environment may have greater incentive to disclose, as there is potentially less risk to their competitive position.

Second, this paper introduces a new economic variable OC that unifies both ownership and competition variables. Mayer (1997) argues that variables such as the ownership structure of the firm may be influenced by the degree of product-market competition. Further, the strategic management and industrial organisation literatures suggest that competition, combined with internal factors of the firm (for example, ownership structure) are important inputs into a firm's decision making approach. Using 1 minus the Herfindahl index to measure competition C, and the percent of shares held by the top 20 shareholders O to measure ownership, we hypothesize that the inclusion of the OC variable to the voluntary disclosure model enhances its explanatory power.

The results of the empirical tests strongly support our hypothesis. Our first test involves regressing voluntary segment disclosures VD on the ownership variable O, and we find that firms having high levels of shares owned by top 20 shareholders are more likely to disclose voluntary segment items. Our second test involves regressing VD on the competition variable C, and the results show that results show that voluntary disclosures and firms in highly competitive industries are positively related and statistically significant. We then test the contribution of our interaction variable OC alone, and, as predicted, VD and the interaction variable OC are positively related, and highly significant. Next we explicitly test the disclosure variables of past investigators and the contribution of our interaction variable OC. The results of this regression show that all variables, including OC, enter the regression significantly with their expected signs. In other words, the OC variable enhances the model's ability to explain voluntary segment disclosures. Our findings are robust to changes in the Australian segment reporting standard and capital market changes of acquisitions and disposals of physical assets.

Comments and suggestions by Mike Bradbury, Peter Clarkson, Robert Faff, Bruce Grundy, John Handley, Paul de Lange, Gordon Richardson, two anonymous referees, as well as seminar participants at Monash University, the University of Melbourne Finance Seminar Series, UNITEC seminar series, the 2003 AGSM Accounting and Finance Research Camp in Sydney and the 2003 AFAANZ conference in Brisbane are gratefully acknowledged. Petra Fleischer provided valuable assistance in data collection.

(Date of receipt of final transcript: May 11, 2006. Accepted by Garry Twite, Area Editor.)

References

Aghion, P. & Howitt, P. 1996, 'Research and development in the growth process', Journal of Economic Growth, vol. 1, pp. 49-73.

Aitken, M. Czernkowski, R. & Hooper, C. 1994, 'The information content of segment disclosures: Australian evidence', Abacus, vol. 30, no. 1, pp. 65-77.

Aitken, M.A.J., Hooper, C. & Pickering, J. 1997, 'Determinants of voluntary disclosure of segment information: A re-examination of the role of diversification strategy', Accounting and Finance, vol. 37, pp. 89-109.

Aoki, M. 1994, 'Monitoring characteristics of the main bank system: An analytical and developmental view', in M. Aoki & H. Patrick (eds), Japanese Main Bank System, Oxford University Press, Oxford UK.

Atiase, R.K. 1985, 'Predisclosure information, firm capitalization, and security price behaviour around earnings announcements', Journal of Accounting Research, vol. 23, pp. 21-36.

Australian Accounting Standards Board, 1986, Financial Reporting by Segments, Approved Accounting Standard, (AASB 1005), Canberra.

Australian Accounting Standards Board, 2000, Segment Reporting, Approved Accounting Standard, (AASB 1005), Canberra.

Bamber, L.S. 1987, 'Unexpected earnings, firm size and trading volume around quarterly earnings announcements', The Accounting Review, vol. 62, pp. 510-32.

Belkaoui, A. & Karpik, P. 1989, 'Determinants of the corporate decision to disclose social information', Accounting, Auditing and Accountability Journal, vol. 2 no. 1, pp. 36-51.

Berle, A. & Means, G. 1932, The Modern Corporation and Private Property, Harcourt, Brace & World, New York.

Berger, P.G. & Hann, R. 2002, 'Segment disclosures, proprietary costs, and the market for corporate control', Working paper, University of Pennsylvania.

Botosan, C.A. 1997, 'Disclosure level and the cost of equity capital', Accounting Review, vol. 72, pp. 323-49.

Botosan, C.A. & Stanford, M. 2005, 'Managers' motives to withhold segment disclosures and the effect of SFAS No. 131 on analysts' information environment', Accounting Review, vol. 80, no. 3, pp. 751-71.

Bradbury, M. 1991, 'Characteristics of firms and voluntary interim earnings disclosures: New Zealand evidence', Pacific Accounting Review, vol. 3, pp. 36-62.

Bradbury, M. 1992, 'Voluntary disclosures of segment information', Accounting and Finance, vol. 32, pp. 15-26.

Chow, C.W. & Wong-Boren, A. 1987, 'Voluntary disclosure by Mexican companies', The Accounting Review, vol. 62, pp. 533-41.

Deegan, C. & Gordon, B. 1996, 'A study of the environmental disclosure practices of Australian Corporations', Accounting and Business Research, vol. 26, pp. 187-99.

Demsetz, H. 1983, 'The structure of ownership and the theory of the firm', Journal of Law and Economics, vol. 26, pp. 375-90.

Diamond, D.W. & Verrecchia, R.E. 1991, 'Disclosure, liquidity, and the cost of capital', Journal of Finance, vol. 46, pp. 1325-60.

Dye, R.A. 2001, 'An evaluation of 'essays on disclosure' and the disclosure literature in accounting', Journal of Accounting and Economics, vol. 32, pp. 181-235.

Foster, G. 1986, Financial Statement Analysis. Prentice-Hall, Englewood Cliffs, N.J.

Harris, M.S. 1998, 'The association between competition and managers' business segment reporting decisions', Journal of Accounting Research, vol. 36, pp. 111-28.

Hayes, R.M. & Lundholm, R.J. 1996, 'Segment reporting to the capital market in the presence of a competitor', Journal of Accounting Research, vol. 34, pp. 261-79.

Healy, P.M., Hutton, A. & Palepu, K.G. 1999, 'Stock performance and intermediation changes surrounding sustained increases in disclosure', Contemporary Accounting Research, vol. 16, pp. 485-520.

Healy, P.M. & Palepu, K.G. 2001, 'Information asymmetry, corporate disclosure, and the capital markets: A review of the empirical disclosure literature', Journal of Accounting and Economics, vol. 31, pp. 405-40.

Holthausen, R. & Leftwich, R. 1983, 'The economic consequences of accounting choice: Implications of costly contracting and monitoring, Journal of Accounting and Economics, vol. 5, pp. 77-117.

Januszewski, S.I., Koke, J. & Winter, J.K. 2002, "Product market competition, corporate governance and firm performance: An empirical analysis for Germany', Research in Economics, vol. 56, pp. 299-332.

Jensen, M.C. & Meckling, W. 1976, 'Theory of the firm: Managerial behavior, agency costs and ownership structure', Journal of Financial Economics, vol. 3, pp. 305-60.

Jiambalvo, J., Rajgopal, S. & Venkatachalam, M. 2002, 'Institutional ownership and the extent to which stock prices reflect future earnings', Contemporary Accounting Research, vol. 19, pp. 117-45.

Kelly, G.J. 1994, 'Unregulated segment reporting: Australian evidence', British Accounting Review, vol. 26, pp. 217-34.

King R., Pownall, G. & Waymire, G. 1990, 'Expectations adjustment via timely management forecasts: Review, synthesis, and suggestions for future research, Journal of Accounting Literature, vol. 9, pp. 133-44.

Kochanek, R.F. 1974, "Segmental financial disclosure by diversified firms and security prices', Accounting Review, vol. 49, pp. 245-58.

Lang, M. & Lundholm, R. 1993, 'Cross-sectional determinants of analyst ratings of corporate disclosures', Journal of Accounting Research, vol. 31, pp. 246-71.

Lev, B. & Penman, S. 1990, 'Voluntary forecast disclosure, nondisclosure, and stock prices', Journal of Accounting and Research, vol. 28, pp. 49-76.

Leuz, C. 1999, 'Proprietary versus non-proprietary disclosures: Voluntary cash flow statements and business segment reports in Germany', working paper, University of Pennsylvania.

Mayer, C. 1997, 'Corporate governance, competition and performance', Journal of Law and Society, vol. 24, pp. 152-76.

Mitchell, J.D., Chia, C.W.L. & Loh, A.S. 1995, 'Voluntary disclosure of segment information: Further Australian evidence', Accounting and Finance, vol. 35, pp. 1-15.

McKinnon, J.L. & Dalimunthe, L. 1993, 'Voluntary disclosure of segment information by Australian diversified companies', Accounting and Finance, vol. 33, pp. 33-50.

Nagar, V., Nanda, D. & Wysocki. P. 2003, "Discretionary disclosure and stock-based incentives', Journal of Accounting and Economics, vol. 34, pp. 283-309.

Nickell, S.J. 1996, 'Competition and corporate performance', The Journal of Political Economy, vol. 104, pp. 724-46.

Nickell, S.J., Nicolitsas, D. & Dryden, N. 1997, 'What makes firms perform well?' European Economic Review, vol. 41, pp. 783-96.

Porter, M.E. 1981, 'The contributions of industrial organization to strategic management', The Academy of Management Review, vol. 6, pp. 609-20.

Prencipe, A. 2004, 'Proprietary costs and determinants of voluntary segment disclosure: Evidence from Italian listed companies', European Accounting Review, vol. 13, pp. 319-40.

Ronen, J. & Livnat, J. 1981, 'Incentives for segment reporting', Journal of Accounting Research, vol. 19, pp. 459-81.

Saloner, G. 1991, 'Modeling, game theory and strategic management', Strategic Management Journal, vol. 12, Special Issue: Fundamental Research Issues in Strategy and Economics, pp. 119-36.

Schmalensee, R. & Willig, R.D. 1989, Handbook of Industrial Organisation, North Holland, New York.

Street, D. & Gray, S. 2001, 'What lies behind non-compliance?' Accounting and Business, pp. 36-7.

Tirole, J. 1990, The Theory of Industrial Organization, MIT Press, Cambridge, MA.

Verrecchia, R.E. 1983, 'Discretionary disclosure', Journal of Accounting and Economics, vol. 5, pp. 179-94.

Verrecchia, R.E. 2001, 'Essays on disclosure', Journal of Accounting and Economics, vol. 32, pp. 97-180.

Wagenhofer, A. 1990, 'Voluntary disclosure with a strategic opponent', Journal of Accounting and Economics, vol. 12, pp. 341-63.

(1.) See Botosan (1997), Deegan and Gordon (1996) and Kelly (1994).

(2.) See Kochanek (1974) and Aitken, Czernkowski and Hooper (1994).

(3.) See, for example, Kochanek (1974) and Aitken, Czemkowski and Hooper (1994).

(4.) As part of the Australian adoption of IFRS, AASB 1005 Segment Reporting has now been reissued as AASB 114 Segment Reporting.

(5.) The 'management approach' has been adopted from the U.S. standard SFAS 131 and the international standard IAS 14R.

(6.) A geographic segment is classified as primary if the entity's risks and returns are affected predominantly by the fact that it operates in different countries or other geographical areas. A business segment would be primary if the entity's risks and returns are affected predominantly by the differences in the products and services it provides.

(7.) See Belkaoui and Karpik (1989).

(8.) See Healy and Palepu (200I).

(9.) See also, Hayes and Lundholm (1996) and Ronen and Livnat (1981).

(10.) In the early 1990s, a special issue of the Strategic Management Journal (Vol, 12, 1991) examines the relation between strategic management and economics and indicates areas for future research using the linkage between the two disciplines.

(11.) Other proxies for competition have been used in segment studies including the four firm concentration ratio and the speed of abnormal profit adjustment. See, for example, Leuz (1999) and Harris (1998). We choose the Herfindahl index as it is widely used in research and practice including the U.S. Department of Justice in its antitrust investigations.

(12.) The percentage of voluntary disclosures has obviously increased over the 2001-2003 period. This increase is possibly due to the revised standard explicitly stating examples of voluntary disclosures that a firm may choose to disclose.

(13.) See Sections 6.1.3 and 6.1.4 of the standard.

(14.) We thank Gordon Richardson for suggesting the acquisition and disposal dummy as a robustness check.

by

Jacqueline L. Birt ([dagger])

Chris M. Bilson ([section])

Tom Smith ([section])

Robert E. Whaley ([double dagger])

([dagger]) School of Accounting and Business Information Systems, Australian National University. College of Business and Economics, Australian National University, CANBERRA ACT 0200. Email: [e-mail]

([section]) School of Finance and Applied Statistics, College of Business and Economics, Australian National University, CANBERRA ACT 0200.

([double dagger]) Owen School of Management, Vanderbilt University, 401 21st Avenue South, Nashville TN 37203.

Table 1
Summary of Estimation Results in Studies of Corporate Disclosure

Sign and significance (at the five percent probability level)
of variables are: ++ significantly positive, + positive but
insignificant,--negative but insignificant, and -- significantly
negative. Market legend: N = NYSE; A = ASX; NZ = NEW ZEALAND;
G = GERMANY

Market
No. of Firms

Agency Variables       Variable Definition

Firm size              Log of total assets

Leverage               Book value of debt to
                       sum of book value of
                       debt and market value
                       of equity

Assets in place        Book value of fixed
                       assets to total assets

Ownership              Percent of ordinary
diffusion              shares not owned by
                       top 20 shareholders

Profitability          Net profit to total
                       assets

Minority interest      1% of subs not held
                       by Top 20 shareholders

Free float             Percent of voting
                       shares held for free
                       trading

Number of              Natural logarithm of
shareholders           number of shareholders

Number of
subsidiaries

Market-to-book         Ratio of market/book
                       equity

Proprietary Cost       Variable Definition
Variables

Competition--4 firm    4 firm ratio
concentration ratio

Competition--speed     Speed of profit
of profit adjustment   adjustment

Abnormal profit        Industry adjusted ROA
                       and ROE

Herfindahl index       Industry concentration

Industry               Dummy variable of
diversification        high/low diversity

Segment                Number of SIC codes
Diversification        across segments to
                       number of segments

Heterogeneity          Heterogeneity in
                       earnings persistence

Other Variables        Variable Definition

Overseas association   Overseas listing

Earnings volatility    Five-year coefficient of
                       variation

Trading volume         Share turnover

Number of segments     Number of segments

Number of industries   Number of industries

Scale of               Industry sales/Firm
operations/Firm size   sales

Foreign sales          % of sales outside
                       domicile

Big '5' or '6'         Big '5' or '6' auditor
auditor

Industry membership    Mining and oil
                       classification

                        Bradbury     Aitken,        Leuz
                         (1992)      Hooper,       (1999)
                                    Pickering
                                      (1994)

Market                     NZ           A            G
No. of Firms               29           65          109

Agency Variables

Firm size                  ++           ++           ++

Leverage                   ++           ++           -

Assets in place            +            +            ++

Ownership                               +
diffusion

Profitability                                        --

Minority interest                       +

Free float                                           ++

Number of                               ++
shareholders

Number of                               ++
subsidiaries

Market-to-book

Proprietary Cost
Variables

Competition--4 firm
concentration ratio

Competition--speed
of profit adjustment

Abnormal profit

Herfindahl index

Industry                            +            +
diversification

Segment
Diversification

Heterogeneity

Other Variables

Overseas association   +            +            +

Earnings volatility    -            -

Trading volume                                   +

Number of segments

Number of industries

Scale of
operations/Firm size

Foreign sales                                    +

Big '5' or '6'                                   +
auditor

Industry membership                 ++

                       McKinnon &     Harris
                       Dalimunthe     (1998)
                         (1993)

Market                     A            N
No. of Firms               65          929

Agency Variables

Firm size                  ++

Leverage                   +

Assets in place

Ownership                  ++
diffusion

Profitability

Minority interest          ++

Free float

Number of
shareholders

Number of
subsidiaries

Market-to-book

Proprietary Cost
Variables

Competition--4 firm                 --
concentration ratio

Competition--speed                  --
of profit adjustment

Abnormal profit

Herfindahl index

Industry               +
diversification

Segment
Diversification

Heterogeneity                       --

Other Variables

Overseas association

Earnings volatility

Trading volume

Number of segments

Number of industries                --

Scale of                            ++
operations/Firm size

Foreign sales

Big '5' or '6'
auditor

Industry membership    ++

                       Mitchell,     Berger &
                         Chia &        Hann
                          Loh         (2002)
                         (1995)

Market                     A            N
No. of Firms              129          1207

Agency Variables

Firm size                  ++           --

Leverage                   ++

Assets in place            +

Ownership                  ++
diffusion

Profitability

Minority interest          ++

Free float

Number of
shareholders

Number of
subsidiaries

Market-to-book                          +

Proprietary Cost
Variables

Competition--4 firm                 +
concentration ratio

Competition--speed                  ++
of profit adjustment

Abnormal profit                     +

Herfindahl index                    +

Industry
diversification

Segment                             --
Diversification

Heterogeneity                       +

Other Variables

Overseas association   ++

Earnings volatility    -

Trading volume

Number of segments                  ++

Number of industries

Scale of
operations/Firm size

Foreign sales

Big '5' or '6'
auditor

Industry membership    ++

Table 2
Summary of Sample Firms by Year and Industry Category

The voluntary disclosure variable, VD, is the number of firms
within each industry that disclosed information in addition to
that required by the standard, and TOTAL is the number of firms
in each industry. COUNT shows the totals across all industries.
O is the median of the ownership variable (expressed as a
percentage) in each industry, and C is the concentration level
within each industry (expressed as a percentage) as measured by
1 minus the Herfindahl index for each industry using the top
500 firms on the Australian Stock Exchange.

CRIF Class Name   Energy   Chemicals   Construction   Diversified
                                        Materials      Metals and
                                                         Mining

CRIF Class No.        1          2            3              4

2001 Sample
  VD                  3          4            2              7
  TOTAL              11          6            5             16
  O (%)              40         41           64             59
  C (%)              51         68           54             59
2002 Sample
  VD                  6          4            2              4
  TOTAL               9          6            4              9
  O (%)              48         56           60             59
  C (%)              50         69           57             54
2003 Sample
  VD                  8          4            3              8
  TOTAL              10          5            6             13
  O (%)              53         50           67             62
  C (%)              63         56           63             57
Full Sample
  VD                 17         12            7             19
  TOTAL              30         17           15             38
  O (%)              48         50           62             60
  C (%)              51         68           57             57

CRIF Class Name   Gold    Precious    Steel and    Paper and
                         Metals and   Aluminium      Forest
                          Minerals                Products and
                                                   Packaging
CRIF Class No.      5         6            7            8

2001 Sample
  VD                3         1            1            1
  TOTAL            13         2            3            8
  O (%)            75        81           46           66
  C (%)            81        42           58           55
2002 Sample
  VD                3         2            2            6
  TOTAL            14         2            4            9
  O (%)            70        84           56           60
  C (%)            84        49           57           52
2003 Sample
  VD                6         2            6            6
  TOTAL            14         2            6            6
  O (%)            69        60           62           63
  C (%)            88        63           67           49
Full Sample
  VD               12         5            9           13
  TOTAL            41         6           13           23
  O (%)            71        68           56           64
  C (%)            84        49           58           52

CRIF Class Name   Building   Construction   Machinery   Conglomerates
                  Products       and                      and Other
                             Engineering                Capital Goods

CRIF Class No.         9           10            11            12

2001 Sample
  VD                   1            2             1             2
  TOTAL                2           10             5             8
  O (%)               69           68            69            66
  C (%)               64           71            34            56
2002 Sample
  VD                   3            5             1             2
  TOTAL                3            8             4             6
  O (%)               88           68            66            61
  C (%)               66           64            40            38
2003 Sample
  VD                   1            6             3             3
  TOTAL                1            7             5             6
  O (%)               52           66            64            51
  C (%)               57           68            39            42
Full Sample
  VD                   5           13             5             7
  TOTAL                6           25            14            20
  O (%)               70           68            67            62
  C (%)               65           68            39            42

CRIF Class Name    Commercial    Transportation   Automobile
                    Services                         and
                  and Supplies                    Components

CRIF Class No.          13              14             15

2001 Sample
  VD                     5               0              3
  TOTAL                 14               2              5
  O (%)                 62              78             57
  C (%)                 69              40             62
2002 Sample
  VD                     9               3              2
  TOTAL                 20               4              4
  O (%)                 67              69             62
  C (%)                 71              48             69
2003 Sample
  VD                    12               1              4
  TOTAL                 15               6              5
  O (%)                 67              63             58
  C (%)                 76              57             78
Full Sample
  VD                    26               4              9
  TOTAL                 49              12             14
  O (%)                 66              66             57
  C (%)                 71              57             69

CRIF Class Name     Consumer       Hotels     Media   Retailing
                    Durables    Restaurants
                  and Apparel   and Leisure

CRIF Class No.          16          17         18        19

2001 Sample
  VD                     0           4          4         7
  TOTAL                  3           7         14        10
  O (%)                 80          86         76        82
  C (%)                 76          78         48        88
2002 Sample
  VD                     2           3          4         8
  TOTAL                  6           6         15        12
  O (%)                 70          53         80        78
  C (%)                 73          77         45        90
2003 Sample
  VD                     4           6         11         9
  TOTAL                  5          14         16        11
  O (%)                 74          63         78        72
  C (%)                 82          83         51        90
Full Sample
  VD                     6          13         19        24
  TOTAL                 14          27         45        33
  O (%)                 74          66         79        78
  C (%)                 76          83         48        90

CRIF Class Name    Food and   Beverages     Food and     Health Care
                     Drug                    Other        Equipment
                  Retailing               Products and   and Supplies
                                            Tobacco
CRIF Class No.       20            21           22             23

2001 Sample
  VD                  1             4            8              2
  TOTAL               4             8           14              4
  O (%)              55            74           60             56
  C (%)              64            78           79             36
2002 Sample
  VD                  1             6           10              6
  TOTAL               3            10           14             11
  O (%)              52            63           70             56
  C (%)              64            64           77             49
2003 Sample
  VD                  3             5            9              7
  TOTAL               4             7           11             10
  O (%)              58            61           44             63
  C (%)              66            70           87             63
Full Sample
  VD                  5            15           27             15
  TOTAL              11            25           39             25
  O (%)              52            66           59             56
  C (%)              64            70           79             49

CRIF Class Name   Health Care   Pharmaceuticals   Banks   Diversified
                   Providers          and                  Financials
                  and Services   Biotechnology

CRIF Class No.          24             25           26         27

2001 Sample
  VD                     3              4            5          5
  TOTAL                  4              8            6         13
  O (%)                 59             45           41         71
  C (%)                 49             66           77         84
2002 Sample
  VD                     3              7            7          2
  TOTAL                  6             12            8         16
  O (%)                 61             56           45         67
  C (%)                 50             68           78         75
2003 Sample
  VD                     2              8            6          7
  TOTAL                  4             12            6         17
  O (%)                 66             46           42         44
  C (%)                 53             70           78         72
Full Sample
  VD                     8             19           18         14
  TOTAL                 14             32           20         46
  O (%)                 61             51           41         57
  C (%)                 50             68           78         75

CRIF Class Name   Insurance     Real Estate   Real Estate     Internet
                                 Investment    Management     Software
                                  Trusts         and       and Services
                                              Development
CRIF Class No.        28            29            30            31

2001 Sample
  VD                   1             2             3             1
  TOTAL                4             9             6             2
  O (%)               57            64            50            48
  C (%)               53            91            35            58
2002 Sample
  VD                   2             4             2             3
  TOTAL                4             7             9             3
  O (%)               73            62            80            57
  C (%)               44            89            30            55
2003 Sample
  VD                   4             8             4             1
  TOTAL                5            11             8             3
  O (%)               51            69            83            52
  C (%)               74            89            49            69
Full Sample
  VD                   7            14             9             5
  TOTAL               13            27            23             8
  O (%)               69            67            80            51
  C (%)               53            89            35            58

CRIF Class Name   It Consulting   Software   Technology
                   and Services               Hardware
                                                and
                                             Equipment
CRIF Class No.            32           33          34

2001 Sample
  VD                       1            6           5
  TOTAL                    2           13           7
  O (%)                   66           69          53
  C (%)                   65           82          82
2002 Sample
  VD                       1            7           5
  TOTAL                    5            9           7
  O (%)                   66           74          57
  C (%)                   68           75          83
2003 Sample
  VD                       2            8           8
  TOTAL                    5           10           8
  O (%)                   64           70          61
  C (%)                   67           83          85
Full Sample
  VD                       4           21          18
  TOTAL                   12           32          22
  O (%)                   65           70          58
  C (%)                   67           82          83

CRIF Class Name   Telecommunications   Utilities   Count

CRIF Class No.              35               36

2001 Sample
  VD                         3                2      107
  TOTAL                      9                6      263
  O (%)                     75               69
  C (%)                     55               42
2002 Sample
  VD                         3                2      142
  TOTAL                      3                4      276
  O (%)                     82               59
  C (%)                     61               37
2003 Sample
  VD                         5                3      193
  TOTAL                      6                6      286
  O (%)                     83               59
  C (%)                     58               41
Full Sample
  VD                        11                7      442
  TOTAL                     18               16      825
  O (%)                     80               67
  C (%)                     57               41

Table 3
Summary Statistics of the Regression Variables

The sample consists of the Top 500 Australian companies for
the years 2001 through 2003. The notation is as follows: O
is the percentage of total shares outstanding not held
directly or indirectly by the directors of the company, C
is 1 minus the level of Herfindahl index based on firm
revenue within the industry, OC is the product of O and C,
RETURN is the log stock return over the year, FAIP is the
fixed assets in place, AUDITOR is an indicator variable set
equal to 1 if the firm is audited by a big 5 (2001) or big
4 auditor (2002 and 2003) and O otherwise, LEV is the book
value of debt divided by total assets, LOGTA is the natural
logarithm of total assets and CROSS is an indicator variable
which takes a value of 1 if a firm is cross-listed and zero
otherwise.

Variable   No. of obs.   Mean         Std Dev      Min

O          825           0.6259       0.2023       0.0200
C          825           0.6653       0.1526       0.2976
OC         825           0.4166       0.1710       0.0110
RETURN     825          -0.0419       0.6015      -2.0424
FAIP       825           0.5986       0.2420       0
AUDITOR    825           0.8230       0.3819       0
LEV        825           0.4485       0.2247       0.0068
LOGTA      825          12.7371       2.0823       8.2779
CROSS      825           0.1321       0.3388       0

Variable    25%          Median       75%          Max

O           0.4900       0.6470       0.7800       1.0000
C           0.5487       0.6833       0.7811       0.9115
OC          0.2887       0.4035       0.5311       0.9022
RETURN     -0.2541       0.0366       0.2685       2.3786
FAIP        0.4257       0.6212       0.8005       1
AUDITOR     1            1            1            1
LEV         0.2898       0.4599       0.5949       0.9837
LOGTA      11.2248      12.4284      14.0398      19.8006
CROSS       0            0            0            1

Table 4
Pair-Wise Correlation Coefficients Among Regression Variables

The sample consists of the Top 500 Australian companies for the
years 2001 through 2003. The notation is as follows: O is the
percentage of total shares outstanding not held directly or
indirectly by the directors of the company, C is 1 minus the
level of Herfindahl index based on firm revenue within the
industry, OC is the product of O and C, RETURN is the log stock
return over the year, FAIP is the fixed assets in place, AUDITOR
is an indicator variable set equal to 1 if the firm is audited
by a big 5 (2001) or big 4 auditor (2002 and 2003) and 0
otherwise, LEV is the book value of debt divided by total assets,
LOGTA is the natural logarithm of total assets and CROSS is an
indicator variable which takes a value of 1 if a firm is
cross-listed and zero otherwise.

Variable      O        C         OC       RETURN

C             0.006
OC            0.793    0.587
RETURN       -0.004    0.031    -0.012
FAIP          0.043   -0.036     0.002    0.093
AUDITOR      -0.013   -0.043    -0.052   -0.006
LEV          -0.048   -0.041    -0.065    0.033
LOGTA        -0.051   -0.101    -0.115    0.111
CROSS        -0.087   -0.016    -0.079    0.007

Variable      FAIP     AUDITOR   LEV      LOGTA

C
OC
RETURN
FAIP
AUDITOR       0.044
LEV          -0.031    0.210
LOGTA         0.302    0.235     0.529
CROSS         0.087    0.078     0.099    0.344

Table 5
Results of the OC Regressions

The sample consists of the Top 500 Australian companies for the
years 2001 through 2003. The voluntary disclosure variable, VD, is the
dependent variable in all regressions. The table reports the results
of regressions of VD on various models involving the ownership variable
(O), the competition variable (C), and the interaction variable (OC).
LLF is the log likelihood function. The likelihood ratio test statistic
(LR) is calculated as LR = -2(RLLF-ULLF) and is distributed as a [[chi]
.sub.k] under the null, where [[chi].sub.k] is the number of
coefficients being restricted, and RLLF (ULLF) is the restricted
(unrestricted) log likelihood function. BIC is the Bayesian Information
Criterion, which is calculated as -2LLF + k ln n where k is the number
of parameters estimated and n is the sample size. T-statistics are
reported in parentheses.

Dependent      No. of                     Coefficient Estimate
Variable        obs      Constant          (and t-ratio) for:

                                       OC         O         C

VD              825       -0.178                 0.428
                         (-1.25)                -1.97
VD              825       -0.458                           0.825
                         (-2.33)                          (2.86)
VD              825       -0.726                 0.428     0.825
                         (-3.03)                (1.97)    (2.86)
VD              825       -0.265      0.853
                         (-2.28)     (3.30)
VD              825       -0.463      0.604      0.025     0.431
                         (-0.69)     (0.42)     (0.03)    (0.44)

Dependent
Variable        LLF       LR     p-value      BIC

VD             -567.78                       1148.99
VD             -565.62                       1144.67
VD             -563.67                       1147.49
VD             -564.25                       1141.93
VD             -563.59    1.32      0.52     1154.04

Table 6
Results of the OC Regressions Using Control Variables

The sample consists of the Top 500 Australian companies for the
years 2001 through 2003. The voluntary disclosure variable, VD,
is the dependent variable in all regressions. The independent
variables are as follows: OC, RETURN, FAIP, AUDITOR, LEV, LOGTA
and CROSS. The likelihood ratio test statistic (LR) is calculated
as LR = -2(RLLF-ULLF) and is distributed as a [[chi].sub.k] under
the null, where k is the number of coefficients being restricted,
and RLLF (ULLF) is the restricted (unrestricted) log likelihood
function. For each regression, the coefficients being restricted
are presented in bold. The RLLF for columns 3 and 4 is the ULLF
in column 2; for column 5 the RLLF is the ULLF in column 3. BIC
is the Bayesian Information Criterion, which is calculated as
-2LLF + k ln n, where k is the number of parameters estimated
and n is the sample size. T-statistics are reported in parentheses.

Independent
Variables        VD         VD         VD         VD

No. of obs.    825        825         825        825
Constant        -0.701     -1.263      (1.81)     -1.328
               (-2.31)    (-3.74)     (-4.38)    (-1.81)
0C                         1.047                   1.163
                           (3.96)                 (0.77)
RETURN          -0.192     -0.193      (0.19)     -0.194
               (-2.56)    (-2.56)     (-2.52)    (-2.57)
FAIP            -0.278     -0.313      (0.31)     -0.305
               (-1.40)    (-1.57)     (-1.56)    (-1.53)
AUDITOR          0.042      0.059       0.06       0.062
                (0.35)     (0.49)      (0.45)     (0.51)
LEV              0.383      0.386       0.39       0.380
                (1.58)     (1.59)      (1.59)     (1.57)
LOGTA            0.056      0.066       0.07       0.068
                (1.91)     (2.23)      (2.25)     (2.30)
CROSS            0.289      0.311       0.31       0.302
                (2.03)     (2.17)      (2.13)     (2.10)
O                                       0.54       1.163
                                       (2.46)     (0.79)
C                                       0.95      -0.232
                                       (3.23)    (-0.22)
ULLF          -554.33    -546.39     -546.0     -545.72
RLLF            n/a      -554.33     -554.3     -546.39
LR              n/a        15.88       16.60       1.34
p-value         n/a         0.00        0.00       0.51
BIC           1155.67    1146.50     1152.50    1158.59

Table 7
Results of Robustness Tests

The sample consists of the Top 500 Australian companies for the
years 2001 through 2003. The voluntary disclosure variable, VD,
is the dependent variable in all regressions. The independent
variables are as follows: OC, RETURN, FAIP, AUDITOR, LEV, LOGTA
and CROSS. There are also four dummy variables used. The structural
break variable, D, takes a value of 1 in the period after the change
in standard (i.e. 2002 and 2003) and 0 otherwise. The acquisition
dummy variable ACQ takes a value of 1 if a company makes a physical
asset acquisition during the year while the disposal dummy variable
DISP takes a value of 1 for those firms that dispose of physical
assets during the year. LLF is the log likelihood function. The
likelihood ratio test statistic (LR) is calculated as LR =
-2(RLLF-ULLF) and is distributed as a [[chi].sub.k] under the null,
where k is the number of coefficients being restricted, and RLLF
(ULLF) is the restricted (unrestricted) log likelihood function. For
each regression, the coefficients being restricted are presented in
bold. The RLLF for column 3 is the ULLF in column 2. BIC is the
Bayesian Information Criterion, which is calculated as -2LLF + k ln
n, where k is the number of parameters estimated and n is the sample
size. T-statistics are reported in parentheses.

Independent Variables       VD           VD

No. of obs.               825          825
Constant                   -1.624       -1.568
                          (-4.64)      (-4.43)
OC                          1.059        1.057
                           (3.96)       (3.95)
RETURN                     -0.219       -0.225
                          (-2.87)      (-2.94)
FAIP                       -0.303       -0.296
                          (-1.51)      (-1.47)
AUDITOR                     0.018        0.017
                           (0.14)       (0.14)
LEV                         0.438        0.440
                           (1.79)       (1.80)
LOGTA                       0.067        0.060
                           (2.26)       (1.96)
CROSS                       0.297        0.291
                           (2.04)       (2.00)
D                           0.503        0.501
                           (5.20)       (5.17)
ACQ                                      0.079
                                        (0.75)
DISP                                     0.083
                                        (0.63)
ULLF                     -532.74      -532.18
RLLF                     -554.33      -532.74
LR                         43.18         1.12
p-value                     0.00         0.57
BIC                      1125.92      1138.23