1 Introduction

In this study we investigate the role played by the corporate governance framework of Australian companies in the decision to voluntarily disclose forward-looking information in their published financial reports. Forward-looking information in the form of management forecasts has been used as a proxy for voluntary disclosure in a number of prior studies (Healy and Palepu 2001). Furthermore, the disclosure of such information may be viewed as one dimension of financial reporting quality since a financial report containing forward-looking information is more likely to be perceived as being of higher quality (Ajinkya et al. 2005; Karamanou and Vafeas 2005; Clarkson et al. 1994). In the present study we examine the disclosure of all types of forward-looking information, including but not limited to management forecasts, in order to provide a broader representation of the type of information disclosed by firms regarding their future prospects. We consider whether forward-looking information is provided within all narrative sections of the annual report (e.g., chairman’s letter, directors’ report, review of operations and discussion and analysis), prior to examining how such disclosures relate to the firm’s standard of corporate governance.

There are numerous techniques that a firm may use to distribute corporate information to external stakeholders, with earnings forecasts frequently being made under the continuous disclosure regime of the Australian Securities ExchangeFootnote 1 (ASX). However, the corporate annual report has traditionally been the principal medium for conveying detailed financial and non-financial information on the firm (Botosan 1997). Firms employ the annual report as an important avenue of disclosure, given its effectiveness as a marketing tool and a means of conveying a certain corporate image or message (Preston et al. 1996). Neu et al. (1998, p. 269) suggest that annual reports “provide organisations with an effective method of managing external impressions”, in part because the annual report “possesses a degree of credibility not associated with other forms of advertising”. Hence, while we acknowledge that the annual report may not be the first avenue of disclosure of forward-looking information, we argue that firms that make high quality disclosures will ensure that their report includes such information.

Prior research indicates that firms with higher standards of corporate governance are more disposed to voluntary disclosure of future information (Hossain et al. 2005; Karamanou and Vafeas 2005; Ajinkya et al. 2005). Hence in this study, we expect to identify a positive relationship between certain corporate governance categories (and the overall standard of corporate governance) and the disclosure of forward-looking information.

This study is motivated by the aim of providing a more detailed examination of the relationship between corporate governance and the voluntary disclosure of forward-looking information in Australian corporate annual reports. Australia provides an interesting setting in which to study such relationships because corporate governance is less regulated than in other regimes such as the United States (Davidson et al. 2005). Indeed, at the time of the present study, there were no formal requirements or recommendations relating to board structure and the use of sub-committees such as the audit committee. Further, while Australian listed companies are required to provide a review of operations in the Directors’ Report, the content of the report at the time of the study was unregulated and the disclosure of forward-looking information was not stipulated (Seah and Tarca 2007).

We provide an in-depth representation of corporate governance within the firm by considering a range of fifteen governance mechanisms. In addition, we investigate the broader category of forward-looking information, rather than the narrower category of earnings forecasts (Clarkson et al. 1994). Hence our study contributes to the small but growing body of literature that examines the effect of corporate governance on the quality of financial disclosures (Karamanou and Vafeas 2005).Footnote 2 Our findings should be of particular interest to corporate regulators and policy makers who expect a link between stronger governance and greater corporate transparency. The study should also interest regulators who advocate the provision of forward-looking information.

Our study focuses on the years 2000 and 2002. The results reveal that the forward-looking information disclosed by sample firms tends to be expressed in terms of expecting an upward movement in such measures as profit, earnings or income. In the year 2000, disclosing firms typically enjoy a higher standard of corporate governance than non-disclosing firms. In that year, the corporate governance category, audit quality, which consists of the presence of an audit committee, its independence, the frequency with which it meets, the adoption of a big 6 auditor and the auditor’s independence, is positively associated with the decision to disclose forward-looking information in financial reports. The corporate governance category, board committees, consisting of the appointment and independence of a compensation committee and the creation of a nomination committee, and the overall efficacy of the corporate governance system are also positively associated with the disclosure of forward-looking information. In 2002, however, none of the governance categories is significantly associated with the firm’s decision to publish forward-looking information in financial reports. This is in spite of the finding that the strength of corporate governance in the sample firms increased between 2000 and 2002. Hence, this result suggests that an increased application of corporate governance mechanisms does not necessarily lead to a higher incidence of disclosure, thereby calling into question the effectiveness of such mechanisms in promoting greater transparency.

The paper is structured in the following way. The next section provides the institutional framework for forward-looking information. The third section discusses prior research and develops the hypotheses. Data and method are described in the fourth section while the results are presented in fifth section. The final section provides the concluding comments.

2 Institutional background

There are no disclosure rules that relate expressly to earnings forecasts. The decision of whether and how to release forward-looking statements, such as profit projections within the annual report, is ultimately determined by managerial discretion. There are, however, a number of financial reporting guidelines that may have implications for the inclusion of earnings forecasts in the annual report. Many countries have mandated the inclusion of a Management Discussion and Analysis (MD&A) or Discussion and Analysis (D&A) within the annual report. Such nations have not compelled the disclosure of forward-looking information within statutory D&A, however it remains a likely vehicle for the voluntary disclosure of such data.

For example, 1980 saw the introduction of statutory MD&A requirements in the United States (US), including discussions of liquidity, capital resources and the results of operations. The incorporation of forward-looking information such as earnings forecasts is promoted but not mandated. In 1989, Canada introduced legislation based on the US model requiring the inclusion of an MD&A within corporate annual reports. Although the provision of forward-looking information is not obligatory, it is strongly espoused by the Ontario Securities Commission. Similarly, in the United Kingdom (UK), the Accounting Standards Board has formally withdrawn Reporting Statement (RS) 1 Operating and Financial Review and replaced it with a statement of best practice, the Reporting Statement “The Operating and Financial Review”.

In Australia, the Corporations Law,Footnote 3 Australian Accounting Standards and the ASX Listing Rules and Guidance Notes all influence the nature of disclosures made within annual reports. D&A requirements were first introduced in 1998 with the release of AASB 1039—Concise Financial Reports.Footnote 4 Given that many companies issue concise reports, AASB 1039 has made significant inroads into introducing D&A requirements into Australian accounting regulation. However, the standard does not specifically espouse the disclosure of forward-looking information within such discussion.

In September 1999, the ASX introduced the requirement for a review of operations to be included in directors’ reports effective from 2000. ASX Guidance Note “Review of Operations and Activities: Listing Rule 4.10” provides guidance on the preparation of the company’s review of operations and financial conditions, including identifying items to be included in such a review. Specifically, as it pertains to forward-looking information, a review of operations should include among other things, “operating results of the company … including a comparison to prior periods and any projections published by the company …” and “investments for future performance”.

Our study’s focus on the years 2000 and 2002 is particularly significant in light of the changes to Australian regulation around that the time. A starting point of 2000 meant that listed companies were required to include a review of operations in their directors’ reports for the first time in that year. However, most had been producing similar information in the D&A in their Concise Annual Reports since 1998, giving time for “good practice” to be established. Furthermore, our comparison year of 2002 preceded a number of important changes to regulation affecting both corporate disclosure and corporate governance.Footnote 5 Hence, the comparatively stable time period chosen for our study enables us to minimize the effects of regulatory change on both disclosure and governance requirements.

3 Prior research and hypothesis development

The UK Cadbury Committee (Committee on the Financial Aspects of Corporate Governance, 1992) is credited with providing the momentum for the emergence of studies in the area of corporate governance (Bujaki and McConomy 2002). Specifically, the implementation of certain corporate governance mechanisms is reported to improve financial reporting quality in general, as well as disclosure transparency, and the standard and extent of voluntary disclosure. For example, studies find a positive relationship between voluntary disclosure and the number of independent directors on the board (Leftwich et al. 1981; Forker 1992), the presence of an audit committee (Forker 1992), the dispersion of shareholdings (Hossain et al. 1994) and a higher level of institutional ownership (Bushee and Noe 2000). Further, Narayanan (2000) demonstrates analytically that voluntary disclosure increases as managerial compensation becomes more sensitive to stock prices. In a recent Australian study, Beekes and Brown (2006) report that firms with a higher overall standard of corporate governance tend to make more informative disclosures. A firm’s standard of corporate governance is captured by a governance rating which takes into account attributes such as the independence of the board, its chairman and major committees.

Several studies have found a relation between prospective information and various corporate governance factors. Karamanou and Vafeas (2005) find that the likelihood of making a management earnings forecast is positively associated with stronger corporate governance in the form of more outside directors on the board, a lower level of managerial share ownership, a higher level of institutional share ownership and a smaller audit committee. Ajinkya et al. (2005) report that firms with more independent boards and greater institutional ownership are more likely to issue a management earnings forecast and to forecast more frequently. In the course of investigating the relationship between the level of prospective information disclosed and a firm’s investment opportunity set, Hossain et al. (2005) also find the percentage of outside directors on the board to be a significant variable in explaining voluntary disclosure. Finally, Nagar et al. (2003) determine that earnings forecasts are disclosed more frequently as the percentage of Chief Executive Officer (CEO) compensation tied to stock prices increases.

It is predicted in this study that a higher standard of corporate governance is associated with increased disclosure of future information in published financial reports. We group the individual corporate governance attributes into four categories based on current research findings. These categories are board autonomy, board committees, the extent of independent ownership and audit quality. These categories are then amalgamated into a summary measure of corporate governance.

3.1 Board autonomy

Board autonomy comprises four governance attributes, namely board independence, the absence of a dominant personality, chair independence and non-executive director shareholdings. It is widely accepted that a positive relationship exists between board independence, measured by the proportion of non-executive directors on the board that are not grey (that is, outsiders with special ties to the company or management), and effective corporate governance (e.g., Core et al. 1999; Young 2000; Ho and Wong 2001). It is argued that outside directors serve as more effective monitors of managerial behaviour, due to greater opportunities and incentives to exercise control (Weisbach 1988; Byrd and Hickman 1992). Specifically, it is argued that board independence may decrease managerial perquisite consumption (Brickley and James 1987). Further, independent directors are less likely to be intimidated by the CEO (Weisbach 1988). Weisbach (1988) finds that the probability of CEO replacement following a period of poor corporate performance is higher for firms with a more independent board of directors. Similarly, Kosnick (1987) argues that demands for greenmail payments are more likely to be resisted by boards with a higher percentage of independent outside directors.

The presence of a dominant personality within the ranks of executive management is thought to hinder effective corporate governance. For example, one individual fulfilling the roles of both CEO and board chair will result in increased managerial domination (Molz 1988). The presence of a dominant personality is found to be associated with poor disclosure, due to the reduced quality of monitoring and increased benefits arising from the withholding of information (Forker 1992). The independence of the board chair is demonstrated to enhance corporate performance (Rechner and Dalton 1991; Donaldson and Davis 1991), improve board effectiveness (Haniffa and Cooke 2002) and enhance the quality of corporate governance (Coulton et al. 2003)

The extent of independence displayed by non-executive directors, as measured by the level of company shareholdings, leads to higher standards of corporate governance (Beasley 1996). Uzun et al. (2004) find that a greater proportion of outside directors on the board is significantly correlated with a lower occurrence of corporate fraud.

Hence, the independent variable of board autonomy forms the basis of the first hypothesis. Prior research investigating the board of directors reveals a positive relationship between board vigilance and the standard of firm disclosure. We therefore expect a more autonomous board to be associated with increased forward-looking disclosures. This gives rise to the following hypothesis:

H1:

A positive association exists between board autonomy and the voluntary disclosure of forward-looking information in corporate annual reports.

3.2 Board committees

Board committees incorporate the presence and independence of the compensation committee, as well as the appointment of a nomination committee. These committees are associated with improved corporate governance (Forker 1992; Davis 2001). While there are no reported studies that examine a linkage between the existence and structure of board committees and the disclosure of forward-looking information, there are studies that find a relation between more independent board committees and a lower level of both financial reporting problems and corporate fraud (Song and Windram 2004; Uzun et al. 2004). Both of these factors can be regarded as indicators of financial reporting quality. Since financial reports containing forward-looking information are more likely to be perceived as being of higher quality (Ajinkya et al. 2005; Karamanou and Vafeas 2005), we can expect the presence and quality of board committees to be associated with forward-looking disclosures. Hence, this leads to our second hypothesis:

H2:

A positive association exists between the presence and quality of board committees and the voluntary disclosure of forward-looking information in corporate annual reports.

3.3 Independent ownership

The third category of corporate governance relates to the type of ownership structure inherent within the company. This category incorporates the degree of institutional ownership and block shareholdings, as well as ownership dispersion. Institutional investors have a stronger incentive to monitor managerial behaviour due to their large shareholdings (Byrd et al. 1998; DeFond 1992). Substantial shareholders are also more inclined to scrutinise managerial performance, thereby improving corporate governance (Coulton et al. 2003). For example, Denis and Serrano (1996) find that external blockholders are instrumental in removing poorly performing managers, after an unsuccessful corporate control bid.

The dispersion of shareholdings also influences the standard of governance. It is argued that within firms that are characterised by greater concentration of ownership, substantial shareholders wield increased power due to the size of their shareholdings. It is easier for fewer substantial shareholders to voice an opinion to which management will be forced to listen (Yeoh and Jubb 2001). This includes requests for disclosing relevant information such as that relating to future expectations.

The above arguments point to an association between the independence of ownership and forward-looking disclosures. Thus, our third hypothesis is as follows:

H3:

A positive association exists between independent ownership and voluntary disclosure of forward-looking information in corporate annual reports.

3.4 Audit quality

There is general consensus that the external audit constitutes a keystone of corporate governance (Cadbury Committee 1992), in that external auditors serve as gatekeepers who monitor managerial behaviour on behalf of firm stakeholders. It is argued that without efficient gatekeeping, the very structure of corporate governance is in jeopardy (Coffee 2001). The positive relationship existing between the external audit function and the standard of corporate governance is obviously strengthened by the quality of the auditor (Yeoh and Jubb 2001). This variable considers a number of facets often associated with improved audit quality, including the presence and independence of audit committees and the frequency of audit committee meetings. This is because an important role of audit committees is to monitor both audit quality and financial reporting quality (Walker 2004; Davidson et al. 2005). Further, as the effectiveness of the auditor may enhance audit quality, auditor size and independence are also incorporated into the fourth independent variable. Hence, audit quality forms the basis of the next hypothesis:

H4:

A positive association exists between audit quality and voluntary disclosure of forward-looking information in corporate annual reports.

3.5 Summary corporate governance measure

This study also examines the combined effect of all corporate governance categories within the firm on corporate disclosure of forward-looking information. Many prior studies investigate just one or two corporate governance mechanisms. However, an important contribution of this research is that it assigns an overall corporate governance score to each firm by combining the individual governance mechanisms within each of the four independent variables. This allows for the possibility that companies may choose from a portfolio of corporate governance mechanisms. However, given that stronger overall corporate governance should lead to improved financial reporting transparency, it is reasonable to expect that a stronger governance system (reflected in a higher corporate governance score) will be associated with more forward-looking disclosures. Hence, a firm’s overall governance score provides the foundation for the fifth hypothesis:

H5:

A positive association exists between the strength of a firm’s corporate governance system and the voluntary disclosure of forward-looking information in corporate annual reports.

3.6 Control variables

Prior literature identifies a number of other factors also likely to have an impact upon a firm’s voluntary disclosure policy. The control variables included in the model are firm size, firm performance, information environment and leverage. Previous disclosure studies indicate that a positive association exists between voluntary disclosure and firm size (Kent and Ung 2003). This relationship applies in the context of both disclosure quality (Lang and Lundholm 1993) and quantity (Coulton et al. 2003). In relation to the voluntary disclosure of forward-looking information in particular, larger firms are also more likely to divulge earnings forecasts than smaller firms (Cox 1987; Choon et al. 2000).

Empirical research finds that firm performance is associated with corporate disclosure, but results are mixed. Many studies document a positive relationship between firm performance and voluntary disclosure (Lev and Penman 1990; Lang and Lundholm 1993; Botosan 1997). With regard to forward-looking information, Cahan and Hossain (1996) report that strongly performing firms are more likely to voluntarily disclose earnings forecasts. However, other studies suggest that relatively poor performers are prone to greater disclosure (Skinner 1994, 1997). Therefore, while a relationship is likely to exist between firm performance and voluntary disclosure, the direction is uncertain.

The firm’s information environment is another factor reported to shape a firm’s disclosure policy. Firstly, the extent of growth opportunities available to the firm is often used to reflect a firm’s information environment (Lang and Lundholm 1993, 1996). In turn, firms with a higher percentage of growth options relative to assets-in-place are associated with increased disclosure (Hossain and Ahmed 2000). Other research suggests that the need for external finance to fund future growth opportunities prompts such firms to increase their voluntary disclosure (Core 2001). Specifically, to decrease the unacceptably high level of information asymmetry characteristic of growth firms, these firms have an incentive to increase their voluntary disclosure to minimise the cost of capital (Strebel 1996).

A number of studies evaluate the association between firm leverage and disclosure. Voluntary disclosure can alleviate information asymmetry, thereby decreasing the borrower’s apparent risk of default, in turn reducing the cost of capital (Baiman and Verrecchia 1996; Sengupta 1998). Hence, managerial disclosure may be aimed at securing capital at minimum cost (Cooke 1989). However, it should also be noted that empirical studies present mixed findings. Some researchers document a positive relationship between leverage and voluntary disclosure (Bradbury 1992) while others fail to find any statistically significant association between leverage and disclosure (Craswell and Taylor 1992; Hossain et al. 1994; Raffournier 1995) or report a significant negative relationship (Meek et al. 1995). Thus the direction of the expected association is not predicted.

4 Data and research method

4.1 Data collection

The data collection process involves the acquisition of the 2000 and 2002 published annual reports of the largest 300 publicly listed firms according to net profit for the year 2000, as identified in the Business Review Weekly (BRW). The years 2000 and 2002 are chosen for consideration in order to provide a contrast between the corporate governance structures in place within firms before and after a number of high-profile corporate collapses, such as Enron Corporation in the US and HIH Insurance Ltd in Australia. In addition, as previously noted, this time period preceded a number of key changes to regulation which may have impacted corporate disclosure and corporate governance. The research design is adapted from the Canadian study of Clarkson et al. (1994) and tailored to an Australian institutional setting, with reference to Choon et al. (2000).

The potential sample size is reduced by excluding listed property trusts, listed investment trusts, funds and trustee companies.Footnote 6 We also exclude foreign listed firms and those companies where corporate annual reports fail to cover at least 6 months. From the remaining 239 firms, we then select 200 companies on the basis that their annual reports can be readily obtained. In relation to the sample for 2002, certain firms have since delisted, reducing the sample size to 183 firms.

4.2 Models

The following logistic regression model is adopted to test Hypotheses 1–4:

$$ {\text{FLI = }}\alpha \,{\text{ + }}\beta _{{\text{1}}} {\text{BOARD + }}\beta _{{\text{2}}} {\text{COMMEE + }}\beta _{{\text{3}}} {\text{OSHIP + }}\beta _{{\text{4}}} {\text{AUDIT + }}\beta _{{\text{5}}} {\text{SIZE + }}\beta _{{\text{6}}} {\text{PERFORM + }}\beta _{{\text{7}}} {\text{INFO + }}\beta _{{\text{8}}} {\text{LEVG + }}\varepsilon $$

where: FLI, Disclosure of forward-looking information in financial report; BOARD, Board autonomy; COMMEE, Presence and value of board committees; OSHIP, Independent ownership; AUDIT, Audit quality; SIZE, Firm size; PERFORM, Firm performance; INFO, Information environment; LEVG, Firm leverage.

Our second model measures the effect of the firm’s overall corporate governance framework (CG) on the voluntary disclosure of prospective information and is used to test Hypothesis 5. The model is specified as follows:

$$ {\text{FLI = }}\alpha {\text{ + }}\beta _{{\text{1}}} {\text{CG + }}\beta _{{\text{2}}} {\text{SIZE + }}\beta _{{\text{3}}} {\text{PERFORM + }}\beta _{{\text{4}}} {\text{INFO + }}\beta _{{\text{5}}} {\text{LEVG + }}\varepsilon $$

where: FLI, Disclosure of forward-looking information in financial report; CG, Summary corporate governance score; SIZE, Firm size; PERFORM, Firm performance; INFO, Information environment; LEVG, Firm leverage.

4.3 Specification of dependent variable

The dependent variable is the voluntary disclosure of forward-looking information in corporate financial reports. This variable reflects but is not limited to earnings forecasts, which are the focus of the study conducted by Clarkson et al. (1994). Their definition of earnings forecasts is adopted to assist with the identification of such statements within Australian financial reports:

Unambiguous quantitative or qualitative statement of the expected direction of change in operating results for the upcoming fiscal year is included in its annual report. A non-forecasting firm is one which hedges discussion so that the expected direction of change was unclear or if it involved either what-if scenarios (i.e., sensitivity analyses) or forecasts of macro variables (e.g., the economy, interest rates) without relating them to the firm’s own operating results. (p. 429).

The application of this definition however does not preclude the investigation of other forward-looking information which does not qualify as an earnings forecast. Rather, provided the projection can be classified in terms of the following four characteristics, it is recognised as a forward-looking statement for the purposes of this study:

  • direction (up, down or no change);

  • type (income/profit, cash flow, sales/revenue);

  • location (Directors’ Report, CEO’s/Managing Director’s Report, Review of Operations, Chairman’s Report, D&A and other); and

  • nature (qualitative, quantitative).

To determine whether sample firms disclose forward-looking information within their annual financial reports, all narrative matter in the full and concise financial reports is read. The dependent variable, FLI, is dichotomous. It is assigned a value of one if the company discloses forward-looking information within its corporate financial report and zero otherwise.

4.4 Specification of independent variables

Table 1 summarises each of the independent variables, together with the corporate governance attributes, the way in which these attributes are measured, and their expected relationship with voluntary disclosure. A similar procedure to Brown and Caylor (2006) is followed in this study. They use detailed corporate governance data encompassing 51 provisions spanning eight governance dimensions to create a broad summary corporate governance measure. Brown and Caylor code each of the 51 factors as either 1 or 0 depending on whether Institutional Shareholder Services (ISS) considers the firm’s governance for that attribute to be minimally acceptable. These binary variables are then summed to derive a corporate governance score. Each of the 51 corporate governance provisions are categorised into one of the eight governance dimensions of ISS.

Table 1 Independent variables
Full size table

Another similar study is that of Gompers et al. (2003) who construct a “Governance Index” using the incidence of 24 governance rules to proxy for the level of shareholder rights at about 1500 large US firms in the 1990s. The main data source for the study is the Investor Responsibility Research Centre (IRRC), which publishes detailed listings of corporate governance provisions used in the defence of a takeover. The researchers divide these provisions into five groups including, for example, tactics for delaying hostile bidders.

A more intricate analysis of corporate governance measures is undertaken by Larcker et al. (2005) who examine the relation between a broad set of corporate governance indicators and various measures of managerial decision making and organizational performance. Their method involves extracting 39 structural measures of corporate governance from a sample of 2,106 firms and combining them into 14 governance constructs using principal components analysis.

In our study, the measures of corporate governance attributes and the research that supports these measures is presented in Table 1. The first independent variable, board autonomy (BOARD), reflects board independence, the absence of a dominant personality within the firm, the independence of the chair and non-executive director shareholdings. Board independence is captured by the percentage of non-executive directors comprising the board that are not grey directors. The absence of a dominant personality refers to the separation of the roles of CEO and board chair. The independence of the chair is concerned with the appointment of a non-executive director to the position of board chair. Lastly, the percentage of non-executive director share ownership is used to reflect director independence.

The corporate governance attributes comprising the category BOARD (as well as the other corporate governance categories) are assessed according to whether they are represented by binary or continuous values. In the case of dummy variables, their actual value is added to the score for BOARD. With respect to continuous variables, their impact upon a firm’s corporate governance index depends on their magnitude with respect to the average value of that variable for all firms. If a firm exhibits a continuous variable greater than or equal to the mean, one point is added to its corporate governance score. Conversely, if the value of a firm’s continuous variable for that governance attribute is lower than the mean, no points accumulate toward the BOARD score. Given that four internal corporate governance attributes are considered to calculate a score for the first independent variable, the maximum value for BOARD is four. However, in order that no one governance variable is implicitly assumed to have a greater impact on the firm’s corporate governance structure than any other, individual scores for each independent variable are standardised to one. This is achieved by dividing each component score by the number of attributes reflected in that component. For example, the score reflecting the variable BOARD is comprised of four governance attributes and is thus divided by four.

The second independent variable relates to board committees (COMMEE) and is comprised of three separate corporate governance attributes. The presence and independence of a compensation committee, as well as the appointment of a nomination committee are determined to calculate the value of the COMMEE score. We use the percentage of committee members who are non-executive and not grey directors to measure the independence of the compensation committee.

The next independent variable considers the ownership structure prevalent within the firm (OSHIP). The extent of institutional ownership is determined by whether or not the largest shareholder within the firm is an institution.Footnote 7 The presence of block holdings within firms is captured by the percentage of ordinary shareholders’ equity held by block shareholders (where holdings of greater than 5% of ordinary shareholders’ equity comprise a block holding). Finally, the percentage of shareholdings held by the top twenty investors captures the concentration of firm shareholdings.

The final measure of corporate governance pertains to the audit function. A firm’s audit quality (AUDIT) is assessed by determining whether the firm has appointed an audit committee and how often that committee meets.Footnote 8 In addition, the extent of that committee’s independence is reflected in the proportion of non-executive directors comprising the committee, where such directors are not grey. Further, audit firm size is used to gauge audit quality, with reference to whether or not the auditor is a Big 6 firm. Although the market perceives Big 6 auditors to be of higher quality (Balsam et al. 2003), certain misgivings have emerged regarding the validity of using this proxy to capture audit quality. As a result, the more recent measure of auditor independence as reflected in fees paid for the provision of non-audit services is also incorporated into the AUDIT variable (DeFond et al. 2002). Specifically, the lower the amount of fees paid by clients, to their auditors for non-audit services, the more independent the auditor.

The final independent variable is a summary measure of corporate governance, which takes into account values calculated for the categories, BOARD, COMMEE, OSHIP and AUDIT. We calculate an overall corporate governance score (CG) for each company, by adding together the standardised values for each of the preceding four independent variables. It follows that the highest possible overall governance score for a firm is four. This approach is similar to that of Yeoh and Jubb (2001) who compute an overall score to reflect corporate governance within a firm, whilst investigating an association between corporate governance and audit quality.

4.5 Specification of control variables

Table 2 presents each of the controls, along with the proxies used to capture their values and expected associations with voluntary disclosure. The size of the firm (SIZE) is measured by the log of sales. Return on assets is used as a proxy for firm performance (PERFORM), while the book-to-market ratio reflects the firm’s information environment (INFO). Lastly, the debt-to-total assets ratio is employed as the surrogate for firm leverage (LEVG). The regressions are repeated using alternative specifications of the control variables.Footnote 9

Table 2 Control variables
Full size table

5 Results

5.1 Nature of forward-looking information disclosed

Table 3 summarises the characteristics of forward-looking information collected from the annual reports for sample firms for the years 2000 and 2002 respectively. The published financial reports of the firms are perused to ascertain whether forward-looking information is included within the narrative material, as well as the type of forward-looking information that is disclosed. Table 3 indicates that in 2000 a total of 282 forward-looking statements are made by 102 firms. The majority (79%) of these statements are qualitative in nature, and predict an upward movement in firm performance (88%). Most statements are made within the chairperson’s and/or the CEO’s reports. A similar trend is identified for 2002, with 103 firms making a total of 297 forward-looking statements. Again, the majority of these statements are qualitative (83%), anticipate an improvement in corporate performance (97%) and appear within the chairperson’s and/or the CEO’s reports.

Table 3 Characteristics of forward-looking information
Full size table

5.2 Descriptive statistics

Table 4 summarises the descriptive statistics for the overall governance scores, the four independent variables and the individual governance attributes for the years 2000 and 2002 respectively. The table shows that the mean value of the overall governance score for all firms is 2.28 for 2000, increasing to 2.70 for 2002. An ANOVA test reveals that these are significantly different at p < 0.001. The scores for both years approximate a normal distribution. Further analysis indicates that, in 2000, the firm with the highest corporate governance score does in fact disclose forward-looking information while the firm with the lowest corporate governance score does not reveal any forward-looking information. Also worthy of note is the fact that the number of firms with the maximum corporate governance score grew from only one firm in the year 2000 to twelve firms for 2002. Of these twelve firms, eight disclose forward-looking information within their financial reports. Two companies exhibit the minimum corporate governance score, and neither of these makes any forward-looking statements within their financial reports. Incidentally, the minimum corporate governance score also shows substantial improvement from 2000 to 2002.

Table 4 Corporate governance variables (2000 vs. 2002)
Full size table

None of the four independent variables deviates substantially from a normal distribution, as evidenced by their values of skewness and kurtosis (not reported). Further, the mean values of three of the four variables increase from 2000 to 2002, indicating an overall improvement in corporate governance. An ANOVA test reveals that these three variables, OSHIP, BOARD and AUDIT, are significantly different at p < 0.001. Indeed, OSHIP (independent ownership) in particular shows a marked increase from 2000 to 2002. Descriptive statistics for the individual governance attributes also indicate that, in general, the mean values increase from 2000 to 2002.Footnote 10

Table 5 provides the mean corporate governance scores and independent sample tests for disclosing and non-disclosing firms. The table shows that the mean overall corporate governance score of disclosing firms (2.40) is higher than that of non-disclosers (2.16). This result is statistically significant (p = 0.0009). However, in 2002, there is no significant difference between the mean scores for disclosing (2.69) and non-disclosing (2.71) firms.

Table 5 Disclosers vs. non-disclosers (2000 vs. 2002)
Full size table

This table also presents the mean values for each of the four governance categories for the 2 years under investigation. In relation to the year 2000, the analysis suggests that firms disclosing forward-looking information exhibit stronger corporate governance in two categories. The mean values for COMMEE (the presence and quality of board committees) and AUDIT (audit quality) are higher for disclosing firms than non-disclosing firms. T-tests indicate that these results are statistically significant (p = 0.0022 for COMMEE and p = 0.0008 for AUDIT). The categories BOARD (board autonomy) and OSHIP (independent ownership) do not differ significantly between disclosing and non-disclosing firms. With respect to 2002, differences between the disclosing and non-disclosing firms in terms of corporate governance are generally not statistically significant. The one exception is the category BOARD, which is marginally significant (p = .0981) but in the opposite direction to that predicted.

The results of Mann-Whitney Tests for the fifteen individual corporate governance attributes that comprise the four categories of corporate governance are also reported in Table 5.Footnote 11 Disclosing firms exhibit stronger governance attributes than non-disclosing firms, for ten of the 15 measures outlined above. At one-tailed significance levels, there are statistically significant differences between disclosing and non-disclosing firms in terms of six corporate governance attributes. Specifically, disclosing firms are more likely to form audit and compensation committees, as well as appoint Big 6 auditors, than non-disclosing firms. Further, the audit committee of the former is likely to be more independent than that of the latter. At a 10% level of significance, both non-executive director ownership and independence of the compensation committee are also significant. In 2002, the results indicate that disclosing firms exhibit stronger corporate governance with respect to only seven of the fifteen attributes outlined above. However, in contrast to the results for 2000, none of the differences in the corporate governance attributes of disclosing and non-disclosing firms is statistically significant.

The average value of the corporate governance score for all firms (i.e., 2.28 for 2000 and 2.70 for 2002 as shown in Table 4) is then used to classify the firms as having either weaker corporate governance (less than mean score) or stronger corporate governance (greater than or equal to the mean score).Footnote 12 Table 6 shows whether or not these firms disclose forward-looking information. In relation to the year 2000, 110 firms are found to have stronger corporate governance, while 90 companies have weaker corporate governance systems. Of the 110 firms with relatively strong governance systems, a majority of 64 firms are disclosers of forward-looking information. Conversely, of the 90 firms with weaker corporate governance, a majority of 52 firms do not disclose forward-looking information. A chi-squared test indicates that this difference is statistically significant (p = 0.025), demonstrating that firms with stronger corporate governance are more likely to disclose forward-looking information. For 2002, the difference between disclosing and non-disclosing firms in terms of their corporate governance has narrowed considerably. Both groups are slightly more likely to disclose forward-looking information and the chi-squared test is not significant (p = 0.193).

Table 6 Comparison of corporate governance for disclosing and non-disclosing firms
Full size table

5.3 Logistic regression results

Table 7 reports the results of the two logistic regression models for the year 2000. The table indicates that all independent variables are moving in the predicted direction.Footnote 13 In the first regression model, only the presence and quality of board committees (COMMEE) and audit quality (AUDIT) are statistically significant. The former returns a Wald Statistic of 2.029 at the 10% significance level (p = 0.077). The latter has an associated Wald Statistic of 4.051, at the 5% level of significance (p = 0.022). Of the control variables, only INFO is statistically significant at a 10% level of significance (p = 0.084). The Chi-square of 19.445 is significant at p = 0.013. The Nagelkerke R 2 indicates that 12.4% of the decision to disclose forward-looking information is explained by the first model. These results provide support for Hypotheses 2 and 4 for the year 2000. However, there is no support for Hypotheses 1 and 3.

Table 7 Year 2000: logistic models of decision to disclose forward-looking information
Full size table

The second logistic regression incorporates the overall measure of corporate governance (CG) and the four control variables (SIZE, PERFORM, INFO, LEVG). The table shows that CG is statistically significant at the 5% level (p = 0.018). This indicates that the strength of the firm’s corporate governance framework in the year 2000 is positively associated with the disclosure of forward-looking information in financial reports. Hence, Hypothesis 5 is supported for the year 2000. Of the control variables, SIZE and INFO are significant (p = 0.045 and p = 0.063 respectively). This suggests that larger firms may be motivated to disclose forward-looking information to mitigate their political costs. Firms may also decide to publish prospective information to alleviate information asymmetry surrounding high growth firms. The Nagelkerke R 2 indicates that 10% of the decision to disclose forward-looking information is explained by the model, while the chi-square statistic resulting from the omnibus test of model coefficients is significant at the 1% level (p = 0.008).

Table 8 reports the results of the two logistic regression models for the year 2002. The table indicates that only one of the independent variables (AUDIT) is moving in the predicted positive direction, but the result is not statistically significant.Footnote 14 Accordingly, none of the hypotheses is supported for the year 2002. Of the control variables, only LEVG is statistically significant, at a 1% level of significance (p = 0.002 in the first model and p = 0.004 in the second model). This result differs from 2000, where leverage exhibits a positive but insignificant relationship with the dependent variable, suggesting that gearing is more powerful in terms of motivating the disclosure of forward-looking information in 2002 compared with 2000. The Chi-square statistic is significant in both models (p = 0.079 in the first model and p = 0.049 in the second model) while the Nagelkerke R 2 indicates that the models respectively explain 10% and 8% of the decision to disclose forward-looking information.

Table 8 Year 2002: logistic models of decision to disclose forward-looking information
Full size table

Overall, these results point to some important differences between 2000 and 2002. With respect to the year 2000, audit quality (AUDIT), the presence and quality of board committees (COMMEE) and the overall efficacy of the corporate governance system (CG) are positively associated with the disclosure of forward-looking information. Also, the information environment (INFO) of the sample firms for 2000 has a significant and positive effect on the disclosure of forward-looking information, as does firm size. However, the disclosure of forward-looking information does not seem to be driven by the same factors in 2002. In spite of the apparent strengthening of corporate governance, none of the governance attributes is significantly associated with the disclosure of forward-looking information. However, firm leverage (LEVG) appears to have a strong positive association with the disclosure of this information.Footnote 15

6 Concluding comments

In this study we investigate the association between a firm’s corporate governance framework and the decision to voluntarily disclose forward-looking information in its published financial reports. The forward-looking information disclosed by sample firms exhibits certain common characteristics. The majority of disclosing firms publish qualitative forward-looking information within reports prepared by the chief executive officer, chairman and managing director. Further, such statements tend to be expressed in terms of expecting an upward movement in such measures as profit, earnings or income. This trend persists for both 2000 and 2002.

Our results show that disclosing firms typically experience a higher standard of corporate governance than non-disclosing firms, but only for the year 2000. With respect to that year, our logistic regression results show that the corporate governance category audit quality is positively associated with the decision to disclose forward-looking information in financial reports. The category board committees is also positively associated with the disclosure of forward-looking information as is the overall efficacy of the corporate governance system. In 2002, none of the governance categories is significantly associated with the firm’s decision to publish forward-looking information in financial reports. This result is in spite of the increase in corporate governance scores for firms from 2000 to 2002 which suggest an across-the-board improvement in corporate governance during this period. Hence, while corporate scandals such as those involving Enron Corporation in the US, Parmalat in Europe and HIH Insurance Ltd in Australia saw corporate governance take global centre stage, it appears that this did not lead to a greater level of disclosure, at least in the context of forward-looking information.Footnote 16

The study has a number of limitations. The use of an index of governance mechanisms to arrive at an overall corporate governance score involves attaching an equal weighting to the various governance attributes. This assumes that every attribute is equally important to all firms. A further problem arises from the need to transform continuous scores to dichotomous variables in order to integrate them into the composite score. Splitting around the mean is a somewhat arbitrary way of dichotomising the variable and classification problems arise for those companies just above or below the mean value. An additional problem arises when comparing across the two years since some firms may be classified differently over the 2-year period because of changes in the means, even though their individual scores do not change. Further, using summary categories of corporate governance characteristics has the potential to mask major underlying relationships (Larcker et al. 2005). The sample in this study is selected from the top 300 firms by net profit, for the years 2000 and 2002, subject to certain exclusions. Examining a non-random sample of firms introduces an inherent bias into the study (Watts 1994) as larger, more profitable firms are expected to disclose more information (including prospective data). The selection criterion may also have led to the non-significant results for our performance control variable. Another potential concern of the study is the “staleness” issue wherein the “annual report may not be a timely disclosure medium” (Clarkson et al. 1994, p. 445). Specifically, if firms pre-empt disclosures in annual reports using alternative media (e.g., press releases), such disclosures would not constitute news to the market and as such, the “predictions of the voluntary disclosure literature would not apply”. Firms may pre-empt disclosures via alternative means of communication or decide to only reveal information within such media, totally bypassing the annual report. Indeed, Chan et al. (2006) report that 2,424 earnings-related prior announcements, wherein management referred to current period or future period profits, revenues, distributions/dividends or production, were made to the ASX over the period September 1994 to December 2001.

In spite of these limitations, the results of this study contribute to the relevant literature in a number of ways. First, we provide a deeper examination of voluntary corporate disclosure. The study addresses the broader category of disclosure of forward-looking information in financial reports, which includes but is not limited to quantitative earnings projections. Second, the study provides a more thorough representation of corporate governance within the firm by considering fifteen governance devices. Third, our comparison of results between 2000 and 2002 reveals a marked reduction in the association between corporate governance and the disclosure of forward-looking information in spite of a strengthening of governance during the period. These findings are particularly relevant to corporate regulators and policy makers, given the global interest in corporate governance and its expected role in improving the integrity and transparency of financial reports. Indeed, our findings cast some doubt on the effectiveness of corporate governance mechanisms to promote transparency and hence improve market confidence.

Future research could extend the study by expanding the time period under consideration. In particular, it may be worthwhile to consider whether the improvements in corporate governance exhibited from 2000 to 2002 have been sustained. Additional research could further expand the study by altering the sample selection criteria to remove any bias based on size or profitability. Finally, an important contribution to the literature would be to examine the association between corporate governance variables and the disclosure of forward-looking information in media other than annual reports.