DISCLOSURE QUALITY IN TUNISIAN ANNUAL REPORTS

Transcription

1 DISCLOSURE QUALITY IN TUNISIAN ANNUAL REPORTS Dr. Raïda Chakroun University of Sfax (Tunisia) Pr. Khaled Hussainey* Plymouth University Correspondence Details Professor Khaled Hussainey Plymouth Business School Plymouth University 205 Cookworthy Building Drake Circus, Plymouth Devon PL4 8AA United Kingdom 1

2 DISCLOSURE QUALITY IN TUNISIAN ANNUAL REPORTS ABSTARCT This paper explores the disclosure quality and its determinants in the Tunisian context. More specifically, we follow Beest and Braam (2012) s approach in measuring disclosure quality on the one hand and we examine if disclosure quality and disclosure quantity share the same determinants on the other hand. For this reason, we use 56 annual reports from non-financial companies listed on the Tunisian Stock Exchange for years 2007 and Our results show that board independence (managerial ownership) affects negatively (positively) disclosure quality. However, the results show that the determinants of disclosure quality and disclosure quantity differ. We contribute to disclosure studies by being the first study to examine the disclosure quality in Tunisia. In addition, this study enables us to provide the stakeholders of the Tunisian companies (like regulators and managers) with a diagnosis of the determinants of disclosure quality and quantity. KEY WORDS Disclosure quality, disclosure quantity, determinants, annual reports 2

3 1. Introduction We contribute to the literature by being the first study to examine the disclosure quality in one of the developing countries. Different proxies for measuring the quality of corporate disclosure have been discussed in a number of review articles in recent years. Accounting researchers have critically examined these proxies and argued that there is an urgent need to develop a measure for disclosure quality (Core, 2001 and Beyer et al., 2010). In particular, Beyer et al. (2010) argue that: A sensible economic definition of voluntary disclosure / financial reporting quality and direct derivation of measures from that definition is missing from the literature. This lack of an underlying economic definition hinders our ability to draw inferences from this work, and we recommend that future research address this issue (p.311). In responding to Beyer et al. (2010), recent efforts has been undertaken to measure the quality of corporate disclosure in developed countries. These include Beattie et al (2004), Beretta and Bozzolan (2004 & 2008), Anis et al. (2010), Bamber and McMeeking (2010) and Beest et al. (2009). This paper measures the quality of corporate disclosure for a sample of Tunisian companies for years 2007 and Prior literature also suggests that disclosure quality may be related to disclosure quantity and hence disclosure quality and quantity share the same determinants (Anis et al., 2012; Bamber and McMeeking, 2010). Therefore we test to see if disclosure quality and disclosure quantity share the same determinants in the Tunisian context. The main purpose of this paper is to measure disclosure quality for a sample of non-financial companies listed on the Tunisian emerging market for the period We also measure the determinants of both disclosure quality and quantity. In order to study this two importance research issues, we follow Beest and Braam (2012) s approach in measuring disclosure quality. We also use quantitative regression techniques to measure the drivers of disclosure quality and quantity. As expected, the empirical results indicate that some corporate governance mechanisms (board independence, managerial ownership) affect disclosure quality. This result is in line with prior research that also models the link between disclosure and corporate governance. Consistent with Anis et al. (2012) and Bamber and McMeeking (2010), the empirical results also indicate that the determinants of disclosure quality differ from the determinants of disclosure quantity. The remainder of the paper is organised as follows. Section 2 discusses the literature review and the development of research hypotheses. Section 3 explains the research design. Section 4 describes the data. Section 5 reports the empirical findings. Section 6 concludes the study. 2. Literature Review 2.1. Institutional Framework Tunisia is an African developing country of the MENA zone. It has an emerging stock market composed of 57 listed companies among which there are 25 financial institutions, where the minority shareholders are not well protected and the corporate disclosure is a weakly regulated field. Kothari (2000) argues that if enforcement of shareholder rights and disclosure standards is weak, then the quality of disclosure tends to be poor, regardless of the disclosure standards. Information disclosure in the annual report is not yet strictly regulated in Tunisia, the thing which makes the research useful in this area. In this way, Tunisian companies are still publishing their annual reports using local accounting standards and not the International Financial Reporting Standards (IFRS). Tunisia is experiencing a predictable harmonization with the (IFRS). This move toward the international standards is a real upheaval in the 3

4 financial information guideline of the company and aims to adopt a common and internationally recognized financial language (Chakroun, 2012). In Tunisia, the legal obligations for the annual reports are set by the Code of Commercial Companies (which have a field of application covering most of the trading companies), the firms' accounting system (1997) that was established through standards in harmony with those of the IASB and the regulation of the Financial Market Council (of which the fields of application extend to all the companies publicly appealing to savings) (Chakroun and Matoussi, 2012). Indeed, article 201 of the Code of Commercial Companies gives no precision about the form and content of this report, but only about the fact that it must be detailed. In addition, article 44 of the regulation of the Financial Market Council, which relates to public offering approved by the April 7, 2000 order of the finance minister, lists the compulsory information to be provided in the annual report. Moreover, the Financial Market Council statement published in 2010 reminds the companies making public offerings of the required mandatory information to include in their annual reports. The legal environment of Tunisia has undergone major changes in the recent years which encourage the Tunisian companies to disclose information at the highest level of quality in their annual reports. This is clearly reflected mainly in the promulgation of the Law No concerning the strengthening of financial security. Than we notice a change in corporate governance environment despite the absence of a formal regulatory framework to mentor it. This is reflected by the publications by the Arab Institute of Business Leaders (in 2008 and update in 2012) of a Guide about Good Governance Practices of Companies and of a Guide of the Annual Report of the Tunisian Companies (in 2009); as well as by the establishment of the Tunisian Center of Corporate Governance (in 2009). To conclude, the emergence of many changes related to the information environment on the Tunisian Stock Exchange, the predictable harmonization with the (IFRS) and the development of corporate governance in the economy, highlighted the strategic importance of the quality of annual reports in Tunisia. We should mention that firms tend to disclose information at the highest level of quality to attract more financial analysts (Lang and Lundholm, 1993 and Healy et al., 1999). This has however created new expectations of the Tunisian financial analysts and portfolio managers relating to the quality of corporate disclosure (Chakroun, 2012). Due to the scarcity of studies on the disclosure quality in the emerging markets and the call for research on this topic by Beyer et al. (2010) and Botosan (2004), we intend to elucidate it in the Tunisian emerging stock market Literature review of measurements methods to assess the quality of financial reporting The disclosure is a theoretical concept that is difficult to measure directly (Marston and Shrives, 1991). More specifically, the quality of disclosure and the decision usefulness of the financial reports information are complex, multidimensional context-sensitive and subjective concepts that cannot be observed directly. Botosan (2004) argues that no universally accepted notion of disclosure quality exists. Core (2001) notes that improved measures of disclosure quality need to be developed. This author has emphasized the importance of developing improved measures of disclosure quality. We should indicate that Hussainey et al. (2003), Hassan and Marston (2010), Marston and Shrives (1991), Beattie et al. (2004) and Beretta and Bozzolan ( ) mention the difficulty to measure the disclosure quality of corporate financial reporting. Furthermore Botosan (2004) show that disclosure quality is not measurable. Various types of measurement methods and proxies 1 have been developed and used in prior empirical research to assess and evaluate the quality of corporate disclosure (Healy and 4

5 Palepu, 2001). Beyer et al. (2010) review prior research that considers different proxies for the quality of corporate disclosure. Hassan and Marston (2010) classify the measures of disclosure provided in prior studies into two approaches. The first approach includes proxies for disclosure, which are not directly based on examining the original disclosure vehicle(s). The second approach provides measures of disclosure obtained by inspecting the original disclosure vehicle(s). We will present the measures of: Beattie et al. (2004), Beretta and Bozzolan (2004a, 2008), Anis et al. (2012) and Beest and Braam (2012). Beattie et al. (2004) and Beretta and Bozzolan (2004a, 2008) are considered the key attempts for measuring the disclosure quality. The work of Anis et al. (2012) and Beest and Braam (2012) are currently at the working paper stage of development. The first pioneering study to develop a measure of disclosure quality of Beattie et al. (2004) provides a general framework applicable to various types of information. This study state that quality is a function of the quantity plus four-dimensional framework for the content analysis of accounting narratives namely: the spread (the number of topics disclosed), the time orientation of the information (historical or forward-looking), the financial orientation (financial/non-financial), and the quantitative orientation (quantitative/qualitative).in addition, this paper presents a computer-assisted methodology, explores the complex concept of quality, and the problematic nature of quality assessment. Beretta and Bozzolan (2004a) are restricted to the disclosure quality of risk information. The authors propose a measure that captures four main dimensions: the content of information (the quantity of disclosure based on pre-determined topics); 2 the economic sign (positive/negative information); the type of information (financial/non-financial information); and the outlook orientation. Beretta and Bozzolan (2004b) argue that the quality of voluntary disclosure should be defined from the user s perspective. In this regard, multidimensional frameworks should be based on a detailed analysis of the information needs expressed by specific segments of users on specific issues. This seems particularly important in the case of risk communication, given the multifaceted nature of risk. Beretta and Bozzolan (2008) are restricted to the disclosure quality of forward-looking information. These authors suggest a multidimensional measure that combines disclosure quantity and richness of information. Richness is a function of both width and depth. Disclosure width consists of disclosure coverage (the extent of disclosure of relevant topics) and disclosure dispersion (the spread of disclosure across different topics). Disclosure depth addresses the question of what information is disclosed. The authors identify three information attributes that represent disclosure depth: outlook dimension; the information measurement type (qualitative/quantitative information; financial/non-financial information); and the economic sign (positive/negative news information). Anis et al. (2012) contribute to existing disclosure literature by providing a multidimensional measure for disclosure quality that is supported by a valid framework (Botosan, 2004) 3. The method of these authors, which operationalize the qualitative characteristics of information aims to assess the quality of different dimensions of information simultaneously to determine the decision usefulness of financial reporting information. As a response to Botosan s (2004) recommendation that disclosure quality measures should use a well-established regulatory framework, Anis et al. (2012)consider the Operating and Financial Review best practice (OFR) framework (ASB, 2006) as a base for developing of their measure of disclosure quality. This measure represents a sum of the following information attributes: forwardlooking orientation, verifiability, relevance, supplementary and complementary financial statements, comprehensiveness, readability, balance and neutrality, and comparability. 5

6 Beest and Braam (2012) examine whether there are differences between IFRS and US GAAP based financial reports in meeting the fundamental and enhancing qualitative characteristics for decision usefulness as defined in the Conceptual Framework of the IASB (2010).Fundamental and enhancing qualitative characteristics are the underlying attributes contributing to the decision usefulness of information. For financial information to be useful, it must be relevant and faithfully represent what it purports to represent. The enhancing qualitative characteristics: understandability, comparability, verifiability and timeliness are complementary to the fundamental characteristics and distinguish more useful information from less useful information(iasb, 2010).Although the enhancing qualitative characteristics are perceived to be less important than the fundamental ones, for a comprehensive assessment it remains important to include them in the analysis. This study adds to the literature by the developing and testing a comprehensive and compound financial reporting quality assessment tool that aims to measure the decision usefulness of financial and non-financial reporting information in annual reports, both in terms of the fundamental and the enhancing qualitative characteristics as defined in the Conceptual Framework of the IASB (2010). The FASB and the IASB published an exposure draft of An improved Conceptual Framework for Financial Reporting (IASB, 2008; FASB, 2008). According to the exposure draft, providing decision-useful information is the primary objective of financial reporting. Decision-useful information is defined as information about the reporting entity that is useful to present and potential equity investors, lenders and other creditors in making decisions in their capacity as capital providers (IASB, 2008). In line with the exposure draft, Beest and Braam (2012) define financial reporting quality in terms of decision usefulness (e.g. Jonas and Blanchet, 2000). Jonas and Blanchet (2000) develop questions referring to the separate qualitative characteristics in order to assess information quality. Their research indicates that qualitative characteristics can be made operational; their measures are based on the frameworks of the FASB (1980) and the IASB (1989). In addition Botosan (2004) argues that disclosure quality is a function of information quality attributes proposed by a regulatory framework. These attributes can be the qualitative characteristics of information as proposed by the conceptual frameworks for financial reporting of the International Accounting Standards Board (IASB 2008, IASB 2010) and the Financial Accounting Standards Board (FASB 1980, FASB 2001).There must be a proper framework to support the chosen information attributes (Botosan, 2004). In fact, the quality definition proposed by Beest and Braam (2012) is supported by regulatory bodies and recommendatory reports. Following Botosan (2004) and Jonas and Blanchet (2000), the measure for disclosure quality introduced by Beest and Braam (2012) is based on the following qualitative characteristics of information: on the fundamental qualitative characteristics (relevance and faithful representation) and on the enhancing qualitative characteristics (understandability, comparability and timeliness). Finally, we can say that there is no clear definition of disclosure quality and that its measurement is recognized as a relevant question that is still open in literature; despite several methods have been proposed for the purpose of constructing a measure for the disclosure quality. In fact, the development of a measurement tool of disclosure allows inter-company, inter-industry and inter-country comparisons to be made and also allows changes over time to be monitored. In addition, an appropriate measure of disclosure can permit much more powerful tests of many research questions that relate to corporate disclosure Disclosure quantity versus Disclosure quality Majority of the previous empirical studies do not make a clear distinction between the quantity and quality of disclosure and as it is difficult to measure disclosure quality, support 6

7 the view that quantity is a proxy for the quality of disclosure (e.g. Hussainey et al., 2003 and Hassan and Marston, 2010). These studies assume that disclosure quality and disclosure quantity are positively correlated. However, Beattie et al. (2004), Anis et al. (2012) and Berretta and Bozzolan (2008) criticize this approach. Hassan and Marston (2010) argue that in the absence of a generally agreed model for disclosure quality as well as relevant and reliable techniques to measure it, prior studies tend to use disclosure quantity as a proxy for disclosure quality assuming that quantity and quality are positively related, although quantity and quality measures may lead to different ranking for a sample of companies. Moving from the basic idea that the quantity of information has an implication in determining its quality in many empirical settings and that disclosure quality is inherently immeasurable (Botosan, 2004), it is generally assumed that the extent of disclosure (i.e., quantity) is an adequate measure of the quality of disclosure. Botosan (2004) argues that the measure of disclosure quality of Beretta and Bozzolan (2004a) just counts the number of information items; hence, it does not differ from quantity-based measures used in prior research. Indeed, Botosan (2004) is based on a regulatory foundation for identifying specific information quality attributes. (Amir and Lev, 1996; Hussainey et al., 2003; Schleicher et al., 2007; Hussainey and Walker, 2009) use the quantity of forward-looking statements as a proxy for disclosure quality. These studies find that this information improves investors ability to anticipate future earnings change. Furthermore, Lang and Lundholm (1993) and Botosan (1997) use researcher constructed indices where the amount of disclosure is used as a proxy for disclosure quality and Botosan (1997) tend to assume that disclosure quantity and quality are positively related. In addition Beretta and Bozzolan (2004a) state that the quality of disclosure depends on the quantity of information disclosed and the richness of its content. However other researchers contend that even if the quantity of information disclosed influences the quality of information, an assessment on disclosure quality cannot be based purely on this association (e.g. Beattie et al., 2004; Beretta and Bozzolan (2008)). Beattie et al. (2004) overemphasise disclosure quantity as a component of disclosure quality. In addition, the authors do not justify their key assumption that firms disclosing more information are more likely to have a greater level of quality. According to Anis et al. (2012), among the limitations of the measures of quality disclosure proposed by the prior research is that these measures overemphasise quantity when calculating disclosure quality. These authors, based on a sample of UK firms, provide the first empirical evidence that disclosure quantity is not a proper proxy for disclosure quality. In fact, firms may disclose more information yet such information could lack accuracy. They show an insignificant negative correlation between the quantity and the quality scores. Even, the simple regression to examine if the quantity could be used as an explanatory variable for disclosure quality indicates that disclosure quantity and quality are not correlated and disclosure quantity is not a proper predictor of disclosure quality. These results argue that disclosure quantity is not a proper proxy for quality. Anis et al. (2012) also show that the determinants of disclosure quality and disclosure quantity are not identical. In addition Beretta and Bozzolan (2008) tests confirm that richness and quantity of disclosure are two independent dimensions and reveal that quantity is not a good proxy for quality in assessing narrative disclosure. The empirical evidence offered by their study supports the hypothesis that the dimensions considered in the disclosure quality framework (quantity measured by the standardized relative quantity and richness with its components and richness with its elements depth and width) give a more realistic picture of disclosure than quantity does and suggest that these dimensions can be complementarily used in assessing the disclosure. 7

8 2.4. Determinants of disclosure quality We indicate that the association between corporate governance mechanisms and disclosure quality is an extension proposed by Anis et al. (2012). These authors argue that firm characteristic determinants for disclosure quality and quantity are different which supports the evidence that disclosure quantity is not a precise proxy for disclosure quality. Further, the meta-analysis conducted by Ahmed and Courtis (1999) showed that the results of the researches dealing with the relationship between the corporate characteristics and the disclosure quality in annual reports are mixed. These mixed results might be attributed to the use of inaccurate disclosure quality measures (Beattie et al., 2004). For this reason, Cohen et al. (2004) highlight the relationship between corporate governance mechanisms and financial reporting quality. These authors state that better corporate governance leads to improved financial reporting. The previous literature cited several factors that explain the corporate disclosure. These factors are principally the characteristics of the company and the corporate governance mechanisms. Our study focuses interest on certain corporate governance mechanisms: the board independence, its size, the leadership structure, the managerial ownership, the family control and the quality of auditor; and on certain characteristics of the company: the size of business and the age of the company. Patelli and Prencipe (2007) show a positive relationship between the independence of the board and voluntary disclosure for a sample of Italian companies. Similarly, results from previous empirical studies (Apostolou and Nanopoulos, 2009; Lim et al., 2007; Chen and Jaggi, 2000) show a positive relationship between the independence of the board and the corporate disclosure. This results can be explained by the fact that when the board is independent, this will lead to a better control of management and therefore to a high quality of disclosure. On the Tunisian context Chakroun and Matoussi (2012) find a negative and significant relationship between the board independence and the extent of voluntary disclosure closely linked to the mandatory one in the annual reports. This result is explained by the fact that independent administrators may be regarded as strangers to the company without being actually independent. The definition of an independent administrator is not defined in the Code of Commercial Companies which does not require companies to include such administrators in their boards. In this case the independent administrators in Tunisia can be considered only managers' advisers. In addition Jouini (2013) show a negative and insignificant relation between the board independence and the level of financial disclosure. H 1: There is a positive relationship between the board independence and the quality of disclosure When we assumed that the culture of the quality of disclosure is not deeply rooted in the minds of most of the Tunisian managers, it is very likely to see, in the large-sized boards, members who encourage the increase of the disclosure quality. That is, when boards are large, it is more likely that they include administrators who tend to favor the best quality of disclosure. Chakroun and Matoussi (2012) state the existence of a positive and significant relationship between the size of the board and the extent of voluntary disclosure. Moreover Jouini (2013) find a positive but insignificant relation between the size of the board and the level of financial disclosure. Therefore, we expect that companies with large-sized boards 4 will disclose a higher quality of information. H 2: There is a positive relationship between the size of the board and the quality of disclosure 8

9 According to the assumption of the interest alignment of the dominant personality in the company with those of the other shareholders (Morck et al., 1988), we expect that the existence of a leadership structure (Combination of functions of GM and CH) within the company helps the disclosure quality to rise. Barako (2007) emphasizes, in a sample of Kenyan firms, the existence of a positive and significant relationship between the leadership structure and the three sub-indexes of voluntary disclosure connected to the general and strategic information, the financial and social information and the information about the board. In addition Chakroun and Matoussi (2012) find a positive and significant relationship between the leadership structure and the extent of voluntary disclosure in a sample of Tunisian firms. H 3: The quality of disclosure is higher in firms where there is a leadership structure than in the other firms. The shareholders-administrators can tend to enhance the disclosure quality in order to clear themselves from the other shareholders (non-administrators) and to show them they do not transfer the wealth of the company to their own accounts. In accordance with the assumption of interest alignment of the controlling shareholders with those of the other shareholders in the firm (Morck et al., 1988), we expect that the managerial ownership helps the disclosure quality to rise. More specifically, the more the part held by the shareholders-administrators is important, the weaker the divergences of interests between them and the other shareholders are. That is when administrators hold a significant part of capital; ownership and management are held by the same persons whose interests converge with those of the shareholders nonadministrators interested by the disclosure quality. Thus, we expect that disclosure quality in the annual reports increases with increases in managerial ownership. A high managerial ownership can help increase the disclosure quality of the company (Li and Qi, 2008). On the Tunisian context Chakroun and Matoussi (2012) find also a positive and significant relation between the managerial ownership and the extent of voluntary disclosure. H 4: There is a positive relationship between the managerial ownership and the quality of disclosure In a family business, the members of the family are involved in its management and have a precise knowledge about their business. We expect that these members do not promote a high quality of information. Thus, the firms whose are controlled by families are expected to disclose a low quality of information than other firms. Indeed Chau and Gray (2002) argue that firms controlled by families provide less voluntary information than the non family ones. Chakroun and Matoussi (2012) show also that the extent of voluntary disclosure not closely linked to the mandatory one is lower for firms controlled by families than in the other companies. H 5: The disclosure quality is lower for firms controlled by families than in the other firms. 3. Research methodology 3.1. Data and sample selection TABLE 1 ABOUT HERE This research focuses on data of companies of the non-financial sector listed on the Tunisian Stock Exchange (observed in the years ). We exclude financial firms because their annual reports differ from those of non-financial firms (Schleicher and Walker, 2010). This gives us a sample of 54 firm-year observations. We choose this period because it is after the promulgation of the Law No concerning the strengthening of financial security. 9

10 To assess the disclosure quality we used a manual content analysis on the annual reports. We consulted the annual reports of the companies which are collected from the Financial Market Council and the stockbrokers in the market as they are not directly downloadable through the Internet. Our data for the characteristics of the companies and the corporate governance mechanisms are collected from the website of the Tunisian Stock Exchange and the annual reports of companies Measurement method to assess the disclosure quality Since we have not electronic versions of annual reports, we cannot apply a computer-assisted methodology for our content analysis. The method selected to assess the disclosure quality of Beest et al. (2009) is applicable for the hard copies of the annual reports of our sample. Our study follows the scoring technique developed in Beest et al. (2009). Their study produces a comprehensive measure to operationalize the fundamental and enhancing qualitative characteristic of annual reports information. In line with Beuselinck and Manigart (2007), Beest et al. (2009) and Jonas and Blanchet (2000), we define disclosure quality in terms of decision usefulness of annual reports. We assess a score that represent a proxy of the disclosure quality of 54 annual reports. The operationalzation of the qualitative characteristics of reporting information is based on 19- item index, of which 3 were related to relevance, 5 to faithful representation, 4 to understandability,6 to comparability and 1 to timeliness. We have dropped two items from the list of items of Beest et al. (2009) which are not applicable for the Tunisian firms (Relevance 3 and Understandability 4). In fact the method of Beest et al. (2009) is adapted to the Tunisian context as Botosan (2004) state that researcher must recognize that effective frameworks for assessing disclosure quality are likely to be context specific. The reports were coded on the items using predefined 5 point Likert scales. To ensure consistency in the scoring all annual reports were read twice. As recommended by Botosan (2004) and by Jonas and Blanchet (2000), the measure of Beest et al. (2009) captures all the qualitative characteristics of information discussed in the conceptual frameworks for financial reporting of the International Accounting Standards Board (IASB 2008) 5 and the Financial Accounting Standards Board (FASB 1980):the fundamental qualitative characteristics (i.e. relevance and faithful representation) and the enhancing qualitative characteristics (i.e. understandability, comparability and timeliness). These frameworks could provide the foundation needed to support a framework for assessing disclosure quality (Botosan, 2004). We should indicate that Beest et al. (2009) used multiple items that were drawn from existing measurement items that were already developed in previous studies (e.g. Jonas and Blanchet, 2000).Appendix A provides an overview of the 19 measured items used to operationalize the fundamental and enhancing qualitative characteristics. This appendix also includes the measurement scales used to assess the value of the distinct items. To compute a standardized outcome for each qualitative characteristic (sub scores), the scores on the related items are added and divided by the total number of items. We measure a sub score for each qualitative characteristic and then we measure a score that presents an aggregate measure for the disclosure quality. The aggregated disclosure quality score is a function of five measures (sub scores) representing the quality attributes: relevance, faithful representation, understandability, comparability and timeliness. The sub scores that compose the aggregated score are equally weighted because there is no reason to prioritize one attribute over the others. Indeed, the ASB (2006) values all attributes equally. We will now discuss these five qualitative characteristics. The fundamental qualitative characteristics (i.e. relevance and faithful representation) are most important and determine the quality of information. The enhancing qualitative characteristics (i.e. understandability, 10

11 comparability and timeliness) can improve decision usefulness when the fundamental qualitative characteristics are established. However, they cannot determine disclosure quality on their own (IASB, 2008). Relevance Information is considered relevant if it is capable of making a difference in the decisions made by users (IASB, 2010, p. 17).The IFRS also provide a more specific definition of relevance: financial information is capable of making a difference in decisions if it has predictive value, confirmatory value or both (IASB, 2010, p. 17).Information would have a predictive value if it can be used as an input to processes employed by users to predict future outcomes (IFRS 2010b, p. 17). Information would have a confirmatory value if it provides feedback about (confirms or changes) previous evaluations (IFRS 2010b, p. 17). Information that has predictive value usually has confirmatory value. Faithful representation Faithful representation is the second fundamental qualitative characteristic as elaborated in the conceptual frameworks. To faithfully represent economic phenomena that information purports to represent, annual reports must be complete, neutral, and free from material error (IASB, 2010). Economic phenomena represented in the annual report are economic resources and obligations and the transactions and other events and circumstances that change them (IASB, 2006). Understandability The IASB (2010) defines understandability as the quality of information that enables users to comprehend its meaning. The IASB (2010) argues that understandability is enhanced when information is classified, characterized and presented clearly and concisely. In addition Boelens et al. (2011) argue that graphs can be used as a measure for understandability. Comparability Comparability is considered as a quality attribute of information that enables users to identify similarities in, and differences between, two sets of economic phenomena (IASB, 2010).In addition comparability as a quality attribute helps users identify the main trends and the analysis of a firm s performance over time (ASB, 2006). According to Jonas and Blanchet (2000), if a firm provides the same information over time, comparability would be achieved. This idea is consistent with using financial statements format as one indicator of the comparability of financial statements quality. We should mention that consistency refers to the use of the same accounting policies and procedures, either from period to period within an entity or in a single period across entities (IASB, 2010) Timeliness The final enhancing qualitative characteristic defined in the conceptual frameworks is timeliness. Timeliness means having information available to decision-makers before it loses its capacity of influencing decisions (IASB, 2010). Timeliness refers to the time it takes to reveal the information and is related to decision usefulness in general (IASB, 2010) Measurement method to assess the voluntary disclosure quantity Healy and Palepu (2001) state, that one of the limitations of the studies on voluntary disclosure is the difficulty in measuring its extent.marston and Shrives (1991) provide a review of the use of disclosure indexes in accounting research. 11

12 Our measure of disclosure quantity is based to the Botosan (1997) 6 's index adapted to the Tunisian context (Appendix B).We have dropped eight items not disclosed by any company in our sample. Based on the previous studies to identify the information expected by the users of the annual reports and on the Guide of the Annual Report of the Tunisian Companies published in 2009; we have added three categories of information: information on intangible assets, social and environmental information and information on governance. We used an unweighted and weighted index based on the views of financial analysts and portfolio managers. According to the unweighted approach, an item takes "1" if disclosed and "0" otherwise. The extent of disclosure is measured by the ratio between the score of the company and its maximum possible score for not to penalize it for non-disclosing items when they are not relevant to its activities. 72 UN_DIS i = J = 1 x ji / M i With: M i : maximum number of items of which disclosure is possible for company "i" ;M i 72, x ij = "1" if j th item is disclosed and = "0" otherwise. It should be noted that for the weighting of the disclosure quantity score, we based on data from an investigation through a questionnaire on a sample of 40 Tunisian financial analysts and portfolio managers in Tunisia 7 (Chakroun and Matoussi, 2012). This method reflects the relative utility of each item and admits that all items provide a different utility to the selected user of the annual report. The respondents were asked to rate the usefulness they attach of the items on a5 points Likert scale. The values attached to the items which could be disclosed in the annual reports are (1=Not useful at all), (2=Little useful), (3=Somewhat useful), (4=Useful) and (5=Very useful). According to the weighted approach, an item takes its "weight" if it is disclosed and "0" otherwise. The weight represents the arithmetic average of the points awarded by the respondents to the item W_DIS i = J = 1 x ij *P j / 3.4. The Association between Disclosure Quantity and Disclosure Quality We examine the extent to which disclosure quality and disclosure quantity are correlated and hence the former can be used as a proxy for the latter. In addition, we examine the extent to which both disclosure quality and disclosure quantity share the same determinants. We compare the determinants of the disclosure quantity with the determinants of the disclosure quality, especially that prior studies show that the determinants of disclosure quality and disclosure quantity are not identical based on a sample of UK firms (e.g Anis et al., 2012). We use the following regression model to examine the determinants of disclosure quality and quantity: Mi J = 1 P j Khaled Hussainey 11/7/13 12:41 PM Comment [1]: OK. Please write here the regression model that we use. Please also explain the definition of both the dependent and independent variables. DIS i = β 0 + β 1 YEAR i + β 2 INDB i + β 3 SIB i + β 4 COMFUN i + β 5 MAN i + β 6 FAM i + β 7 AGE i + β 8 QAU i + β 9 LSIZE + ε i Where; DIS = disclosure quality (quantity). Disclosure quality is measured through the fundamental qualitative characteristics (relevance and faithful representation) and enhancing qualitative characteristics (understandability, comparability and timeliness) qualitative information characteristics and their aggregation. Disclosure quantity is measured by a weighted and an unweighted score. YEAR = 1 in 2008 and = 0 in INDB is the independence of the board. SIB is the size of the board. COMFUN is the combination of functions of General 12

13 Manager and Chairman. MAN is managerial ownership. FAM is family control. AGE is the age of the company. QAU is the quality of auditor, and LSIZE is the size of business. Table 2 shows the definition of each of the variables and the data source. 4. Descriptive Statistics TABLE 2 ABOUT HERE We first present the descriptive statistics of the proxies of the disclosure quality and then of the proxies of the disclosure quantity. Afterwards, we present a summary of the descriptive statistics of the independent variables. TABLE 3 ABOUT HERE Table 3 shows that the means of the sub scores of disclosure quality: relevance (R_DISQUA), faithful representation (FR_DISQUA), understandability (U_DISQUA) and comparability (C_DISQUA) are close with a little superiority to (U_DISQUA). We notice that the mean observed for the (C_DISQUA) sub score is relatively low and is of the order of In other words, the firms in our sample tend to be weakly concerned by the qualitative characteristic of comparability. The highest mean is observed for the sub score of timeliness (T_DISQUA).Then it appears that timeliness is the highest qualitative characteristic for the companies of our sample. The mean and median of the aggregate disclosure quality score (DISQUA) rise to 2.90 and 2.86 respectively. In addition, its minimum is 1.95 and its maximum is 4. This result indicates that the disclosure quality of the companies of our sample tend to have a medium level since the values of the mean and the median are close to the neutral value 3. Furthermore, by examining the means and medians values of the disclosure quantity scores W_DIS and UN_DIS), we notice that these values are very close. Such result means that there is no difference between the weighted and unweighted measures of the voluntary disclosure quantity. Moreover, we can see that the boards of directors are generally not independent: the mean and median of the INDB variable reached 28 % and 29 % respectively. The standard deviation of this variable is very close to its mean and increases to 23 %. This could be explained by the variability between the companies of our sample regarding the independence of their boards. The review of the SIB variable reveals that the boards of directors tend to be large. The mean of this variable is 8.81 and its median is For the COMFUN variable, we notice that 62 % of the companies of our sample have a Chairman who is at the same time the General Manager. The mean and the median of the MAN variable are respectively 59 % and 63%. These results enable us to ascertain that the firms in our sample are characterized by a very strong property of administrators. For variable FAM, we can say that more than the third of the observations represent family-controlled companies. This high proportion reflects a characteristic of the Tunisian economic tissue which is the dominance of the familycontrolled businesses. By looking at the control variables, we can see that the mean of the AGE variable rises to For the QAU variable, we notice that only 33 % of the observed companies have a «Big 4» auditor. Finally, the mean of the variable size of business measured by the natural logarithm of total assets is of

14 5. Empirical Results and Discussion 5.1. Correlation Analyses TABLE 4 ABOUT HERE Table 4 shows a significant positive (negative) correlation between the disclosure quality score and the managerial ownership (the independence of the board). More specifically Pearson's correlation coefficients between the disclosure quality and the managerial ownership and between the disclosure quality and the independence of the board stand respectively at 34 % and 33 % and they are significant at 5%. In addition, this table shows some significant correlations between some independent variables such as the correlations between the size of the board, on the one hand and the independence of the board and the size of business, on the other hand. Hence, these results push us to conduct further multicollinearity analyses. TABLE 5 ABOUT HERE Moreover, table 5 shows that the highest correlations between the sub scores of disclosure quality are observed between the sub score of understandability (U_DISQUA) and the sub score of comparability (C_DISQUA), on the one hand and between the sub score of faithful representation (FR_DISQUA) and the sub scores of understandability (U_DISQUA) and of comparability (C_DISQUA), on the other hand. We observe, also, a strong and positive correlation between the scores of disclosure quantity and all the sub scores of disclosure quality, except the sub score of timeliness. This indicates that disclosure quantity and qualitative characteristics of information are correlated and disclosure quantity could be a predictor of disclosure quality. Consequently the prevailing assumption in the literature that disclosure quantity and quality are correlated, and thus quantity presents a proper proxy for quality could be precise and should be tested by multivariate analyses. Furthermore, the correlation between the weighted and unweighted disclosure quantity scores stand at 99% and it is significant. This result could be interpreted by the fact of the non reliability of the weighting of items. Finally, we focus on the correlation between the quantity and quality scores. Pearson correlation shows a significant positive correlation (0.71) between the quality and the quantity scores (weighted and unweighted). As discussed earlier, it seems that the disclosure quantity could be a proper proxy of disclosure quality. Moreover the correlation analysis yields logical results about the strong and significant correlations between the aggregate score of disclosure quality and all its sub scores Results and Discussion of the Multivariate Analyses Results Related to the Multiple Regression Models of Disclosure Quality In order to measure the influence of corporate characteristics on disclosure quality in annual reports, we conducted OLS regressions. Before explaining the results of the OLS regression analysis, the model was tested on multicollinearity. Table 4 shows that the variance inflation factor (VIF) was smaller than the threshold value "3" for each of the variables, which indicates the absence of the problem of multicollinearity. TABLE 6 (PANEL A) ABOUT HERE Panel A of table 6 presents a summary of the results of the multivariate tests of all the developed hypotheses. It should be noted that prior studies offer mixed results when examining the effect of firm characteristics on disclosure quantity or proxies for disclosure quality. The results vary according to the disclosure measure being used, sample 14

15 characteristics, the country object of the study or the choice of variables used in the model (firm characteristics and corporate governance mechanisms). As a result, we use theoretical arguments adapted to the Tunisian context as a base for explaining our findings. We discovered that the coefficient of variable INDB is negative, with a high value and also significant. Then, we can conclude that the independence of the board tends to decrease the disclosure quality. This result corroborates the results of the work of Chakroun and Matoussi (2012) and Jouini (2013) conducted in the Tunisian context and allows us to disprove H 1. Indeed, independent administrators may be regarded as strangers administrators to the firm without being actually independent since the definition of an independent administrator is not clear in Tunisian Law. Besides, the coefficient of the MAN variable is positive, with a high value and significant. The positive found relationship allows us to confirm H 4 and affirm the assumption of the aligning of the interests of the shareholders-administrators with those of the other shareholders (Morck et al., 1988). This result could be explained by the fact that the shareholders-administrators, who have a close idea about the business, can tend to improve the quality of disclosure in order to clear themselves from the other shareholders. The coefficient of the SIB variable has the positive expected sign but it is insignificant which suggests that the size of the board tends to increase the disclosure quality. Likewise, the coefficient of the COMFUN variable has the positive expected sign but it is insignificant. This result could be interpreted by the fact that the leadership structure could increases partially the disclosure quality. Also, the coefficient of the FAM variable has the negative expected sign but it is insignificant. It could be explained by the little separation between ownership and management in the firms controlled by families. Indeed, the family members have a good knowledge about their business through their participating in the management and therefore they are not concerned with the improvement of disclosure quality. To conclude, the insignificant coefficients of the variables SIB, COMFUN and FAM allow us to invalidate our hypotheses H 2, H 3 and H 5.The results of the model with the aggregate score of disclosure quality as dependent variable suggest to validating only our hypothesis H 4. Furthermore, by comparing the R 2 of the regressions of panel B and panel C of table 6, it appears that these values are significantly higher for the regressions with the fundamental qualitative characteristics as dependent variables than for the regressions with the enhancing qualitative characteristics as dependent variables. TABLE 6 (PANEL B) ABOUT HERE Panel B of table 6 shows that there is no significant relationship between the corporate characteristics and the disclosure quality score on relevance. It shows also a negative and significant relation between the board independence and the disclosure quality score on faithful representation and a positive and significant relation between the managerial ownership and the disclosure quality score on faithful representation. We mention that these results are similar to those found for the model with the aggregate score of disclosure quality as dependent variable. TABLE 6 (PANEL C) ABOUT HERE In addition, when we observe Panel C of table 6, we notice the existence of the same results for the disclosure quality based on the sub scores of faithful representation and of understandability. We can say that the positive significant relation between the managerial ownership on the one hand; and the disclosure quality sub scores on faithful representation and on understandability on the other hand; allows us to strengthen H 4. 15